Skip to main content

39th International Symposium on Intensive Care and Emergency Medicine

Brussels, Belgium, 19-22 March 2019

Accepted abstracts for 39th International Symposium on Intensive Care and Emergency Medicine

P001 Prognostic value of a genetic polymorphism of AQP5 in sepsis depends on a source of infection

V Pisarev1, A Chumachenko1, I Tyurin2, R Cherpakov2, A Tutelyan3

1Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, V.A.Negovsky Institute of General Reanimatology, Moscow, Russia; 2V.M.Buiyanov City Clinical Hospital, Anesthesia-Reanimatology Department, Moscow, Russia; 3Central Institute of Epidemiology, Moscow, Russia

Introduction: The purpose of the study was to determine whether the preferential localization of the infection and age affect the prognostic value of the genetic marker AQP5 (1364A/C, rs3759129) in outcome prediction in sepsis patients. Studies by Adamzik and colleagues have demonstrated that aquaporin AQP5 polymorphism (1364A/C, rs3759129) associates with increased 30-day survival in sepsis patients presumably due to increased gene expression that enhance the leukocyte migration. To increase the informative value of the prediction and decrease the cost, it might be crucial to determine at a pre-test level the subset of patients who might benefit most from the prognostic genotyping.

Methods: Sepsis and septic shock were defined in patients according to SEPSIS-3 (2016) recommendations. Study groups (n=152) included ICU patients with abdominal sepsis (AS, including pancreatitits, peritonitis, cholecystitis, appendicitis; n=98) and sepsis patients with other sources of infections. AQP5 polymorphism was studied by analyzing PCR products in a 2% agarose gel using a AQP5 1364A/C specific tetra primer set. Data were analyzed by Kaplan-Meyer plot and Fisher test, and odds ratios were calculated.

Results: Distribution of alleles (A and C) and genotypes (AA, CA and CC) AQP5 1364A/C in patients with sepsis or sepsis subgroups (sepsis with no septic shock and sepsis shock patients) versus control group (healthy volunteers) did not differ. Although there was a trend to preferential survival of sepsis patients with genotype C AQP5 despite the source of infection, only patients with AQP5 CC or CA genotype and abdominal sepsis (Sepsis-3), or a subgroup of the same AQP5 genotype experiencing septic shock, demonstrated increased 30-day survival versus AA homozygotic patients (P<0.002).

Conclusions: The informative value of detecting the AQP5 CC or CA genotype for prognosis of 30-day survival versus AA homozygotic patients is increased only in abdominal sepsis patients.

P002 Depressed expression of FCER1A gene is associated with increased mortality in infected surgical patients

R Almansa1, C Andrés2, M Martín-Fernández3, S Montero4, C Jambrina5, C Doncel6, J Sánchez-Crespo5, M Heredia-Rodríguez7, J Rico4, C González8, E Sánchez-Barrado5, M Lorenzo-López7, S Martín4, L Muñoz-Bellvis8, M Vaquero5, E Tamayo7, C Aldecoa4, J Bermejo-Martín6

1Hospital Clínico Universitario de Valladolid/IECSCYL, BioSepsis (Group of Biomedical Research in Sepsis), Valladolid, Spain; 2Hospital Clínico Universitario de Valladolid, Clinical Analysis Service, Valladolid, Spain; 3Hospital Clínico Universitario de Valladolid/IECSCYL, BioSepsis (Group for Biomedical Research in Sepsis), Valladolid, Spain; 4Hospital Universitario Rio Hortega, Anesthesiology and Reanimation Service, Valladolid, Spain; 5Hospital Clínico Universitario de Salamanca, Anesthesiology and Reanimation Service, Salamanca, Spain; 6Hospital Clínico Universitario de Valladolid/IECSCYL, BioSepsis (Group for Biomedical Research in Sepsis), Valladolid, Spain; 7Hospital Clínico Universitario de Valladolid, Anesthesiology and Reanimation Service, Valladolid, Spain; 8Hospital Clínico Universitario de Salamanca, Department of General and Gastrointestinal Surgery, Salamanca, Spain

Introduction: Increasing evidence supports a central role for “immunosuppression” in sepsis. It is necessary to develop biomarkers of immune dysfunction that could help to identify patients at risk of poor outcomes [1]. The decreased expression of human leucocyte antigen (HLA)-DRA is proposed as a major feature of immunodepression and its persistent decrease is associated with mortality in sepsis [2]. In a previous study, we evidenced that FCER1A (Fc Fragment Of IgE Receptor Ia) is the gene showing the lowest expression levels of the entire transcriptome in sepsis [3]. Here we studied the association between FCER1A expression and mortality in infected surgical patients.

Methods: FCER1A and HLA-DRA expression levels were quantified by droplet digital PCR in blood of 257 infected surgical patients. 26 patients died within 28 days (10.11%). Spearman test was used to evaluate the association between gene expression and the Sequential Organ Failure Assessment (SOFA) score. Areas under Receiver Operating Curves (AUROC) were used to determine the gene expression cut-off values predicting mortality. Kaplan-Meier survival curves were obtained and differences in survival between groups were evaluated using the Log rank test. Cox regression was employed to assess mortality risk at 28 days.

Results: Gene expression levels of FCER1A and HLA-DRA correlated inversely with patients’ severity (r: -0.5 p<0.001; r: -0.3, p<0.001 respectively). Both genes showed significant AUROCs to predict survival, but FCER1A showed the best accuracy (Fig. 1). Patients with low levels of FCER1A or HLA-DRA had an increased risk of mortality and died 3 days earlier than non survivors with higher expression levels of these genes (Fig. 2, Table 1-2).

Conclusions: Depressed FCER1A gene expression is associated with severity and increased mortality in surgical patients with infection.


1 Hotchkiss R et al. Lancet Infect Dis 13(3): 260–268, 2013

2 Cazalis MA et al. Crit Care 10;17(6):R287, 2013

3 Almansa R et al. J Infect 70(5):445-56, 2015

Table 1 (abstract P002). Predictive capacity of FCER1A gene expression cut-off for 28-day mortality in surgical patients with infection. (COX regression)
Table 2 (abstract P002). Predictive capacity of HLA-DRA gene expression cut-off for 28 day mortality in surgical patients with infection. (COX regression)
Fig. 1 (abstract P002).

AUROCs for differential diagnosis of mortality in surgical patients with infection

Fig. 2 (abstract P002).

Kaplan-Meier survival curves. Kaplan-Meier survival curves were established after stratification based on calculated thresholds (optimal operating points of FCER1A (A) and HLA-DRA (B) expression levels)

P003 Genetic markers of nosocomial pneumonia, acute respiratory failure and renal insufficiency in critically ill patients

M Khadzhieva1, O Belopolskaya1, T Smelaya2, A Kuzovlev2, L Salnikova3

1N.I. Vavilov Institute of General Genetics, Russian Academy of Sciences, Moscow, Russia; 2Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, Moscow, Russia; 3Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, N.I. Vavilov Institute of General Genetics, Russian Academy of Sciences, Moscow, Russia

Introduction: Severe pulmonary and renal conditions such as acute respiratory distress syndrome (ARDS), respiratory failure, and deterioration in kidney function often occur in patients with nosocomial pneumonia (NP). The emergence and course of infection is genetically determined, hence host genetic landscape may influence an ability to resist infection.

Methods: Variants for genotyping were selected using the PheWAS Catalog which presents genotypic data for 13835 Caucasian patients, 1358 phenotypes and 3144 single nucleotide polymorphisms (SNPs) with P < 0.05 [1]. SNPs with the lowest P-values for phenotypes with both, respiratory and renal manifestations were selected: intergenic variants rs7130588 and rs4980785, rs347344 (EDIL3) and rs2470893 (CYP1A1). CYP1A1 gene was associated with pneumonia and ARDS in our previous investigations, so we included in our analysis three sites of CYP1A1 gene (rs2606345, rs4646903 and rs1048943) studied on a smaller sample. Genotyping was performed on 7 sites for a sample of resuscitation patients with or without NP and other pulmonary complications (n = 354 and n = 216, respectively).

Results: Allele rs2606345-G of the CYP1A1 gene was protective against ARDS and an increase in creatinine level (Fig. 1). The rs7130588-G allele was associated with lung complications and with the development of severe respiratory insufficiency (Fig. 2).

Conclusions: The SNPs rs2606345 and rs7130588 can influence the aggravation of pulmonary and renal symptoms through genetically mediated response to infection.


1. Denny JC et al. Nat Biotechnol 31: 1102–1110, 2013.

Fig. 1 (abstract P003).

Protective effect of the rs2606345-G allele (CYP1A1 gene) on the risk of ARDS development (left) and an increase of serum creatinine level on the 14th day after hospitalization (right)

Fig. 2 (abstract P003).

Association of the rs7130588-G allele with lung complications (LC) (NP, ARDS, pleurisy, abscess, etc.) (left) and with the development of severe respiratory failure (right)

P004 Mesenchymal stem cells regulate LPS–stimulated macrophages polarization balance by paracrine transforming growth factor beta

F Liu

School of Medicine, Zhongda Hospital, Southeast University, Department of Critical Care Medicine, Nanjing, Jiangsu, China

Introduction: An uncontrolled inflammatory response plays a major role in the sepsis related organ dysfunction. Mesenchymal stem cells(MSCs) can improve survival of sepsis experimental models by modulating the inflammatory response. Macrophages have been considered as important immune effector cells and their polarization imbalance aggravates the disordered inflammation reaction. The project aims to identify the effects of MSCs on macrophages polarization against dysregulated inflammatory response.

Methods: RAW264.7 cells were plated in the lower chambers of transwell system in the presence or absence of Lipopolysaccharide (LPS). Then, MSCs were seeded in the upper chambers and incubation for different time. Finally, transforming growth factor beta (TGF-β) receptor (TGF-βR) inhibitor was added in transwell system. The phenotype of RAW264.7 cells were analyzed by flow cytometry, the levels of inflammatory cytokines were detected by Enzyme-linked immunosorbent assay (ELISA).

Results: Our data showed that LPS increased the level of interleukin (IL)-6 in RAW264.7 cells (p<0.001) (Fig. 1). In line with IL-6 expression, LPS induced the expression of M1 macrophage (p<0.001). Moreover, LPS stimulated RAW264.7 cells co-culture with MSCs in transwell system, MSCs inhibited the expression of IL-6 and M1 macrophages, while increased M2 macrophages (p<0.001). Compared with LPS group, the concentration of TGF-Β was obviously increased in MSCs treatment groups (p<0.001), furthermore, there were no significantly difference between MSCs directed and indicted groups. More significantly, TGF-βR inhibitor abolished the impact of MSCs on LPS stimulated RAW264.7 cells (p<0.001) (Fig. 2).

Conclusions: MSCs polarized M1 macrophages into M2 macrophages and decreased pro-inflammatory cytokine levels by paracrining TGF-β.

Fig. 1 (abstract P004).

Lipopolysaccharide (LPS) promoted RAW 264.7 M1 polarization. (a): LPS 500 ng/ml stimulated RAW264.7 for 12, 24, 48 and 72 hours, the level of IL-6 significantly increased in a time-dependent manner. (b, c): Flow cytometry showed LPS enhanced the expression of M1 macrophages at all time points. *** P<0.001 compared with control group

Fig. 2 (abstract P004).

TGF-β secreted by MSCs regulated the M1 to M2 in LPS-Stimulated RAW264.7. LPS stimulated RAW264.7 co-culture with MSCs in transwell system for 24, 48 and 72 hours. (a): LPS increased the level of IL-6, whereas MSCs inhibited expression of IL-6. (b, c): MSCs reduced M1 macrophages while increased M2 macrophages in LPS stimulated RAW264.7. (d): The concentration of TGF-β was obviously increased in MSCs directed or in-directed group. (e, f, g): LPS increased IL-6 and M1 macrophages, and MSCs inhibited the IL-6 and M1 macrophages while increased M2 macrophages. TGF-βR inhibitor reversed the effect of MSCs on LPS-stimulated RAW264.7. *P<0.05, **P<0.01, *** P<0.001 compared with control group. # # # P<0.001 compared with LPS stimulated group. &&& P<0.001 compared with MSCs treatment group

P005 Chronomics in ICU: effect of timing of septic shock onset on circadian rhythm profiles of melatonin and cortisol

E Sertaridou1, I Chouvarda2, K Arvanitidis3, E Filidou4, G Kolios3, I Pnevmatikos1

1University Hospital of Alexandroupolis, Intensive Care Unit, Alexandroupolis, Greece; 2Aristotle University of Thessaloniki, Faculty of Medicine, Thessaloniki, Greece; 3Democritus University of Thrace, Faculty of Medicine, Alexandroupolis, Greece; 4Democritus University of Thrace, Faculty of Mrdicine, Alexandroupolis, Greece

Introduction: Circadian rhythmicity of melatonin and cortisol has been found to be affected by sepsis in both experimental and clinical studies.

Methods: In this study, we evaluated the potential effect of septic shock on circadian rhythms of urinary excreted aMT6s, a melatonin’s metabolite and cortisol in 26 patients, divided into two groups: Group A (N=15) included subjects with septic shock upon admission to the ICU and Group B (N= 11) included patients who developed septic shock during ICU stay. Urine samples were collected every 4 h over a 24-h period, whereas data were available during entry and before discharge from the ICU in Group A and during entry, septic shock and before exit from the ICU in Group B. Circadian analysis was performed leading to the estimation of mesor (mean value), amplitude of the oscillation and acrophase (phase shift of maximum values in hours).

Results: Circadian markers of aMT6s and cortisol exhibited inverse changes, both within and between groups, since amplitude of aMT6s was reduced in entry in relation to exit (437.2±309.2 vs 674.1±657.6 ng/4h, p<0.05), whereas amplitude of cortisol was increased upon admission compared to exit (13.3±31 ng/4h vs 8.7±21.2 ng/4h p<0.05), in Group A. Furthermore, in Group B, mesor of aMT6s was increased during septic shock (2492.2± 1709.1 ng/4h) compared to both entry (895.4±715.5 ng/4h) and exit (1308.6± 1214.4 ng/4h, p<0.05 for all comparisons). However, cortisol’s mean values were reduced during septic shock (10±5.3 ng/4h) compared to both entry (30±57.9 ng/4h) and exit (14.4±20.7 ng/4h, p<0.05 for all comparisons) and correlated with higher APACHE II score and longer ICU and hospital stay (p<0.05 for all comparisons).

Conclusions: Septic shock induced inverse changes of aMT6s and cortisol circadian rhythm profiles, depending on timing of onset.

P006 Investigation of the relationship between organ damage, microcirculatory dysfunction and reactive oxygen intermediate formation in experimental sepsis

J Kaszaki, A Rutai, R Fejes, S Tallósy, M Poles, D Érces, M Boros, A Szabó

Universitiy of Szeged, Institute of Surgical Research, Szeged, Hungary

Introduction: Sepsis is dysregulated response to an infection, which can lead to progressive microcirculatory dysfunction, release of reactive oxygen intermediates (ROI) and life-threatening organ dysfunction. Our aim was to investigate the relationship between organ damage - characterized by the Sequential Organ Failure Assessment (SOFA) scores, microcirculatory failure and ROI production, in a large animal model of experimental sepsis.

Methods: Fecal peritonitis was induced in anesthetized minipigs (n=28; 0.5g/kg autfeces containing 5-9 x106 CFU bacteria i.p.), control animals (n=9) received sterile saline i.p. Invasive hemodynamic monitoring and blood gas analyses were performed between 16-24 hrs, the signs for failure of circulatory, respiratory and urinary systems were evaluated in accordance with the SOFA score. The microcirculatory perfusion rate in the sublingual region was measured by orthogonal polarization spectral imaging technique (Cytoscan A/R). The leukocyte-origin ROI production was determined by lucigenine (mostly O2-.) and luminol-based (H2O2) chemiluminescence methods.

Results: Between 16-24 hrs after induction the SOFA score indicated moderate organ failure in 19 animals (M: 1.9; 25p: 1.5, 75p: 2.9) and the change was statistically significantly higher in 9 pigs, suggesting severe organ dysfunction (M: 4.1; 25p: 3.5, 75p: 5.2). The microcirculation was significantly deteriorated in all cases, independently of SOFA score data. The H2O2 production was significantly lower in septic animals as compared to controls, while the lucigenine enhanced ROI production correlated with the SOFA score-indicated moderate and severe organ dysfunction.

Conclusions: Sublingual microcirculatory parameters are not correlating with the severity of SOFA score-indicated organ dysfunction in abdominal sepsis. The measurement of ROI production of the whole blood seems to be better biomarker for the detection of the progression of events from moderate to severe organ damages.

Grant supports: NKFIH K116689; EFOP-3.6.2-16-2017-00006

P007 Increased rate of mechanical ventilation in septic patients with left ventricular dysfunction

A Newsome, S Smith, T Jones

The University of Georgia College of Pharmacy, Pharmacy, North Augusta, United States

Introduction: The purpose of this study was to characterize differences in sepsis management in patients with and without left ventricular (LV) dysfunction. Septic patients with LV dysfunction have higher mortality, and limited guidance exists for sepsis management of patients with LV dysfunction. The possibility exists that the cornerstones of sepsis management may contribute to these poor outcomes.

Methods: A retrospective chart review was conducted from May 2016 - January 2018 at two centers. Adult patients who had a diagnosis of sepsis, were treated with vasopressors for > 3 hours, and had an echocardiogram within 12 months were included. Patients were divided into two groups: reduced ejection fraction (EF) of < 40% and preserved EF defined as EF ≥40%. Information about patient outcomes and sepsis management were collected. The primary outcome was the need for mechanical ventilation (MV). Categorical and continuous data were analyzed using the Chi-Squared and Mann-Whitney U tests, respectively. The IRB has approved this project.

Results: A total of 37 patients with EF < 40% and 42 patients with EF ≥40% were included. No significant differences in fluid management, vasoactive agent maximum rate or duration, or steroid use were observed. Net fluid balance between low and preserved EF was positive 4.6 liters vs. 5.1 liters (p = 0.814), respectively. The number of patients that needed MV was higher in the low EF cohort (86% vs. 57%, p = 0.004), and this cohort had fewer MV-free days (20, IQR 0-25 vs. 24 (IQR 0 -28), p=0.064.

Conclusions: No significant differences were observed with regard to sepsis management, reflecting current guidelines. The significantly increased need for MV is a provocative result. A potential mechanism is the inability of a patient with reduced LV dysfunction to maintain appropriate cardiac and respiratory function in the face of fluid overload. Prospective analysis of the role of fluid balance in septic patients with LV dysfunction is warranted.

P008 Biomarkers of myocardial injury and cytokine plasma levels in septic patients

M Assuncao1, FR Machado2, MK Brunialti3, O Rigato4, R Salomao3

1Hospital Israelita Albert Einstein, Department of Critical Care, Sao Paulo, Brazil; 2Federal University of Sao Paulo, Department of Anesthesiology, Pain and Intensive Care, Sao Paulo, Brazil; 3Federal University of Sao Paulo, Division of Infections Diseases, Sao Paulo, Brazil; 4Hospital Sirio Libanes, Department of Critical Care, Sao Paulo, Brazil

Introduction: The relationship between myocardial injury and systemic inflammation in sepsis response is not well understood [1]. It´s proposed to evaluate the association between myocardial injury biomarkers, high-sensitive troponin T (hs-cTnT) and N-terminal pro-brain natriuretic peptide (NT-ProBNP), with inflammatory mediators (IL-6, IL-1Β , IL-8, IL-10, IL-12 / IL-23p40, IL17A, IL- 21 and TNF-α ) and biomarkers, C protein reactive (CPR) and procalcitonin (PCT), in septic patients

Methods: This was a prospective cohort study performed in three intensive care units, from September 2007 to September 2010 enrolling patients with sepsis (infection associated with organ dysfunction), and septic shock (hypotension refractory by fluids infusion requiring vasopressor). Blood samples were collected up to 48h after the development of first organ dysfunction (D0) and on the 7th day after inclusion in the study (D7)

Results: Ninety-five patients were enrolled, with median age 64 years (interquatile?48–78), APACHE II: median 19 (14-22), SOFA: median 8 (5-10); 24.2% were admitted in ICU with sepsis and 75.8% with septic shock. Hospital mortality was 34.7%. In D0, NT-ProBNP correlated with IL-8 (r = 0.495, p <0.001) and IL-10 (r = 0.471, p <0.001). In D7, hs-cTnT and NT-ProBNP correlated with PCT (r = 0.446, p < 0.001 and r = 0.495, p < 0.001; respectively). NT-ProBNP D0 was higher in non-survivors than in survivors on mortality in seventh day (p = 0.029) and in-hospital mortality (p = 0.030). hs-cTnT D7 (p = 0.030) and NT-ProBNP D7 (p <0.001) were significantly higher in non-survivors on in-hospital mortality. NT-ProBNP D7 (OR 9.28; IC95% 2.05-41.94, p=0,004) and hs-cTnT D7 (OR 10,93; IC95% 2.139 – 55.795, p=0,04) were independently associated with in-hospital mortality

Conclusions: NT-ProBNP plasma levels at D0 correlated with IL-8 and IL-10, and both NT-ProBNP and hs-cTnT at D7 correlated with PCT. In addition, NT-ProBNP has been shown to be an important predictor of mortality


1. Landesberg G et al. Chest. 2015;148:93-102.

P009 Repeated measures of heparin-binding protein correlate with mean arterial pressure and systemic vascular resistance index in septic shock: a pilot study on biomarker kinetics from a Swedish intensive care unit

J Tverring1, N Nielsen2, F Kahn3, A Linder3, P Åkesson3

1Division of Infection Medicine, BMC, B14, Faculty of Medicine, Department of Clinical Sciences, Lund, Sweden; 2Division of Anesthesiology and Intensive Care, Department of Clinical Sciences, Lund, Sweden; 3Division of Infection Medicine, BMC, B14, Lund, Sweden

Introduction: Heparin-binding protein (HBP) acts proinflammatory on immune cells and induces vascular leakage through cytoskeletal rearrangement and cell contraction in the endothelium and is a promising novel prognostic biomarker in sepsis and septic shock. However, studies on repeated measures of HBP are lacking. Our objective was to describe the kinetics of plasma HBP during septic shock and correlate it to hemodynamic parameters.

Methods: We included patients with septic shock (sepsis-3) on admission to Helsingborg hospital’s intensive care unit (ICU) during September 2016 to February 2018. Patients were sampled from ICU admission and every 4 hours for 72 hours or until death or ICU discharge. The plasma samples were analyzed for HBP and converted using the natural log (lnHBP) for normality. lnHBP was then evaluated against mean arterial pressure (MAP) as primary analysis and against systemic vascular resistance index (SVRI) as a secondary analysis, using mixed-effects linear regression models, treating patient id as a random intercept and adjusting for hemodynamic parameters.

Results: A total of 22 patients were included with median age 67 years, 9 females (41%), 7 surgical admissions (32%), median SOFA-score 12 points on day one and 6 deaths from all causes within 90 days (27%). Plasma HBP ranged from 0 to 932 ng/ml with a median of 47 ng/ml (lnHBP range 1.6 to 6.8, median: 3.9). An increase lnHBP was significantly associated with a decrease in MAP (Coef. -2.58 mmHg, 95% CI: -0.62 to -4.55, p=0.010, n=22), when adjusting for heart rate (HR), noradrenaline (NA), vasopressin (VP), dobutamine (DBT) and levosimendan (LS). In a secondary subgroup analysis, an increase in lnHBP was also significantly associated with a decrease in SVRI (Coef. -94.2 dyne*s*cm-5*m-2, 95% CI: -1.3 to -187.1, p=0.047, n=13), when adjusting for MAP, HR, NA, VP, DBT, LS and cardiac index.

Conclusions: Repeated measures of plasma HBP during septic shock were correlated with important hemodynamic parameters in this small pilot study.

P010 Mid-regional pro-adrenomedullin (MR-proADM) as early mortality predictor in septic shock

V Lovati, F Marsigli, E Pierucci

Azienda Ospedaliero Universitaria Policlinico Sant´Orsola Malpighi, Dipartimento di Scienze Mediche e Chirurgiche, Bologna, Italy

Introduction: Mid-regional pro-Adrenomedullin (MR-proADM) comes from the synthesis of the hormone adrenomedullin (ADM), which is overexpressed during inflammation and progression from sepsis to septic shock. Thus, MR-proADM can be a useful biomarker for the clinical management of septic patients [1]. The aim of our study was to understand the ability of MR-proADM to predict 30-day (30-d) mortality and to find a correlation between MR-proADM and Sequential Organ Failure Assessment (SOFA) score in the first 24 hours from Intensive Care Unit (ICU) admission.

Methods: We evaluated 28 consecutive septic shock patients according to 2016 Sepsis III definitions. Clinical data from the medical records included demographics, comorbidities, laboratories, microbiology and biomarker levels. Whole blood samples for biomarker profiling were collected at 24, 72 and 120 hours from ICU admission. MR-proADM measurement was detected in EDTA plasma using a sandwich immunoassay by TRACE® (Time Resolved Amplified Cryptate Emission) technology (Kryptor Thermo Fischer Scientific BRAHMS).

Results: Overall 30-d mortality rate was 50.0%. MR-proADM [odds ratio (OR) = 1.195], SOFA score (OR = 2.174) and Lactate (Lac) levels (OR = 1.956) in the first 24 hours were associated with 30-d mortality in univariate logistic analysis (P value < 0.05, Table 1). 30-d mortality rate was not associated with procalcitonin (PCT) levels (OR = 1.002). Further linear regression analysis showed significant correlation between MR-proADM and SOFA score at 24 hours from ICU admission (P value<0.001, Fig. 1, Table 2).

Conclusions: MR-proADM demonstrated superior accuracy to predict 30-d mortality compared to PCT levels and is directly linked to SOFA score at 24 hours from admission. MR-proADM may aid early identification of poor prognosis septic patients who could benefit a more intensive management.


1. Andaluz-Ojeda D. et al. Ann Intensive Care 7, 15 (2017)

Table 1 (abstract P010). Univariate logistic analysis between 30-day mortality, MR-proADM, SOFA score, lactate levels, PCT levels
Table 2 (abstract P010). Linear regression analysis between MR-proADM and SOFA score related to Fig. 1
Fig. 1 (abstract P010).

Linear regression analysis between MR-proADM and SOFA score at 24h from ICU admission: SOFA score = Coeff. x MR-proADM + Const. Coeff. = 0.3211 L/nmol Const. = 8.5158

P011 Pre-sepsin as diagnostic and prognostic marker in sepsis: prospective evaluation by test and confirmation cohorts

N Melachroinopoulos1, S Pouriki2, A Prekates3, K Toutouzas4, C Mathas5, E Giamarellos-Bourboulis6

1National and Kapodistrian University of Athens, Athens, Greece; 2Hippokrateion General Hospital, Athens, Greece; 3Tzaneion General Hospital, Intensive Care Unit, Piraeus, Greece; 4National and Kapodistrian University of Athens, 1st Department of Propedeutic Surgery, Athens, Greece; 5Konstantopouleion General Hospital, Intensive Care Unit, Athens, Greece; 6National and Kapodistrian University of Athens, 4th Department of Internal Medicine, Athens, Greece

Introduction: Biomarkers have not yet been studied in prospective studies using Sepsis-3. The diagnostic and prognostic validity of pre-sepsin (soluble CD14) was studied in both one test and one confirmation cohort.

Methods: The test cohort was the prospective clinical study INTELLIGENCE-1 ( NCT03306186) enrolling patients with documented infections and at least one qSOFA sign. The confirmation cohort was the prospective clinical study INTELLIGENCE-2 ( NCT03306186) with patients admitted in the emergencies with at least one qSOFA sign. Blood samples were collected within the first 24 hours of the presence of the qSOFA criteria and pre-sepsin was measured in plasma using the PATHFAST assay. Patients were classified as sepsis and non-sepsis using Sepsis-3 definitions; 28-day mortality was recorded.

Results: In the test cohort, 62 patients were classified as non-sepsis and 111 as sepsis. Using ROC curve analysis, it was found that the best trade-off between sensitivity and specificity was provided at 350 pg/ml. The odds ratio for sepsis with presepsin above 350 pg/ml was 4.04 (p<0.0001) providing diagnostic sensitivity 80.2%. In logistic regression analysis, it was found that Charlson’s comorbidity index more than 2, history of type 2 diabetes mellitus and of chronic obstructive pulmonary disease and presepsin more than 350 pg/ml were the only variables independently associated with sepsis. Presepsin above 350 pg/ml was associated with sensitivity 91.5% for 28-day mortality. The odds ratio for mortality with presepsin above 350 pg/ml was 6.84 (p: 0.001). In the confirmation cohort, 59 patients were enrolled. The sensitivity of presepsin above 350 pg/ml for the diagnosis of sepsis was 85.7% and for the prediction of 28-day mortality 100%.

Conclusions: Using a test and confirmation cohort approach, presepsin above 350 pg/ml was proved a valuable indicator for the diagnosis of sepsis and outcome prognosis among the most severe patients with one qSOFA sign.

P012 Toll-like receptors as biomarkers of sepsis in the emergency department

C Graham, LY Leung, R Lo, YK Leung, K Hung

The Chinese University of Hong Kong, Accident and Emergency Medicine Academic Uint, Hong Kong, China

Introduction: We aimed to investigate circulating TLRs gene signatures in Emergency Department (ED) patients at high risk of developing sepsis. Sepsis is “life-threatening organ dysfunction due to dysregulated host responses to infection”. Toll-like receptors (TLRs) are proteins that play a key role in the immune system’s response to infection. Thus, TLRs may act as early markers to identify patients at high risk of sepsis.

Methods: This is single-center, prospective study conducted in the ED of Prince of Wales Hospital, Hong Kong (July-September 2017). Patients presented with suspected infection were recruited. Blood samples were collected and buffy coat TLR mRNA levels were measured by real-time polymerase chain reaction (PCR). Beta-2-microglobulin (B2M) was used as a control gene.

Results: Among 67 patients recruited (median age 69 years, IQR: 56-84; 46.3% male), we analyzed TLR gene signatures in 21 infection patients and 13 sepsis patients. We recruited 10 gout patients and 10 healthy controls (HC). Median buffy coat TLR-3 mRNA levels were lower in sepsis patients compared with infection, gout and HC groups (0.26 vs 1.67 vs 1.15 vs 1.25 ng/ng B2M, p<0.05). Higher TLR-7 levels were found in infection patients than in the gout and HC groups (0.46 vs 0.28 vs 0.30 ng/ng B2M, p<0.05), whereas lower TLR-9 levels were found in sepsis than infection and HC groups (0.015 vs 0.034 vs 0.025 ng/ng B2M, p<0.05). Receiver operator curve analysis of TLR-3, -7 & -9 for discriminating sepsis and non-sepsis patients, the areas under the curve (AUC) were 0.82, 0.61 and 0.68 respectively. The combination of TLR-3, -7 and -9 demonstrated the largest AUC: 0.94.

Conclusions: TLRs mRNA signatures in buffy coat vary among different pathological conditions and have the potential to be an early marker to identify patients at high risk of development of sepsis. Combinations of TLR-3, -7 and -9 could further improve the diagnostic potential of the prediction of sepsis development.

P013 Evaluation of cell-free DNA (cfDNA) as predictor of mortality and severity in hospitalized septic patients

L Maia, P Frizera Vassallo, R Caldeira Machado Berger, V Garrone Barauna, V M. Curty, D Zaniqueli

UFES, Physiology, Vitória, Brazil

Introduction: Study of the expression of cell free DNA (cfDNA) in the search for new biomarkers for infection, sepsis and septic shock.

Methods: The population studied was all patients included in the sepsis protocol from March 2017 to January 2018, hospitalized patients of a federal public hospital. Plasma samples were collected for quantification of cfDNA, which after centrifugation were stored at -80 ° C and then thawed and analyzed by fluorescence using a Varioskan Flash fluorometer). CfDNA values were expressed as ng/mL. The patients were divided into 2 groups: Infection and sepsis/septic shock. We analyzed mortality, Sequential Organ Failure Assessment Score (SOFA score), qSOFA (quick SOFA), comorbidities, cfDNA and laboratory parameters of 111 patients.

Results: Among the 111 patients, 28% were classified as infection and 72% sepsis/septic shock. Overall lethality was 33%, infection 9.7%, and sepsis/septic shock 42.5% (p<0.001). The mean of cfDNA, SOFA and lactate was higher according to the classification of infection and sepsis/septic shock: CfDNA (159.4±117.3 and 282.7±358.6, p=0.006), SOFA (1.9±2.1 and 6.6±4.3, p<0.001), QSOFA (positive in 25% and 75%, lactate (1.6±0.8 and 3.8±3.5, p<0.001). We analyzed leukocytes, creatinine, CRP (C reactive protein), INR (International Normalized Ratio), as predictors of severity and only CRP showed no association with disease severity (P=0.84). Levels of cfDNA and qSOFA showed worse prognostic utility as a predictor of sepsis / septic shock when compared to lactate and SOFA: OR 1.00 (95% CI 0.41-2.45), p=0.98 for cfDNA, OR 2.4 (95% CI 1.37-4.21), p=0.002 for SOFA and OR 2.00 (95% CI 0.94-4.28), p=0.072 for lactate. Negelkerke R Square was 0,633 for cfDNA. In addition, area under the curve for cfDNA mortality was 0.60 (95% CI 0.46-0.73) and SOFA 0.81 CI 95% 0.19-0.91).

Conclusions: Our study suggests that cfDNA and qSOFA have worse prognostic accuracy when compared to lactate and SOFA, variables already used in clinical practice and easily measured.

P014 LCN2 expression correlates with organ failure in surgical patients with infection

M Martín-fernández1, R Almansa1, S Montero2, J Almeida Cristo-Barbosa3, A Ortega1, A Hernández Valero3, E Gómez Sánchez4, E Gómez Pesquera4, J Rico-Feijoo2, MC Esteban-Velasco5, JM Calvo-Vecino3, M Vaquero3, C Aldecoa2, E Tamayo4, J Bermejo-Martín1

1Hospital Clínico Universitario de Valladolid/IECSCYL, BioSepsis (Group for Biomedical Research in Sepsis), Valladolid, Spain; 2Hospital Universitario Río Hortega, Anesthesiology and Reanimation Service, Valladolid, Spain; 3Hospital Universitario de Salamanca, Anesthesiology and Reanimation Service, Salamanca, Spain; 4Hospital Clínico Universitario de Valladolid, Anesthesiology and Reanimation Service, Valladolid, Spain; 5Hospital Universitario de Salamanca, Department of General and Gastrointestinal Surgery, Salamanca, Spain

Introduction: The aim of this study is to develop a “molecular equivalent” to Sequential Organ Failure Assessment (SOFA) score, which could identify organ failure in an easier, faster and more objective manner, based on the evaluation of Lipocalin-2 (LCN2/NGAL) expression levels by using droplet digital PCR (ddPCR). Sepsis has been classically defined as the exuberant, harmful, pro-inflammatory response to infection. This concept is changing [1] and the presence of a life-threatening organ dysfunction caused by a dysregulated host response to infection is now considered a central event in the pathogenesis of sepsis [2].

Methods: LCN2 expression levels were quantified by ddPCR in blood of a total of 257 surgical patients with a diagnosis of infection. Spearman analysis was used to evaluate if LCN2 correlated in a significant manner with SOFA score. Area under the receiver operating curve (AUROC) analysis and multivariate regression analysis were employed to test the ability of LCN2 to identify organ failure and mortality risk.

Results: Spearman analysis showed that there was a positive, significant correlation between LCN2 expression levels and SOFA score (Fig. 1). AUROCs analysis showed that LCN2 presents a good diagnostic accuracy to detect organ failure and mortality risk (Fig 2). In the multivariate regression analysis, patients showing LCN2 expression levels over the Optimal Operating Points (OOPs) identified in the AUROCs showed a higher risk of developing organ failure (Table 1) and a higher mortality risk (Table 2).

Conclusions: Quantifying LCN2 expression levels by ddPCR is a promising approach to improve organ failure detection and mortality risk in surgical patients with infection.


1. Bermejo-Martin JF et al. J Infect 72: 525–536, 2016.

2. Singer M et al. JAMA 315: 801–810, 2016.

Table 1 (abstract P014). Multivariate analysis for evaluating the risk of organ failure based on the LCN2 expression levels. Adjusting variables were [age], [chronic cardiac disease], [cancer], [immunosuppression], [hypertension], [chronic respiratory disease], [chronic renal failure], [respiratory focus], [abdomen focus]
Table 2 (abstract P014). Multivariate analysis for evaluating the risk of mortality based on the LCN2 expression levels. Adjusting variables were [age], [chronic renal failure], [diabetes], [respiratory focus]
Fig. 1 (abstract P014).

Dot plot showing the correlation between LCN2 expression levels and SOFA score in surgical patients with infection

Fig. 2 (abstract P014).

AUROCs for differential diagnosis of organ failure and mortality in surgical patients with infection

P015 Prospective validation of an 18-mRNA score for diagnosis of infection in critically ill patients

TE Sweeney1, AR Moore2, J Roque2, B Shaller2, T Asuni2, JE Levitt3, JG Wilson2, P Khatri4, M Remmel1, D Rawling1, O Liesenfeld1, AJ Rogers2

1Inflammatix, Burlingame, United States; 2Stanford University, Medicine, Palo Alto, United States; 3Stanford University, Levitt, Palo Alto, United States; 4Stanford University, Immunity, Transplantation and Infections, Palo Alto, United States

Introduction: There is an urgent need for improved diagnostics for acute infections to help physicians decide whether to treat with antibiotics. We previously described an 18-host-mRNA diagnostic consisting of an 11-mRNA Sepsis MetaScore (SMS) to determine the presence of an infection, and a 7-mRNA bacterial-viral score (BVS) to discriminate between bacterial and viral sources [1,2].

Methods: We collected PAXgene™ RNA blood from 165 patients enrolled in the Stanford ICU Biobank from 2016-2017. Infection status was adjudicated by Stanford physicians post-hoc using clinical data from the electronic medical record. Low-RNA samples were removed, and target mRNAs were quantitated using NanoString nCounter™ and difference-of-geometric-mean mRNA scores were calculated by Inflammatix [2], blinded to clinical phenotyping. Primary outcome was performance of the two scores in correctly diagnosing physician-adjudicated bacterial or viral infection. Secondary outcome was comparison to procalcitonin (PCT), which was drawn only according to treating physician preference.

Results: Of 165 patients, physicians adjudicated patients as: 29 noninfected, 102 infected (71 bacterial, 14 viral, 2 fungal, 15 mixed infections), and 33 with uncertain status. The SMS had an AUROC of 0.83 for separating infection of any type from noninfected status. The BVS had an AUROC of 0.95 for separating bacterial from viral infection. Both SMS and BVS were substantially better than PCT across all adjudicated patients and in matched pairs (PCT AUROC 0.7 for any infection vs noninfected; PCT AUROC 0.82 for bacterial vs. viral; Fig. 1). When used together, the SMS and BVS were able to separate patients based on infection status (Fig. 2).

Conclusions: We prospectively validated an 18-mRNA host-response diagnostic module in a blinded, independent study, confirming high accuracy for the presence and type of infection in a critically ill population.


1) Sweeney TE et al, Sci Transl Med, 287ra71, 2015

2) Sweeney TE et al, Sci Transl Med, 346ra91, 2016

Fig. 1 (abstract P015).

ROC curves of the infection-diagnostic Sepsis MetaScore (SMS) and the bacterial-vs-viral BVS score, shown both for all patients, and in head-to-head comparison with procalcitonin

Fig. 2 (abstract P015).

Plotting patients according to transcriptomic scores in two dimensions (x-axis, SMS; y-axis, BVS) shows accurate discrimination by adjudicated infection type

P016 Low IgG and IgM levels do not predict mortality in cancer patients with septic shock

G Oliveira, C Park, J Almeida, M Mourão, S Rizk, J Fukushima, L Hajjar

Instituto do Cancer do Estado de São Paulo, Intensive care unit, São Paulo, Brazil

Introduction: Sepsis is an inflammatory state due to an exacerbated immune response against infection. In cancer patients, sepsis presents a 10-fold higher mortality than in general population and leads to longer intensive care unit (ICU) and hospital lengths of stay. It has been shown that reduced levels of circulating immunoglobulins (Ig) might be a surrogate marker of unfavorable outcome in sepsis [1]. The aim of this study was to evaluate the association between Ig levels in plasma and 60-day mortality rate in cancer patients with septic shock.

Methods: From December 2017 to November 2018, we conducted a prospective study in the intensive care unit (ICU) of Cancer Institute of State of Sao Paulo, an 84-bed ICU linked to University of Sao Paulo. Patients ≥18 years old with cancer and septic shock were enrolled. Descriptive statistics were computed for demographic and outcome variables. Laboratory data and Ig levels were collected at ICU admission and at days 1, 2 and 3. A multivariate analysis was performed to evaluate predictors of 60-day mortality.

Results: A total of 190 patients were included in the study. The 30-day and 60-day mortality were 40.5% and 45.3%, respectively. No significant differences in IgM and IgG levels were observed between survivors and non-survivors. In both groups, the median IgM levels were low and the median IgG levels were normal. In the multivariate analysis for 60-day mortality, a favorable status performance measured by the Eastern Cooperative Oncology Group (ECOG) was associated with better survival; metastatic disease, higher Sequential Organ Failure Assessment (SOFA) score at admission and higher levels of initial lactate were associated with increased mortality.

Conclusions: Low levels of serum endogenous immunoglobulins are not predictors of 60-day mortality in cancer patients with septic shock.


1. Bermejo-Martín JF et al. J Intern Med. 276(4):404-12, 2014.

P017 Measurement of plasticity (deformability) of neutrophils and monocytes provides a rapid and early indicator of sepsis

R Sheybani1, M Hem1, A Shah1, M Samoszuk1, H Omran1, T Caffery2, C Thomas2, H Tse1, H O’Neal2

1Cytovale, San Francisco, United States; 2Louisiana State University Health Sciences Center, Baton Rouge, Louisiana, United States

Introduction: Cytovale has developed a rapid biophysical assay of the host immune response which can serve as a rapid and reliable indicator of sepsis. Neutrophils and monocytes undergo characteristic structural and morphologic changes in response to infection. One type of response is the generation of neutrophil extracellular traps (NETs), these have been proposed as potential mediators for widespread tissue damage. During NETosis there is a fundamental reorganization of a cell’s chromatin structure – a signal that we have shown is sensitively measured by the Cytovale cytometer. We hypothesized that quantification of plasticity (deformability) of leukocytes in the peripheral blood provides an early indicator of sepsis. The Cytovale assay uses microfluidic cytometry to measure the plasticity of up to 100,000 white blood cells from EDTA-anticoagulated, peripherally-collected whole blood and provides a result in 5 minutes.

Methods: In two prospective studies conducted in two academic medical centers in Baton Rouge, LA, the Cytovale test was performed on peripheral blood samples obtained from 500 patients who presented to the emergency department with signs or symptoms suggestive of infection. The two studies included high acuity patients (400 patient study) and low acuity patients (100 patient study). An adjudicated reference diagnosis of sepsis or no sepsis was established for each subject, using consensus definitions, by review of the complete medical records.

Results: The Receiver Operator Curve (ROC) performance of the Cytovale assay for both studies demonstrated an Area Under the Curve (AUC) greater than 0.85 (Fig. 1).

Conclusions: Measurement of neutrophil and monocyte plasticity by a novel assay provides an accurate and rapid indication of sepsis in patients who present to an emergency room with signs or symptoms of infection.

Fig. 1 (abstract P017).

ROC performance of Cytovale test

P018 Plasma hepatocyte growth factor in sepsis and its association with mortality: a prospective observational study

F Peng, C Liang, W Chang, Q Sun, J Xie, H Qiu, Y Yang

Zhongda Hospital, School of Medicine, Southeast University, Department of Critical Care Medicine, Nanjing, China

Introduction: Sepsis and septic shock are commonly associated with endothelial cell injury. Hepatocyte growth factor (HGF) is a multifunctional protein involved in endothelial cell injury and plays a pivotal role in sepsis. This study assesses its correlation with relevant endothelial cell injury parameters and prognostic value in patients with sepsis.

Methods: A prospective, observational cohort study was conducted in patients with sepsis admitted to the department of critical care medicine at the Zhongda Hospital from November 2017 to March 2018. The plasma HGF level was collected on the first 24h after admission (day 1) and day 3, then was measured by enzyme-linked immunosorbent assay. The primary endpoint was defined as all-cause 28-day mortality. Furthermore, we analyzed the correlation of HGF with relevant endothelial cell injury markers.

Results: Eighty-six patients admitted with sepsis were included. HGF levels of non-survivors were elevated upon day 1 (1940.62 ± 74.66pg/mL vs. 1635.61 ± 47.49pg/mL; p = 0.002) and day 3 (1824.82 ± 137.52pg/mL vs. 1309.77 ± 83.49pg/mL; p = 0.001) compared with that in survivors, and showed a strong correlation with von Willebrand factor (r = 0.45, p <0.0001), lactate (r = 0.35, p = 0.0011), pulmonary vascular permeability index (r = 0.38, p = 0.0241), first 24 h fluid administration (r = 0.38, p <0.0001) and sequential organ failure assessment score (r = 0.40, p = 0.0001) (Fig. 1). Plasma levels were able to discriminate prognostic significantly on day 1(AUC: 0.72, 95%CI: 0.60-0.84) and day 3 (AUC: 0.77, 95%CI: 0.63-0.91) (Fig. 2).

Conclusions: HGF levels are associated with sepsis and are correlated with established markers of endothelial cell injury. Elevated HGF level in sepsis patients is a predictor of mortality.

Fig. 1 (abstract P018).

Correlation between HGF levels with markers of endothelial cell injury and SOFA

Fig. 2 (abstract P018).

Prognostic value of HGF level for sepsis patients

P019 Calprotectin, a powerful biomarker for the diagnosis of bacterial infections

A Havelka1, P Venge2, A Larsson2

1Department of Molecular Medicine and Surgery, Karolinska Institute, Stockholm, Sweden; 2Department of Medical Sciences, Uppsala University, Uppsala, Sweden

Introduction: The rapidly growing problem with antibiotic resistance has resulted in demand for more specific and restricted use of antibiotics. Biomarkers which can diagnose an infection in early stage and distinguish between bacterial and viral infections could possibly reduce the use of antibiotics. Calprotectin is one of the most abundant proteins in the cytosol of neutrophil granulocytes and is released upon activation of neutrophils. The aim of this study was to investigate the performance of calprotectin as a marker for bacterial infection and its possibility to distinguish between bacterial and viral infections.

Methods: The study group consisted of 432 subjects including 144 healthy, noninfected, control patients and 288 patients with confirmed etiology of their infections, 185 patients with bacterial infection, 54 with viral infection, 26 with mycoplasma infection, and 23 with a bacterial infection as a secondary infection to influenza. Calprotectin was measured in serum samples with a particle enhanced turbidimetric assay (Gentian AS, Norway). Heparin Binding Protein (HBP) and Procalcitonin were analyzed by sandwich immunoassays (Hycult Biotech and Thermo Fisher Scientific).

Results: Performance of Calprotectin in the diagnosis of bacterial infections as well as in distinguishing between bacterial and viral infections was compared to performance of Procalcitonin and HBP. Calprotectin was superior in diagnosis of bacterial infections as well as in differentiating bacterial from viral infections. Results are presented in Table 1. Interestingly, calprotectin was the only biomarker with ability to distinguish between mycoplasma and viral infections.

Conclusions: Calprotectin is a promising biomarker for diagnosis of bacterial infections. Our results indicate that Calprotectin is superior to Procalcitonin and HBP in diagnosis of bacterial infections and in differentiation between bacterial and viral infections including mycoplasma infections.

Table 1 (abstract P019). Diagnostic performance of the studied biomarkers

P020 A personalized randomized trial of validation and restoration of immune dysfunction in severe infections and sepsis: PROVIDE

N Antonakos1, I Tsangaris2, D Markopoulou3, N Rovina4, A Prekates5, E Antoniadou6, V Theodorou7, A Stefos8, G Vlachogianni9, E Giamarellos-Bourboulis1, G Dimopoulos2, M Netea10

1National and Kapodistrian University of Athens, 4th Department of Internal Medicine, Athens, Greece; 2National and Kapodistrian University of Athens, 2nd Department of Critical Care Medicine, Athens, Greece; 3KAT General Hospital, Intensive Care Unit, Athens, Greece; 4National and Kapodistrian University of Athens, Intensive Care Unit 1st Department of Respiratory Medicine, Athens, Greece; 5Tzaneion General Hospital, Intensive Care Unit, Piraeus, Greece; 6G.Gennimatas General Hospital, Intensive Care Unit, Thessaloniki, Greece; 7Dimocriteion University of Thrace, Intensive Care Unit, Alexandroupolis, Greece; 8University of Thessaly, Department of Internal Medicine, Larissa, Greece; 9Aghios DImitrios General Hospital, Intensive Care Unit, Thessaloniki, Greece; 10Radboud University, Department of Internal Medicine, Nijmegen, Netherlands

Introduction: Based on the recent post-hoc analysis of previous trials of the efficacy of anakinra in patients with macrophage-like activation syndrome (MALS), PROVIDE ( registration NCT03332225) is the first double-blind, double-dummy on-going trial aiming to the impact of immunotherapy according to personalized needs.

Methods: Adult patients with septic shock by the Sepsis-3 classification due to lung infection or primary bacteremia or acute cholangitis are screened using two consecutive measurements of ferritin and of HLA-DR/CD14 co-expression for MALS (ferritin above 4,420 ng/ml) or immunosuppression (HLA-DR/CD14 less than 30%) and randomized into immunotherapy with either anakinra (targeting MALS) or recombinant IFNγ (targeting immunosuppression) and into placebo treatment. Main exclusion criteria are primary and secondary immunodeficiencies and solid and hematologic malignancies.

Results: 101 patients have been screened so far. Most common infections are community-acquired pneumonia (41.6%), hospital-acquired pneumonia (26.7%) and primary bacteremia (12.9%). Mean +/- SD SOFA score is 12.6 +/- 2.9 and Charlson’s comorbidity index 5.21 +/- 2.44; 25 patients have MALS (24.8%); two immunosuppression (2%); the majority remain unclassified for immune state.

Conclusions: Current screening suggests greater frequency of MALS than recognized so far in a setting of septic shock due to lung infection or primary bacteremia or acute cholangitis.

P021 Development of an algorithm to predict mortality in patients with sepsis and coagulopathy

D Hoppensteadt1, A Walborn2, M Rondina3, J Fareed1

1Loyola University Medical center, Pathology, Maywood, United States; 2Loyola University Medical center, Pharmacology, Maywood, United States; 3University of Utah and the GRECC, Internal Medicine and the Molecular Medicine Program, Salt Lake City, United States

Introduction: Sepsis is a systemic response to infection which involves inflammation, infection response, hemostatic dysregulation, endothelial dysfunction, and platelet activation. The purpose of this study was to develop an equation incorporating biomarker levels at ICU admission to predict mortality in patients with sepsis, to test the hypothesis that using a combination of biomarkers of multiple systems would improve predictive value.

Methods: Plasma samples were collected from 103 patients with sepsis at the time of ICU admission. Biomarker levels were measured using commercially available, ELISA methods. Clinical data, including the ISTH DIC score, SOFA score, and APACHE II score were also collected. 28-day mortality was used as the primary endpoint. Stepwise linear regression modeling was performed to generate a predictive equation for mortality.

Results: Differences in biomarker levels between survivors were quantified and using the Mann-Whitney test and the area under the receiver operating curve (AUC) was used to describe predictive ability. Significant differences (p<0.05) were observed between survivors and non-survivors for PAI-1 (AUC=0.70), procalcitonin (AUC=0.77), HMGB-1 (AUC=0.67), IL-6 (AUC=0.70), IL-8 (AUC=0.70), protein C (AUC=0.71), Angiopoietin-2 (AUC=0.76), endocan (AUC=0.58), and platelet factor 4 (AUC=0.70). A predictive equation for mortality was generated using stepwise linear regression modeling. This model incorporated procalcitonin, VEGF, the IL-6:IL-10 ratio, endocan, and PF4, and demonstrated a better predictive value for patient outcome than any individual biomarker (AUC=0.87).

Conclusions: The use of a mathematical modeling approach resulted in the development of a predictive equation for sepsis-associated mortality with performance than any individual biomarker or clinical scoring system. Furthermore, this equation incorporated biomarkers representative of multiple physiological systems that are involved in the pathogenesis of sepsis.

P022 The effects of biomarker clearances as markers of improvement of severity in abdominal septic shock during blood purification

T Taniguchi1, K Sato2, M Okajima2

1Kanazawa University, Anesthesiology and Intensive Care Medicine, Kanazawa, Japan; 2Kanazawa University Hospital, Intensive Care Unit, Kanazawa, Japan

Introduction: Recently the clearances of biomarkers (BM) such as procalcitonin (PCT) and presepsin (p-SEP), appear to be better indicators than single cutoff values to diagnose septic complications or predict outcomes. Moreover, blood purification (BF) such as endotoxin absorption therapy (PMX) and continuous renal replacement therapy (CRRT) has been carried out for abdominal septic shock (ASS). However, there are few studies about BM clearances in ASS during BF. Therefore, the present study retrospectively evaluated the effects of BM clearances on improvements of severity in critical patients with ASS during BF.

Methods: Thirty-three patients (M/F 21/12, mean age 69 years) were entered. Septic shock was defined in sepsis-3 criteria. PMX was undergone twice and CRRT was undergone for 5 days. BM levels were measured for 5 days after ICU admitted. Moreover, SOFA scores were measured for 5 days after ICU admitted. BM clearances were determined at the entering ICU and 1, 3, and 5 days after ICU admitted. The improvements of severity were determined the differences of SOFA scores after ICU admitted. Primary outcome is the correlation between BM clearances and the improvement of severity. Secondary outcomes are the changes of BM after ICU admitted, and mortality in ICU.

Results: Two of 33 patients died after ICU admission. PCT and lactate levels were improved 5 days after ICU admitted (61.1 to 10.8 ng/mL; p<0.05, 4.2 to 1.1 mmol/L; p<0.05). CRP and p-SEP levels were not improved. SOFA scores decreased (13.4 to 6.5; p<0.05) at 5 days after ICU admitted. There were significantly correlations between PCT and lactate clearances and the improvement of severity (Y=4.02 +0.11X; R2=0.07, p=0.0102, Y=2.75 +0.03X; R2=0.226, p<0.0001). There were not significantly correlations between CRP and p-SEP clearances with the improvement of severity.

Conclusions: The present study showed that PCT and lactate clearances significantly correlated the improvement of severity in critical patients with ASS during BF.

P023 Decreased thrombin generation potential is associated with increased thrombin generation markers in sepsis associated coagulopathy

J Fareed1, F Siddiqui1, R Laddu1, D Hoppensteadt1, M Rondina2, E Brailovsky3

1Loyola University Medical Center, Pathology, Maywood, United States; 2University of Utah School of Medicine, Department of Internal Medicine, Salt Lake City, United States; 3Loyola University Medical Center, Cardiology, Maywood, United States

Introduction: Sepsis associated coagulopathy (SAC) is commonly seen in patients which leads to dysfunctional hemostasis. The purpose of this study is to determine the thrombin generation potential of baseline blood samples obtained from SAC patients and demonstrate their relevance to thrombin generation markers.

Methods: Baseline citrated blood samples were prospectively collected from 49 patients with SAC at the University of Utah clinic. Citrated normal controls (n=50) were obtained from George King Biomedical (Overland Park, KS). Thrombin generation studies were carried out using a flourogenic substrate method. TAT and F1.2 were measured using ELISA methods (Seimens, Indianapolis, IN). Functional antithrombin levels were measured using a chromogenic substrate method.

Results: The peak thrombin levels were lower (82 ± 40nM) in the DIC patients in comparison to higher levels observed in the normal plasma (133 ± 10nM). The AUC was lower (561 ± 280) in the DIC group in comparison to the normals (624 ± 18). The DIC group showed much longer lag time (4.1 ± 2.1) in comparison to the normal group (2.1 ± 2.2). Wide variations in the results were observed in these parameters in the DIC group. The F1.2 levels in the DIC group were much higher (570±48 pmol) in comparison to the normal (210 ± 25 pmol). The TAT levels also increased in the DIC group (27.9 ± 5.1 ng/ml) in comparison to the normal (2.8 ± 0.8 ng/ml). The functional antithrombin levels were decreased in the DIC group (64 ± 11%).

Conclusions: These results validate that thrombin generation such as F1.2 and TAT are elevated in patients with DIC. However thrombin generation parameters are significantly decreased in this group in comparison to normals. This may be due to the consumption of prothrombin due to the activation of the coagulation system. The decreased functional AT levels observed in the DIC group are due to the formation of the complex between generated thrombin and antithrombin.

P024 Relationship of markers of inflammation, infection, and endothelial function to mortality and severity of coagulopathy in patients with sepsis-associated DIC

D Hoppensteadt1, A Walborn2, M Rondina3, J Fareed1

1Loyola University Medical center, Pathology, Maywood, United States; 2Loyola University Medical center, Pharmacology, Maywood, United States; 3University of Utah School of Medicine, Internal medicine, Salt Lake City, United States

Introduction: Sepsis-associated disseminated intravascular coagulation (DIC) is a complex clinical scenario involving derangement of many processes, including hemostasis. Assessment of markers including inflammation, endothelial function, and endogenous anticoagulants may provide insight into DIC pathophysiology and lead to improved methods for assessment of patient condition and response to treatment.

Methods: Citrated plasma samples were collected from 102 patients with sepsis and suspected DIC at ICU admission and on days 4 and 8. DIC score was determined using the ISTH scoring algorithm (e.g. platelet count, PT/INR, fibrinogen and D-Dimer). CD 40 Ligand (CD40L), Plasminogen inhibitor 1 (PAI-1), nucleosomes, Procalcitonin (PCT), Microparticle tissue factor (MP-TF) and Prothrombin 1.2 (F1.2) were measured using commercially available ELISA kits. Protein C activity was measured using a clot-based assay. Interleukin 6 (IL-6), Interleukin 8 (IL-8), Interleukin 10 (IL-10), Tumor necrosis factor alpha (TNFα), and Monocyte chemoattractant protein (MCP-1) were measured using biochip technology.

Results: Significant differences in levels of Protein C (p=0.009), PCT (p=0.0005), IL-6 (p=0.019), IL-8 (p=0.0149), PAI-1 (p=0.015), were observed between survivors and non-survivors. Significant variation of Protein C (p=0.002), nucleosomes (p=0.05), PCT (p<0.0001), IL-6 (p=0.001), IL-8 (p=0.003), IL-10 (p=0.011), TNFα (p=0.021) and MCP-1 (p=0.021) were observed based on severity of DIC score.

Conclusions: Markers from multiple systems perturbed in DIC were associated with mortality, suggesting that while these systems may not be routinely evaluated in the normal course of patient care, dysfunction of these systems contributes significantly to mortality. In addition, numerous inflammatory cytokines showed an association with DIC score. This suggests that the measurement of additional markers in sepsis-associated DIC may be of value in the prediction of mortality and may be helpful in guiding treatment for these patients.

P025 Usefulness of plasminogen activator inhibitor-1 (PAI-1) as a predictive marker for identification of sepsis-induced DIC

J Maruyama, K Hoshino, Y Irie, S Miyagawa, R Hokama, M Koie, M Nakashio, Y Kawano, T Kitamura, H Ishikura

Fukuoka University Hospital, Department of Emergency and Critical Care Medicine, Fukuoka, Japan

Introduction: There is a crosstalk between inflammation and coagulation and disseminated intravascular coagulation (DIC) especially sepsis-induced DIC is a one of the most significant causes of mortality in intensive care units. Meanwhile, there are various types of DIC and sepsis-induced DIC is a type of suppressed fibrinolysis DIC. In this study, we aimed to identify coagulation/fibrinolysis markers useful for discriminating whether sepsis-induced DIC or not.

Methods: This is a single-center retrospective observational study of 233 patients with DIC according to the Japanese Association for Acute Medicine (JAAM) DIC criteria (JAAM DIC score ≥4) from July 2017 to June 2018. We divided DIC patients into sepsis and non-sepsis using Sepsis-3 diagnosed criteria and univariate and multivariate logistic regression analyses were performed to identify an independent predictive marker of sepsis-induced DIC among coagulation/fibrinolysis markers on ICU admission.

Results: Sepsis-induced DIC (S-DIC) group (n=62) was significantly higher DIC score and SOFA score rather than non-sepsis-induced DIC (NS-DIC) group (n=171) [DIC score; 5 (5-7) vs. 4 (4-5), P<0.01. SOFA score; 11 (9-14) vs. 6 (4-10), P<0.01.].

About coagulation/fibrinolysis markers, S-DIC group was significantly lower the FDP, D-dimer, and PIC rather than NS-DIC group (P<0.01), and higher the PAI-1 (P<0.01). Moreover, PAI-1 was identified as one of the independent predictive markers of sepsis-induced DIC by multivariate logistic regression.

Conclusions: Recently, we reported that PAI-1 was a useful predictive marker of mortality in sepsis. From this study we suspected that PAI-1 is a useful marker for discriminating sepsis-induced DIC. Furthermore, we confirmed that sepsis-induced DIC was a type of DIC with suppressed fibrinolysis. Therefore, we recommend measuring PAI-1 against sepsis and sepsis-induced DIC patients.

P026 Does EAA value reflect severity of condition in patients admitted to intensive care unit?

T Ikeda1, S Ono1, S Suda1, T Nagura1, M Tomino2, M Sugi2, Z Wajima2

1Tokyo Medical University, Hachioji Medical University, Division of Critical Care Medicine, Tokyo, Japan; 2Tokyo Medical University, Hachioji Medical University, Department of Anesthesiology, Tokyo, Japan

Introduction: The Endotoxin Activity Assay (EAA) is a rapid immunodiagnostic test based on chemiluminescence. It was approved by the FDA in 2003 as a diagnostic reagent for risk assessment of severe sepsis in the ICU. Ascertaining endotoxin levels in the bloodstream is important in targeting patients and determining the appropriate timing for initiation of treatment. It has high sensitivity and specificity for endotoxin, and is considered to be useful in predicting clinical symptoms and determining prognosis. The usefulness of the EAA has yet to be fully clarified.

Methods: A total of 142 patients admitted to the ICU between January 2014 and June 2018 with suspected sepsis or sepsis were enrolled. The EAA was conducted within 24 hr after admission. Patient characteristics were determined, together with levels of IL-6, procalcitonin, presepsin, and PaO2/FiO2. Thereafter, the patients were classified into 5 groups depending on their EAA value: 1) < 0.2; 2) from ≤ 0.2 to < 0.4; 3) from ≤ 0.4 to < 0.6; 4) from ≤ 0.6 to < 0.9; and 5) ≤0.9). The transition of various markers was also examined. The Spearman rank correlation, Wilcoxon rank sum test, and a non-repeated ANOVA were used for the statistical analysis. A P-value of < 0.05 was considered statistically significant.

Results: The EAA values showed a positive correlation with both the APACHE II (r=0.48) and SOFA scores (r=0.56)(P<0.01), although that with the latter was stronger. A significant correlation was also observed with levels of procalcitonin (r=0.45) and presepsin (r=0.51). The EAA showed a high value (P<0.01) only in patients showing a positive result for blood culture; no other marker showed a significant difference between those showing a positive or negative result. At 28 days, significantly higher values were observed in the non-survival group (P<0.05) in values for the EAA and other markers.

Conclusions: EAA value tended to correlate with disease severity (APACHEII and SOFA scores) in patients admitted to the ICU.

P027 Intraperitoneal microdialysis detects early peritonitis caused by intraabdominal bacterial infection in a pig model

S Hødnebø, S Pischke, A Barratt-Due, E Lindholm, T Tønnessen

Oslo University Hospital, Division of Emergencies and Intensive Care, Cardiothoracic clinic, Oslo, Norway

Introduction: Common complications following abdominal surgery are intestinal leaks, with subsequent abdominal sepsis. Early diagnosis is important to allow early intervention. The current clinical methods are insufficient for early detection. We hypothesized that intraperitoneal microdialysis allows detection of peritonitis prior to changes in standard clinical parameters in a pig model.

Methods: Bacterial peritonitis was induced in 5 pigs by bowel perforation and intraperitoneal fecal instillation, one pig underwent sham surgery. Intraperitoneal microdialysis catheters were placed in each abdominal quadrant. The observation time was 10 hours.

Results: In peritonitis pigs the intraperitoneal lactate increased during the first two hours and remained elevated throughout the observation time (Table 1), whereas the arterial lactate remained within reference range (<1.6 mM). Intraperitoneal glucose decreased significantly. Hemodynamics were hardly influenced during the first two hours, and decreased thereafter. Sham surgery did not influence in any of the parameters.

Conclusions: A rapid and pronounced increase in intraperitoneal lactate and decrease in intraperitoneal glucose was observed after instillation of intraabdominal feces. Systemic lactate increase was absent, and the hemodynamic response was delayed. Postoperative intraperitoneal microdialysis is applicable in detecting peritonitis earlier than standard clinical monitoring and should be evaluated in a clinical study in order to explore if early intervention based on MD data will reduce ICU length of stay, morbidity and mortality.

Table 1 (abstract P027). Intraperitoneal microdialysis lactate values, arterial lactate values, mean arterial pressure (MAP) and cardiac output (CO) at different time points after induced bacterial peritonitis in pigs

P028 Validation of B·R·A·H·M·S PCT direct, a new sensitive point-of-care device for rapid measurement of procalcitonin

L Velly1, P Hausfater1, M Seidel2, J Lotz3, C Rechner4, C Brochet5, M Oppert6

1Emergency dpt Pitié-Salpêtrière hospital APHP and Sorbonne Universités GRC-14 BIOSFAST, Paris, France; 2Unfallkrankenhaus Berlin, Berlin, Germany; 3Universitätsmedizin Johannes Gutenberg-Universität Mainz, Mainz, Germany; 4Thermo Fisher Scientific, Hennigsdorf, Germany; 5Biochemistry dpt Pitié-Salpêtrière hospital, APHP, Paris, France; 6Klinikum Ernst von Bergmann gGmbH, Potsdam, Germany

Introduction: Procalcitonin (PCT) is a highly sensitive and specific biomarker for bacterial infection.

B·R·A·H·M·S PCT direct is a new point of care test (POC) for fast measurement of PCT in whole blood (capillary or venous) with a measuring range of 0.1-10 μg/L.

Methods: This multicentre study examined the correlation of capillary and venous whole blood samples measured on B·R·A·H·M·S PCT direct compared to established reference methods B·R·A·H·M·S PCT sensitive KRYPTOR and Elecsys B·R·A·H·M·S PCT using EDTA plasma. The design was based on the related CLSI Guidelines EP09-A3.

Results: 279 patients for venous EDTA whole blood and 93 patients for capillary blood (fingertip) were included in this study. The Pearson correlation coefficient (log transformed) between venous or capillary whole blood and the reference method was r2=0.95. The concordance to reference methods was 93% for venous blood and 95% for capillary blood related to a clinical cut-off of 0.5 μg/L, with a sensitivity of 92% (venous and capillary blood) and a specificity of 95% (venous blood) and 97% (capillary blood), respectively. The concordance at the clinical cut-off 0.25 μg/L was 93% for venous blood and 91% for capillary blood. No significant bias was observed compared to the reference method.

Conclusions: This study found a very good correlation and diagnostic accuracy of the new, sensitive POC device for rapid measurement of PCT (TAT 20 min). The B·R·A·H·M·S PCT direct test allows an accurate measurement of PCT in whole blood at a point of care comparable to established lab based B·R·A·H·M·S PCT assays.

P029 Is procalcitonin a reliable biomarker for infection in trauma patients? Our experience

E De sanso1, C Morena2, MA Palazzo3, E Gamberini2, F Avolio4, E Russo3, V Agnoletti2

1University of Piemonte Orientale A. Avogadro, Anesthesia and Intensive Care, Turin, Italy; 2Anesthesia and Intensive Care AUSL Romagna Anesthesia and Intensive Care Unit, AUSL Romagna Anesthesia and Intensive Care Unit, Cesena, Italy; 3Anesthesia and Intensive Care AUSL Romagna Anesthesia and Intensive Care Unit, Anesthesia and Intensive Care AUSL Romagna Anesthesia and Intensive Care Unit, Cesena, Italy; 4Anesthesia and Intensive Care AUSL Romagna Anesthesia and Intensive Care Unit, Cesena, Italy

Introduction: Procalcitonin (PCT) is a serum biomarker suggested by the Surviving Sepsis Campaign to aid in determination of the appropriate duration of therapy in septic patients. Trauma patients have a high prevalence of septic complications, often difficult to distinguish from inflammatory response. PCT values typically declined after 72h from trauma and increased only during secondary systemic bacterial infections. The aims of the study are to evaluate reliability and usefulness of PCT serum concentration in trauma.

Methods: We retrospectively analyzed data from 40 trauma patients admitted to ICU at Bufalini Hospital - Cesena, from July 2017 to August 2018. We collected data about antimicrobial therapy, Injury severity score (ISS), first arterial Lactate in emergency room, SOFA score and Sepsis severity. Plasma PCT concentration was measured using an automate analyzer (Modular E-Brahms) on 1st day of antimicrobial therapy and every 48h hours. Antimicrobial therapy was stopped according to a local protocol; however medical judgment was considered the overriding point for therapeutic decision.

Results: Median ISS of patients was 33.5, inter quartile range (IQR) 13.5. PCT mean concentration at the starting of antimicrobial treatment was 13.07 μg/L (d.s 40.1), median 0.72 (IQR 10.13). No significative correlation (Spearman´s Rho Test) was found between PCT at day 1 of antimicrobial therapy and ISS (Rho -0.186), between first arterial Lactate in ER and PCT (Rho 0.158). Daily course of PCT was not related to distance from trauma (Rho -0.116). In 21 of 40 patients (52.2 %) PCT measurement led physician to save days of antimicrobial therapy compared with standard clinical practice. We couldn´t find any cut off value.

Conclusions: Our experience suggests that PCT could help physician to optimize duration of antimicrobial therapy in trauma patients. No standard approach can be recommended at present.

P030 Α randomized prospective clinical trial to assess the role of procalcitonin (PCT)-guided antimicrobial therapy to reduce long-term infections sequelae (PROGRESS)

E Kyriazopoulou1, A Panagaki1, L Liaskou1, E Drakou2, K Marousis2, V Apostolopoulos3, A Makina4, V Kolonia5, S Lagou6, EJ Giamarellos-Bourboulis1

14th Department of Internal Medicine, National and Kapodistrian University of Athens, Athens, Greece; 21st Department of Internal Medicine, G.Gennimatas General Hospital, Athens, Greece; 31st Department of Internal Medicine, Thriasio Elefsis General Hospital, Athens, Greece; 42nd Department of Internal Medicine, Thriasio Elefsis General Hospital, Athens, Greece; 52nd Department of Internal Medicine, Sismanogleio General Hospital, Athens, Greece; 63rd Department of Internal Medicine, National and Kapodistrian University of Athens, Athens, Greece

Introduction: Long duration of antimicrobial treatment may predispose to colonization and subsequent infections by multidrug-resistant organisms (MDRO) and Clostridium difficile. PROGRESS ( registration NCT03333304) is an on-going trial aiming to use PCT for the restraining of this calamity.

Methods: Adult patients with sepsis by the Sepsis-3 classification and any of five infections (pneumonia community-acquired; hospital- acquired or ventilator-associated; acute pyelonephritis; primary bacteremia) are randomized to PCT-guided treatment or standard of care (SOC) treatment. In the PCT arm antibiotics are discontinued when PCT on or after day 5 is decreased by more than 80% of the baseline or remains below 0.5 ng/ml; in the SOC arm antibiotics are discontinued at the discretion of the attending physician. Patients are followed for six months. Primary endpoint is the rate of infections by MDRO and/or C.difficile or death. Serial stool samples are cultured for MDRO and screened for glutamate dehydrogenase antigen and toxins of C.difficile.

Results: 201 patients have been enrolled so far. Mean ± SD SOFA score is 4.4 ± 2.4. Most common diagnoses are community-acquired pneumonia (57.2%) and acute pyelonephritis (40.3%). At baseline, 10.4% were colonized by MDRO and 3% by C.difficile. Residency in health-care facilities was the only variable associated with C.difficile colonization (odds ratio 7.04; 95% CI: 1.22-40.44). MDRO colonization was associated with residency in health-care facilities (odds ratio 6.74; 95% CI: 2.07-21.93) and hospitalization the last three months (odds ratio 4.29; 95% CI: 1.48-12.45)

Conclusions: The PROGRESS trial is the first trial assessing the probable benefit from PCT guidance to reduce ecological sequelae from long-term antibiotic exposure. Analysis of baseline patient characteristics indicates that PROGRESS is a real-world trial so that results can have major clinical impact.

P031 Prospective multi-site validation of 11-gene host response signature for influenza diagnosis

S Thair1, S Schaffert1, M Shojaei2, T Sweeney3, B Tang2, P Khatri1

1Stanford Univeristy, Biomedical Informatics Research, Stanford, United States; 2Nepean Hospital, Department of Intensive Care Medicine, Sydney, Australia; 3Inflammatix, Burlingame, United States

Introduction: Influenza causes 650 000 deaths per year globally. It caused 80 000 deaths in the US in 2017. Rapid diagnostic tests (RIDTs) have very low sensitivity, whereas RT-PCR based tests have high sensitivity, but requires trained technicians. There are no blood-based diagnostics able to identify influenza infection and distinguish it from other infections. We have previously described a blood-based 11-gene influenza meta-signature (IMS) score to differentiate influenza from bacterial and other viral respiratory infections.

Methods: We prospectively validated the IMS in a multi-site validation study by recruiting 654 individuals (608 patients with suspected influenza, 46 healthy controls) in 10 community or hospital clinics across Australia. We assayed the IMS and 15 genes from viral genome of 3 influenza strains to generate the Blood Flu Score (BFS) as a measure of viremia using Nanostring from whole blood RNA.

Results: Using clinically determined phenotypes, the IMS score distinguished patients with influenza from healthy (AUC=0.95), non-infected (AUC=0.83), bacterial (AUC=0.88), other viruses (AUC=0.77) (Figure 1A). Interestingly, probes of BFS were found in all phenotypic groups (non-infected, bacterial, and other viral infections) to varying degrees, and positively correlate with the IMS score (r=0.53). IMS AUROCs improve when the BFS is used to inform the phenotypic groups: healthy (AUC=0.92), non-infected (AUC=0.90), bacterial (AUC=0.93), other viruses (AUC=0.88) (Figure 1B). Patients who were clinically influenza negative but had a high IMS and BFS were admitted less often, yet had ~4-fold higher mortality than those who were clinically influenza negative with low IMS and no BFS (Table 1).

Conclusions: Collectively, our prospective multi-center validation of the IMS demonstrates its potential in diagnosis of influenza infections.

Table 1 (abstract P031). Summary of outcomes for patients. Clinically negative influenza patients with a high IMS and BFS were admitted less often yet had ~ 4 fold higher mortality than those who were influenza negative with low IMS and no BFS
Fig. 1 (abstract P031).

IMS area under the receiver operator characteristic (AUROC) curves for distinguishing patients with influenza from healthy, non-infected, bacterial, other viruses or all other infections combined when using using (A) clinically informed phenotyping only or (B) using clinical phenotyping and BFS ≥19. All samples with 15 >BFS<19 were removed in (B)

P032 Modulation of innate immune responses in experimental pyelonephritis by preceding osteomyelitis

S Goumenos1, G Renieris2, D Droggiti2, L Sabracos2, P Papagelopoulos1, O Savvidou1, E Giamarellos-Bourboulis2

1National and Kapodistrian University of Athens, 1st Department of Orthopedic Surgery, Athens, Greece; 2National and Kapodistrian University of Athens, 4th Department of Internal Medicine, Athens, Greece

Introduction: Previous findings of our group suggest that patients with Gram-negative hospital-acquired severe sepsis have better prognosis when sepsis is developing after recent multiple trauma through stimulation of favorable interleukin (IL)-10 responses [1]. Under a similar rationale, we investigated if preceding osteomyelitis may affect experimental osteomyelitis.

Methods: Sham or experimental osteomyelitis was induced in 32 male New Zealand white rabbits after drilling a hole at the upper metaphysis of the left tibia and implementing diluent or 5log10 of Staphylococcus aureus using foreign body. After three weeks, the foreign body was removed and experimental pyelonephritis or sham surgery was induced after ligation of the right pelvo-ureteral junction and instillation of 6log10 of Escherichia coli in the renal pelvis. Survival was recorded and circulating mononuclear cells were isolated and stimulated for the production of tumour necrosis factor-alpha (TNFa) and IL-10. At death or sacrifice, tissue outgrowth and myeloperoxidase (MPO) were measured.

Results: Four sham-operated rabbits (S), 16 rabbits subject to sham surgery and then pyelonephritis (SP) and 12 rabbits subject to osteomyelitis and then pyelonephritis (OP) were studied. Survival after 14 days of group SP was 56.3% and of group OP 100% (log-rank 6.59; p: 0.010). Lab findings are shown in Figure 1. Il-10 production was blunted. Negative correlation between E. coli outgrowth and tissue MPO was found at the right kidney of the OP group (rs: -0.767, p: 0.016) but not of the SP group (rs: -0.318, p: 0.340).

Conclusions: Preceding staphylococcal osteomyelitis provides survival benefit to subsequent experimental osteomyelitis through down-regulation of innate immune responses leading to efficient phagocytosis.


1. Mandragos E, et al. J Infect 2017; 74:163-171

Fig. 1 (abstract P032).

See text for description.

P033 Synthetic analogue of leu-enkephalin prevents neutrophil activation by bacterial compounds

V Likhvantsev1, O Grebenchikov2, A Prykhodko1, A Kuzovlev2

1Vladimirsky Moscow regional research and clinical center, Moscow, Russia; 2Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, Moscow, Russia

Introduction: Activation of neutrophils is a mandatory step and a sensitive marker of a systemic inflammatory response syndrome (SIRS) which is closely related to development of multiple organ failure. The search for drugs that can prevent SIRS and reduce mortality in critically ill patients remains significant. The aim of this study was to study the anti-inflammatory effect of the synthetic analogue of leu-enkephalin (Dalargin) on human neutrophils.

Methods: The study was conducted on isolated from the blood of healthy donors neutrophils. Their activation was assessed by fluorescent antibodies to markers of degranulation CD11b and CD66b (SD11b-FITC and CD66b-AlexaFluor647 (BD Biosciences, USA). As inductors of inflammation lipopolysaccharide (LPS) and the peptide formyl Met-Leu-Pro (fMLP) were used. 100mkM fMLP and dalargin in concentrations of 50 and 100 μ g / ml were added to neutrophils at a concentration of 4 ppm / ml and incubated for 30 min at 37°C; then antibodies were added and incubated for 30 min on ice; then fluorescence was assessed by flow cyto flow meter Beckman-Coulter FC 500. Non-parametric criteria were used; data were presented as a median and 25%—75% interquartile intervals. The statistical significance was estimated using Mann-Whitney test. The difference was considered statistically significant at P<0.05

Results: Synthetic analogue of leu-enkephalin in various concentrations has an anti-inflammatory effect on both intact and pre-activated with bacterial components neutrophils, reducing their activation and degranulation in a dose-dependent manner (Figs. 1, 2).

Conclusions: Synthetic analogue of leu-enkephalin prevents neutrophil activation by bacterial compounds. This has a potential of translation into clinical practice for sepsis treatment.

Fig. 1 (abstract P033).

Synthetic analogue of leu-enkephalin reduces the expression of the degranulation marker CD11b in intact and pre-activated human neutrophils (* - significant difference compared to controls, Mann-Whitney test)

Fig. 2 (abstract P033).

Synthetic analogue of leu-enkephalin reduces the expression of the CD66b degranulation marker in intact and pre-activated human neutrophils (* - significant difference compared to controls, Mann-Whitney test)

P034 Potential role of endothelin receptors in the therapy of experimental sepsis

A Rutai, R Fejes, SZ Tallósy, M Poles, L Juhász, A Mészáros, M Boros, J Kaszaki

University of Szeged, Institute of Surgical Research, Szeged, Hungary

Introduction: The endothelin system plays important roles in circulatory regulation through vasoconstrictor ET-A and ET-B2 receptors and vasodilator ET-B1 receptors (ETAr; ETBr, respectively). Tissue hypoxia during the progression of sepsis is associated with microcirculatory and mitochondrial disturbances. Our aim was to investigate the possible influence of ETAr antagonist, ETBr agonist or combined treatments on oxygen dynamics, microcirculatory and mitochondrial respiration parameters in experimental sepsis.

Methods: Male Sprague-Dawley rats (n=8/group) were subjected to faecal peritonitis (0.6 g/kg faeces ip) or sham-operation. Septic animals were treated with sterile saline solution, or received the ETAr antagonist ETR-p1/fl peptide (100 nmol/kg iv), ETBr agonist IRL-1620 (0.55 nmol/kg iv) or same doses as combination therapy, 22 hr after sepsis induction. Invasive hemodynamic monitoring and blood gas analyses were performed during a 90-min observational window. Intestinal microcirculation (perfusion rate, red blood cell velocity - RBCV) was investigated by intravital videomicroscopy. Complex I and II-linked (CI; CII,) mitochondrial respiration (oxidative phosphorylation - OxPhos) was evaluated by high resolution respirometry (O2k, Oroboros, Austria).

Results: The septic reaction was characterized by significant hypotension and decreased microperfusion, oxygen extraction and CI - CII-linked OxPhos values. The ETAr antagonist treatment significantly increased the oxygen extraction, RBCV and CII-linked OxPhos capacity. The ETBr agonist treatment prevented the sepsis-induced hypotension, decrease in oxygen extraction, and significantly increased the perfusion rate. The combined therapy amplified the beneficial mitochondrial and microcirculation effects of selective ETAr antagonist and ETBr agonist compounds.

Conclusions: The combination of ETAr antagonism and ETBr agonism may offer a novel tool for a simultaneous microcirculatory and mitochondrial resuscitation strategy in sepsis. Grant supports: NKFIH K116689

P035 Pharmacokinetics and pharmacodynamics of nivolumab in Japanese patients with immunosuppressive sepsis: determining safety and tolerability in a multicenter, open-label, phase 1/2 study

E Watanabe1, O Nishida2, Y Kakihana3, M Odani4, T Okamura5, T Harada5, S Oda6

1Chiba University Graduate School of Medicine, Department of General Medical Science, Chiba, Japan; 2Fujita Health University School of Medicine, Department of Anesthesiology & Critical Care Medicine, Aichi, Japan; 3Kagoshima University Graduate School of Medical and Dental Sciences, Department of Emergency and Intensive Care Medicine, Kagoshima, Japan; 4Ono Pharmaceutical Co., Ltd., Data Science, Osaka, Japan; 5Ono Pharmaceutical Co., Ltd., Clinical Development Planning, Osaka, Japan; 6Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, Chiba, Japan

Introduction: Sepsis often induces immunosuppression, which is associated with high mortality rates. Nivolumab is a human IgG-4 antibody directed against the programmed cell death 1 (PD-1) immune-checkpoint inhibitor, which disrupts PD-1-mediated signaling and restores antitumor immunity. Nivolumab is an approved anti-cancer drug that may have the potential to improve sepsis-induced immunosuppression.

Methods: This multicenter, open-label study investigated the safety, pharmacokinetics and pharmacodynamics of a single intravenous infusion of 480 or 960 mg nivolumab in Japanese patients with immunosuppressive sepsis (lymphocytes ≤ 1100 /μL). The dosing of nivolumab was set using the predicted steady state concentration of nivolumab at 3 mg/kg every 2 weeks (Q2W), which was the approved dosage for cancer patients at the time of planning.

Results: Five and eight patients were assigned to the 480 and 960 mg groups, respectively. The mean (standard deviation) peak serum drug concentration in the 480 mg group was comparable to the predicted median concentration (90% PI [prediction interval]) at the end of infusion with 3 mg/kg Q2W (132 [39.5] μg/mL vs. 117 [56.4-239] μg/mL) (Table 1). In addition, the median (range) concentration on day 28 in the 960 mg group was within the range of the predicted minimum concentration (median [90% PI]) after dosing with 3 mg/kg Q2W (33.1 [6.47-44.8] μg/mL vs. 57.7 [19.0-163] μg/mL). Lymphocyte counts and monocytic human leukocyte antigen DR-1 appeared to increase over time in both groups (Figures 1 and 2). Adverse events (AEs) were observed in four patients in each group. Drug related-AEs were observed in only one patient in the 480 mg group (Table 2). No deaths related to nivolumab occurred.

Conclusions: A single dose of 960 mg nivolumab appeared to be well tolerated and sufficient to maintain nivolumab blood concentration in patients with sepsis. Results suggest both 480 and 960 mg nivolumab therapy could improve relevant immune indices.

Table 1 (abstract P035). See text for description
Table 2 (abstract P035). See text for description
Fig. 1 (abstract P035).

Change in lymphocyte count over time

Fig. 2 (abstract P035).

Change in monocyte HLA-DR antigen expression over time

P036 Biological effects of low dose aspirin in critically ill patients with the systemic inflammatory syndrome (SIRS): a pilot feasibility multi-center randomised placebo-controlled trial (PROTECTIN trial)

N Luethi1, L Cioccari1, G Eastwood1, L Peck1, H Young1, SL Cutuli1, P Lloyd-Donald1, M Bailey2, C French3, N Orford4, J Dwivedi5, T Duong6, E Ryan6, G Reid6, R Bellomo1

1Austin Hospital, Intensive Care Unit, Heidelberg, Australia; 2Australian and New Zealand Research Centre, Melbourne, Australia; 3Western Health, Intensive Care Unit, Footscray, Australia; 4University Hospital Geelong, Intensive Care Unit, Geelong, Australia; 5Bankstown Hospital, Intensive Care Unit, Bankstown, Australia; 6University of Melbourne, School of Chemistry, Department of Biochemistry and Molecular Biology, Melbourne, Australia

Introduction: The systemic inflammatory response syndrome (SIRS) accompanies tissue trauma and infection and, when severe or dysregulated, contributes to multiple organ failure and critical illness. Observational studies in man and animal have shown that low-dose acetyl-salicylic acid promotes resolution of inflammation and might attenuate excessive inflammation by increasing the synthesis of specialised pro-resolving lipid mediators (SPMs).

Methods: We randomly assigned patients with SIRS who were expected to stay in ICU for more than 48 hours to receive enteral aspirin (200 mg per day) or placebo for 7 days or until death or discharge from the ICU, whichever came first. The primary outcome was IL-6 serum concentration at 48h after randomisation. The secondary outcomes included safety and feasibility outcomes. In one center, additional blood samples were taken during the first three days for exploratory analysis of SPMs using reversed-phase high-performance liquid chromatography - tandem mass spectrometry (RP-HPLC-MS/MS).

Results: From March 2015 through December 2017 a total of 48 patients across four general ICUs in Australia underwent randomization (Table 1). Compared to placebo patients, IL-6 serum concentration after 48h in aspirin-treated patients was not significantly lower (40 [16-166] pg/ml vs 44 [7.4-85] pg/ml; p=0.66). There were no significant differences for control vs. aspirin-treated patients in the change of pro-resolving/anti-inflammatory lipids between the time points (Figure 1, 2). There were no between-group differences with respect to ICU or hospital mortality, number of bleeding episodes or requirements for red cell transfusions (Table 2).

Conclusions: In patients admitted to the ICU with SIRS, low-dose aspirin did not result in a decreased concentration of inflammatory biomarkers compared with placebo.

Table 1 (abstract P036). See text for description
Table 2 (abstract P036). See text for description
Fig. 1 (abstract P036).

Boxplots of biomarker levels at baseline and 48h after randomisation

Fig. 2 (abstract P036).

Boxplots of the fold change of the total pro-resolving/anti-inflammatory lipid mediators (28 lipids) for control and aspirin-treated patients

P037 Ketone body 3-hydroxybutyrate protects against muscle weakness and stimulates muscle regeneration in septic mice

C Goossens, S Derde, S Vander Perre, R Weckx, G Van den Berghe, L Langouche

KU Leuven, Laboratory of Intensive Care Medicine, Leuven, Belgium

Introduction: Debilitating muscle weakness and impaired muscle regeneration is prevalent in ICU patients. Remarkably, premorbid obesity has been shown to protect against this weakness in both ICU patients and septic mice, which coincided with markers of elevated ketogenesis [1]. We here assessed whether ketone body supplementation could directly protect the muscle during sepsis.

Methods: In a resuscitated, antibiotic-treated mouse model of prolonged (5 days) abdominal sepsis (cecal ligation and puncture), lean ill mice received standard parenteral nutrition (5.8kcal/day) supplemented with D,L-3-hydroxybutyrate (PN+3HB; 150mg/day; n=17) or isocaloric glucose (PN+gluc; 187.5mg/day; n=17). Pair-fed healthy mice served as controls (n=15). Markers of muscle weakness and regeneration were assessed.

Results: Compared to controls, absolute maximal EDL muscle force was reduced with 44% in PN+gluc septic mice (p<0.0001) and with 29% in PN+3HB septic mice (p=0.0001 vs. controls and p=0.002 vs. PN+gluc). Specific maximal muscle force, which is corrected for muscle mass, was reduced with 29% in PN+gluc septic mice compared to controls (p=0.0001), whereas PN+3HB septic mice maintained their specific maximal muscle force up to control levels (p=0.1 vs. controls and p=0.01 vs. PN+gluc), implying preservation of muscle quality, but not quantity with 3HB. Furthermore, PN+3HB increased gene expression of regeneration marker Myod1 (p≤0.03 vs. PN+gluc and controls), and of myogenic regulatory factors Myog and Myf5 (p<0.05 vs. PN+gluc). Stimulation of regeneration markers with PN+3HB coincided with a decreased expression of Hdac4, Hdac5 and upregulation of downstream gene Mef2c (known to stimulate muscle regeneration; p<0.05 vs. PN+gluc).

Conclusions: PN+3HB in lean septic mice protected against muscle weakness and elevated muscle regeneration markers. These data identify nutritional 3HB supplementation as a potential preventive therapy for muscle weakness, requiring further investigation.


[1] Goossens et al. JCSM 8:89-101, 2017

P038 Vitamin C and thiamine in septic shock: a retrospective before-after analysis.

A Casazza1, U Suppo2, M Delorenzo2, L Marinelli2, S Testa2, E Bellazzi1, R Boschi1, D Ciprandi1, C Gigliuto1, R Preda1, R Vanzino1, M Vetere1, L Carnevale1

1ASST Pavia, Anaesthesia and Intensive Care Vigevano, Vigevano, Italy; 2Università degli Studi di Pavia, Scuola Specialità MEU, Pavia, Italy

Introduction: Sepsis is associated with excessive ROS production, NF-kB, iNOS and inflammatory mediators overexpression. Vitamin C is a cellular antioxidant, it increases eNOS and decreases NF-kB; it has several immune-enhancing effects and is crucial for endogenous vasopressors synthesis. Vitamin C reserves in sepsis are often as poor as in scurvy [1]. In recent studies, intravenous high Vitamin C dose seems to reduce organ failure and improve outcome in septic shock.

Methods: We treated all septic shock patients admitted to our ICU in 7 months (from 3/2018 to 9/2018) with intravenous Vitamin C 1.5g/6h and Thiamine 200 mg/12h (for its synergistic effects) [2] as adjunctive therapy for 4 consecutive days and we compared data to septic shock patients admitted in the previous 7 months period. We enrolled 24 patients: 13 received Vitamins supplementation, 11 standard of care. We analysed 28-days mortality, SOFA at 48 and 96 hours, PCT variation from baseline in first 5 days, vasoactive therapy length and DAF 28 (Days Alive and Free from vasopressors, mechanical ventilation and RRT in 28 days follow up). Patients with end stage kidney disease were ruled out. We analysed data with Mann-Whitney and Wilcoxon tests.

Results: Vit C group showed lower 28-days mortality (23% vs 54.5%: NS); SOFA improvement at 48 (-2.4±1.5 vs -0.9±1.9: p=0.01) and 96 hours (-4.2±2 vs -1.9±2: p<0.01) was higher in Vit C group; Vit C patients had faster PCT reduction without statistical significance. Mean vasoactive therapy length was quite similar. DAF was 16.5 (±9.1) days in Vit C group and 9.3 (±8.9) in controls (p=0.03). 3 control patients needed RRT, none in Vit C group.

Conclusions: Despite small study size, we found that Vit C has positive effects on survival and improves SOFA score (Fig.1) and DAF (Fig.2) in septic shock. No Vit C patient developed oxalate nephropathy nor worsened renal function.


1. Marik PE et al. Crit Care 22:23, 2018

2. Marik PE et al. Chest 151:1229-1238, 2017

Fig. 1 (abstract P038).

SOFA variation

Fig. 2 (abstract P038).

DAF variation

P039 A double- blind, randomized, placebo-controlled clinical study of the efficacy of intravenous clarithromycin as adjunctive treatment in patients with sepsis and respiratory and multiple organ dysfunction syndrome: INCLASS study

E Karakike1, M Roumpoutsou1, N Karampela2, E Massa3, E Antypa4, K Psaroulis5, P Chaloulis6, E Pappa7, V Kolonia8, EJ Giamarellos-Bourboulis1, I Tsangaris9

1National and Kapodistrian University of Athens, 4th Department of Internal Medicine, Athens, Greece; 2Korgialeneio Benakeio General Hospital, Intensive Care Unit, Athens, Greece; 3Hippokration General Hospital, Intensive Care Unit, Thessaloniki, Greece; 4G. Gennimatas General Hospital, Intensive Care Unit, Thessaloniki, Greece; 5Aghios Dimitrios General Hospital, Intensive Care Unit, Thessaloniki, Greece; 6Theageneion General Hospital, Intensive Care Unit, Thessaloniki, Greece; 7Laiko General Hospital, Intensive Care Unit, Athens, Greece; 8Sismanogleion General Hospital, 2nd Department of Internal Medicine, Athens, Greece; 9National and Kapodistrian University of Athens, 2nd Department of Critical Care Medicine, Athens, Greece

Introduction: In the light of new insight on pathogenesis of sepsis and after inconclusive randomized clinical trials (RCTs), the benefit of macrolides as adjunctive, low-cost and promising molecules in sepsis, remains to be assessed. The INCLASS study ( NCT: 03345992) is an ongoing RCT aiming to evaluate clarithromycin as immune modulator in high-risk septic patients.

Methods: Adult patients with sepsis according to Sepsis-3 definitions with respiratory failure (defined as PaO2/FiO2< 200) and total SOFA score equal to or more than 7 can be enrolled provided that they present with one of the following infections: hospital-acquired pneumonia; healthcare-associated pneumonia (HCAP); ventilator-associated pneumonia (VAP); intra-abdominal infections (IAIs) and primary gram-negative bacteremia . Patients are blindly randomized to receive either 1gr of intravenous clarithromycin or placebo once daily for four consecutive days. The primary endpoint is survival at 28 days. The study is powered for 110 patients.

Results: Sixty-nine patients have been enrolled so far. The most common infections are VAP 33%, IAI 22% and HCAP 20%. Most common isolated pathogens are Acinetobacter baumannii 37%; Klebsiella pneumoniae 27% and Escherichia coli 20%. Mean ± SD Charlson’s comorbidity index is 5.0 ± 2.7 and APACHE score 21.9 ± 6.6. SOFA score on enrollment is 10.4 ± 2.5.

Conclusions: The INCLASS study is an on-going pragmatic trial enrolling very severe patients with high SOFA score.

P040 IVIG administration in ECMO patients with toxin-mediated shock

M Peetermans, R Wan, C Meadows, N Ioannou, G Glover, L Camporota, D Wyncoll, N Barrett, A Retter

St. Thomas´ Hospital, London, Critical Care, London, United Kingdom

Introduction: Toxin-producing gram-positive organisms cause some of the most severe forms of septic shock [1,2]. Adjunctive therapies such as intravenous immunoglobulins (IVIG) have been proposed for these patients [3,4]. However, at patient presentation, the presence of a toxin-producing organism is most often unknown.

Methods: We reviewed the use of IVIG in our patients requiring extracorporeal membrane oxygenation (ECMO) in a 2-year period between February 2016 and March 2018.

Results: In 44% (15/34) of the patients that received IVIG for presumed toxin-mediated shock, group A Streptococcus or Panton-Valentine leukocidin producing S. aureus was isolated, but the clinical characteristics of these 15 patients were not significantly different from the ones with other final diagnoses, except for a predisposing influenza infection and the presence of an often very high procalcitonin level. These 34 patients were extremely unwell at presentation with a SOFA score of 15 ± 3, high lactate levels (8.6 ± 5.8 mmol/L) and need for vasopressors (equivalent norepinephrine dose of 0.93 ± 0.62 μ g/kg/min). They had very high inflammatory parameters with a procalcitonin ≥ 100 ng/mL in more than half of patients (18/34). IVIG use in these patients was generally safe, with only 1 possible transfusion reaction. The mortality of 35% (12/34) was lower than predicted based on the SOFA scores.

Conclusions: IVIG administration can be considered in a selected group of patients presenting with acute and very severe septic shock, as part of a multimodal approach [5].


1. Burnham JP and Kollef MH. Intensive Care Med 2015;41:1707-10.

2. Gillet Y, et al. Lancet, 2002;359:753-9.

3. Alejandria MM, et al. Cochrane Database Syst Rev 2013:CD001090.

4. Shankar-Hari M, et al. Crit Care 2012;16:206.

5. Nandhabalan P, et al. Crit Care 2018;22:215.

P041 Immunological effect of the adsorbing membrane oXiris on septic patients with AKI

F Turani1, S Busatti1, V Cotticelli1, R Barchetta1, F Leonardis2, F Candidi1, M Dauri2

1Aurelia And European Hospital, Anesthesia And Intensive Care, Rome, Italy; 2University Of Tor Vergata, Anesthesia And Intensive Care, Rome, Italy

Introduction: Extra corporeal treatments are used in septic patients to decrease the inflammatory mediators, but definitive conclusions are lacking . More over in many studies the effect of AKI isn’t evaluated and this may be an important bias. . The aim of this study is to evaluate in septic patients with AKI: 1- the effect of the adsorbing membrane Oxiris on the immunological response 2- the different response in survivors and non survivors

Methods: From our local data base we analyzed retrospectively 50 septic shock patients with AKI (KDIGO classification) submitted to CRRT with the adsorbing membrane oXiris (Baxter, USA ) . At basal time ( T0 ) and at the end of the treatment ( T1 ) we evaluated the following variables: IL 6 IL 10 Procalcitonin Endotoxin (EAA). All data are expressed as mean ±SD or median and IQR. Student T test or Mann- Whitney was used to compare values changes. P < 0.05 was considered statistically significant.

Results: Thirty patients with sepsis /septic shock and AKI were enrolled in this study. 10 patients had AKI 3, 15 patients AKI 2, 5 patients AKI 1. The duration of treatment was 38± 10 hours. 20 patients had citrate as anticoagulation and 10 heparine continous ev. At Table 1 are shown the main results of this study in all the patients. Survivors vs non survivors had a significant decrease of IL 6, Procalcitonin and EAA.

Conclusions: Data of this study confirm on clinical ground previous study “in vitro” [1] that the adsorbing membrane oXiris has important immunological effect during septic shock with AKI. This must be confirmed in a RCT.


1 Malard et al. Intensive Care Medicine Experimental (2018) 6:12

Table 1 (abstract P041). Septic patients with AKI

P042 Hemoadsorption (CytoSorb) therapy in septic shock and MODS patients - a prospective multi-centre investigator initiated study

R Paul1, P Sathe2, R Senthil3, S Prasad4, M Aleem1, S Prashant2

1Apollo Health City, Department of Internal Medicine and Critical Care, Telangana State, India, India; 2Ruby Hall Clinic, Department of Critical Care Medicine, Pune, India; 3Apollo Hospital, Greams Lane, Department of Critical Care Medicine, Chennai, India; 4Narayana Institute of Cardiac Sciences, Department of Anesthesiology and Critical Care, Bangalore, India

Introduction: Sepsis is common and often fatal, representing a major public health problem. Hemoadsorption (CytoSorb) therapy aims to reduce cytokines and stabilise the overall immune response in septic shock patients.

Methods: A prospective, multi-centre, investigator initiated study to evaluate Hemoadsorption (CytoSorb) Therapy in septic shock patients admitted to a tertiary ICU’s in India during 2016 to 2018. All centres followed a common protocol and received ethics committee approval.

Results: A total of 45 patients were administered CytoSorb in addition to standard of care. A total of 26 patients (62%) survived out of 45 patients. Among survival group, 17 patients (71%) were administered CytoSorb within 48 hours of ICU admission resulting in significant reduction in Sepsis Scores, APACHE II (24.42 vs 19.33) and SOFA (12.46 vs 8.71) post CytoSorb therapy. Also there was reduction in inflammatory markers like Cytokines IL6 in most of the patients. All patients in survivor group showed a significant improvement in MAP (69.2 vs 76.8) and reduction in vasopressors (Epinephrine 0.3 to 0.03 mcg/kg/min, Nor-Epinephrine 0.34 to 0.04 mcg/kg/min) after CytoSorb therapy. No device related adverse effect was observed in any of the patients. Among the non-survivor group, (16 patients, 38%) we observed that CytoSorb was administered after 48 hours of ICU admission. Although a few patients showed improvement in SOFA score, majority did not show a significant improvement with MAP (68.05 vs 60.4 mm of Hg) and required increased demand in Vasopressors.

Conclusions: In this multi-centered prospective IIS study, we could observe clinical benefits of Hemoadsorption (CytoSorb) therapy in Septic shock patients if the therapy was initiated early. Larger randomised study are required to establish the above clinical benefits in larger patient population.

P043 A single centre experience with hemoadsorption (CytoSorb) in varied causes of sepsis and MODS

Y Mehta1, C Mehta1, A Kumar1, J George2, A Gupta3, S Nanda1, G Kochar1, A Raizada3

1Medanta The Medicity, Institute of Critical Care and Anesthesiology, New Delhi, India; 2Medanta The Medicity, Institute of Critical Care and Anesthesiology, Clinical Co-Ordinator, New Delhi, India; 3Medanta The Medicity, Department of Biochemistry, New Delhi, India

Introduction: Sepsis and the multiorgan failure is a leading cause of mortality in the intensive care unit. Promising new therapies continue to be investigated for the management of septic shock. We tried to evaluate a novel Hemoadsorption therapy (CytoSorb) through a retrospective evaluation of patient’s data in our centre. We used it as an adjuvant therapy in our patients with Sepsis due to varied causes.

Methods: We retrospectively analysed data of 100 Sepsis & Septic shock patients admitted between 2016 to 2018, who had received CytoSorb as adjuvant therapy along with standard of care. Institutional ethics committee approval was taken before initiating the study.

Results: A total of 100 patients (77 Male and 23 Females) with a mean age of 52.53 years were administered CytoSorb in addition to standard of care. A total of 40 patients survived out of 100 patients. Among 40 patients who survived 28 patients (70%) were administered CytoSorb within 48 hours of ICU admission. There was a significant reduction in scores like APACHE (26.4 vs 18.02) and SOFA (15.05 vs 10.32) post CytoSorb therapy. All the survivors showed a significant improvement in MAP (62.8 vs 68.2) and reduction in vasopressors need (Epinephrine 0.1 to 0.05 mcg/kg/min, Nor-Epinephrine 0.5 to 0.05 mcg/kg/min) after CytoSorb initiation. Out of 60 Patients who didn’t survive, 43 patients (72%) received CytoSorb therapy after 48 hours of ICU admission. In majority of these patients, there was no improvement of SOFA Score, MAP and increased need of Vasopressor demand as compared to the survival group.

Conclusions: Retrospective analysis showed significant reduction of vasopressors, Sepsis Score and improvement in MAP in survived group versus non-survived group. Looking into the positive outcome of this case series, randomized controlled studies are required to define the potential benefits of this new treatment option.

P044 Extracorporeal cytokine removal in septic shock patients: two-years experience

A Gulleroglu, E Gedik, H Sahinturk, A Ozdemirkan, P Zeyneloglu, A Pirat

Baskent University Faculty of Medicine, Anesthesiology and Critical Care, Ankara, Turkey

Introduction: Septic shock is a life-threatening multiple organ dysfunction that has high morbidity and mortality in critically ill patients, due to a dysregulated host response to infection. The aim of this study was to evaluate the efficacy of therapeutic cytokine removal (CytoSorb®) in the management of patients with septic shock.

Methods: We retrospectively analyzed patients admitted to ICU with septic shock between June 2015 and November 2017. Patients included in the study were diagnosed according to The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3), received maximal supportive care including continuous veno-venous hemodiafiltration (CVVHDF) for acute kidney injury and Cytosorb® haemoadsorption column was added to return limb of the CVVHDF circuit. Demographic data, procalcitonin and leukocyte levels before and after therapeutic cytokine removal and duration of Cytosorb® haemoadsorption column application and APACHE II scores were recorded.

Results: The mean age of 48 patients included in the study was 54±16.4 years (74% male) and the mean body mass index was 23.1±5.9. The mean APACHE II score was 21.5 with an expected and actual mortality rates of 40% and 25%, respectively. 18% of the patients were admitted with sepsis and 82% of them with septic shock. 37.5% (n=18) of the cases were solid organ transplant recipients. CVVHDF was applied in all patients during therapeutic cytokine removal. Treatment was combined with ECMO in 8 patients. While the mean duration of CVVHDF was 102.1 hours, the duration of Cytosorb® haemoadsorption column application was 24.1±13.4 hours. Procalcitonin (16.5 ± 25ng/ml vs 25±34ng/ml) and leucocyte levels (14048±10490/mm3 vs 9278±8693 mm3) after therapeutic cytokine removal were found significantly lower than the pretreatment values (respectively p=0.0013, p=0.006).

Conclusions: Therapeutic cytokine removal applied with CVVHDF in septic shock patients have positive contributions to biochemical parameters and provide survival advantage.

P045 Potential benefits of a hemoadsorption column in septic ICU patients

M Popescu, A Marcu, C David, D Tomescu

Fundeni Clinical Institute, Anaesthesia and Critical Care, Bucharest, Romania

Introduction: Recent studies have focused on demonstrating the potential benefits of immunomodulation in the management of septic patients. The aim of our study was to assess the effects of a hemoadsorption column (CytoSorb®) in critical ill septic patients.

Methods: After ethical approval was obtained, we prospectively included 39 patients admitted to the general ICU of Fundeni Clinical Institute. Three consecutive sessions of renal replacement therapy (continuous venovenous hemodiafiltration) in combination with CytoSorb® were applied after ICU admission. Clinical (heart rate, arterial pressure, temperature, Glasgow coma scale) and paraclinical data (PaO2, serum bilirubin and creatinine, platelet count, white blood cell count, pH, C-reactive protein and procalcitonine), vasopressor support and need for mechanical ventilation were recorded before and after the three sessions.

Results: The mean age in the study group was 57±15 years. Median number of organ dysfunction at the time of ICU admission was 4 [1-5] and the mean SOFA score was 10.0±3.7. The use of CytoSorb® was associated with a non-significant increase in PaO2/FiO2 ratio from 216±80 to 248±94 (p=0,278) and creatinine levels from 1.8±1.3 to 1.6±1.2 mg/dL (p=0.685). Although we observed a non-significant increase in C-reactive protein levels from 136±66 mg/L to 144±72 mg/L (p=0.577), we noted a significant decrease in procalcitonine levels from a median of 20.0 [2.2, 100.0] ng/dL to a median of 9.1 [0.7, 55.2] ng/dL (p=0.041). A significant decrease in platelet count was also noted from 135384±99263 /mm3 to 78470±61624 /mm3 (p=0.002). Mean SOFA score decreased non-significantly from 10.0±3.7 to 9.1±4.1 (p=0.533).

Conclusions: The use of CytoSorb was associated with a slight non-significant improvement in organ function and a decrease of procalcitonine levels. Thrombocytopenia remains one of the most important complications of renal replacement therapy.

P046 Removal of circulating NETs-related components with an immobilized polymyxin B filter

N Takeyama, T Gocho, Y Maruchi, N Takenaka, H Mori, M Islam, MA Huq

Aichi Medical University, Department of Emergency and Critical Care Medicine, Aichi, Japan

Introduction: Circulating cell-free neutrophil extracellular traps (NETs) would induce a microcirculatory disturbance of sepsis. The removal of NETs remnants from the circulation could reduce NETs-dependent tissue injury. To address this issue, we evaluated the effect of hemoperfusion with a polymyxin B cartridge (PMX-DHP; Toray, Japan), which was originally developed for the treatment in patients with Gram-negative bacterial infection, on circulating cell-free NETs in patients with septic shock and in phorbol myristate acetate (PMA)-stimulated neutrophils obtained from healthy volunteer.

Methods: Ex vivo closed loop hemoperfusion was performed through a circuit formed by connecting the small PMX module to a tube and a peristalsis pump. Whole blood from healthy volunteers incubated with or without PMA or from septic shock patients were applied to circuit and perfused. Blood was collected at 0, 1 and 2 hr after perfusion. Circulating cell-free NETs were assessed by myeloperoxidase (MPO)-, neutrophil elastase (NE)-, and cell free (cf)-DNA.

Results: Plasma MPO-DNA, NE-DNA and cf-DNA levels were significantly increased at 2 hr after PMA stimulation when compared with plasma levels without PMA. When either blood from septic shock patients or PMA-stimulated neutrophils obtained from volunteers were applied to circuit, circulating MPO-DNA, NE-DNA and cf-DNA were significantly reduced in perfusion with PMX filter than in perfusion without PMX filter at times 1 and 2 hr.

Conclusions: In the ex vivo experiments, MPO-DNA, NE-DNA and cf-DNA were found to decrease after ex vivo perfusion through PMX filters. Selective removal of circulating components of NETs may improve the remote organ damage in patients with septic shock.

P047 A retrospective study of septic shock patients who were treated with direct hemoperfusion with polymyxin B-immobilized fibers based on the levels of endotoxin activity assay

S Sekine, H Imaizumi, I Saiki, A Okita, H Uchino

Tokyo Medical University, anesthesiology/ICU, Tokyo, Japan

Introduction: The purpose of this study was to evaluate the outcomes for septic shock patients with direct hemoperfusion with polymyxin B-immobilized fibers (PMX-DHP) and endotoxin activity assay (EAA).

Methods: According to the levels of EAA, 41patients were classified for three groups (low group (GL); EAA <0.4, intermediate group (GM); EAA >0.4 or EAA <0.6, high group (GH); EAA >0.6). In order to evaluate the severity of illness, acute physiology and chronic health evaluationII (APACHE II) score, the sequential organ failure assessment (SOFA) score, catecholamine index (CAI) were recorded. And the presence of PMX-DHP treatments were also recorded. Blood samples were obtained to measure EAA levels, inflammatory markers (procalcitonin (PCT), C-reactive protein (CRP), and white blood cell count (WBC)), serum lactate level as an indicator of tissue hypoxia, and for blood culture. APACHE II score, SOFA score, CAI, inflammatory markers, serum lactate levels (Lac) and blood culture results were examined for diagnosis of septic shock and prognosis of 30-days mortality. Each values were also compared to EAA levels.

Results: 41 septic shock patients were included (GL/ GM/ GH: 12/ 13/ 16). In GH, APACHE II and SOFA score was significantly higher than that in GL (p< 0.05). EAA levels were significantly increased in gram-negative bacteremia patients compared to the patients with gram-positive bacteremia or fungemia. There was no relationship between EAA levels and other inflammation markers, CAI, and Lac. In GM, 30-days mortality in patient with PMX-DHP treatments was lower than that of without PMX-DHP treatments (0.14 (1/7) vs 0.5 (3/6), p=0.27). In GH, 30-days mortality in patient with PMX-DHP treatments was same as that of without PMX-DHP treatments (0.5 (3/6) vs 0.5 (5/10), p=1.0).

Conclusions: These results of this study suggest PMX-DHP treatment may improve the outcome of septic shock patients with intermediate EAA levels.

P048 Exploring alternative trial designs for pragmatic clinical studies: a Bayesian decision-theoretic model applied on a real ongoing one-stage trial

E Karakike1, A Alban2, EJ Giamarellos-Bourboulis1, SE Chick2

1National and Kapodistrian University of Athens, 4th Department of Internal Medicine, Athens, Greece; 2INSEAD, Technology and Operations Management, Fontainebleau, France

Introduction: Numerous inconclusive randomized clinical trials (RCTs) in sepsis in the past years suggest a need to re-think trial design to improve resource allocation and facilitate policy adoption decisions. The INCLASS study ( NCT: 03345992) is an ongoing RCT evaluating clarithromycin as an immune modulator in high-risk septic patients with clinical and cost-effectiveness outcomes. We aim to compare the original one-shot trial with an alternative sequential design that balances trial costs and value of information.

Methods: Adult patients with sepsis, respiratory failure and total SOFA score of at least 7, are randomized to receive intravenous clarithromycin or placebo adjunctive to standard-of-care therapy. For the cost-effectiveness study, efficacy is measured in Quality-Adjusted Life Years (QALYs) by EQ-5D-3L questionnaire at 90 days. The endpoint is the Incremental Net Monetary Benefit (INMB) of clarithromycin compared to placebo, defined as WTP x (Increment in QALY) – (increment in costs), where WTP is willingness to pay per QALY gained. Fixed and variable costs of trial execution (including administrative, insurance, supplies, tests) are calculated; hospitalization cost is extracted from patient records; medical care beyond day 28 is recorded; cost of adoption in the general population is estimated. Previous data from RCTs using clarithromycin are used to form a prior belief about the INMB. Known incidence of sepsis with respiratory failure allows estimation of the population to benefit from trial decision. A Bayesian model is used to determine the sequential design that maximizes trial value.

Results: We will compare the performance of the sequential trial design with the one-shot design of INCLASS trial in terms of sample size, cost, social-welfare, and probability of correctly identifying the best treatment.

Conclusions: In this protocol we validate a Bayesian model for sequential clinical trials and assess the benefits for the patient population and health care system.

P049 The effect on the outcome of critically ill patients with catecholamine resistant septic shock and acute renal failure through implementation of adsorption therapy

G Schittek1, H Simonis1, O Huhn2, J Soukup2, J Soukup2

1Medical University of Graz, Department of Anesthesiology and Intensive Care Medicine/ Division of General Anesthesiology and Intensive Care Medicine, Graz, Austria; 2Carl-Thiem-Klinikum, Klinik für Anästhesiologie, Intensivtherapie und Palliativmedizin, Cottbus, Germany

Introduction: CytoSorb-Adsorption has been described as an effective way for hemodynamic stabilisation in septic shock [1]. Aim of this study was to examine whether the adsorption-therapy could influence patient-outcome with catecholamine resistant septic shock (CRSS) and acute renal failure(ARV). Furhtermore we tried to identify clinical constellations that would predict an effective use of adsorbers.

Methods: We evaluated 44 adult patients with CRSS, ARV and CytoSorb-Therapy during the period 11/14-3/18. Data were collected according to the Cytosorb-Registry and with permission of the local chamber of physicians [AS 88(bB)/2015]. Furthermore we collected data from a matched patient group from the year 2012 (N=14) with septic shock and increasing noradrenaline dependency and ARV. The efficacy was assesed by means of laboratory tests, catecholamine dependency and outcome. Calculations were done with non-parametric-tests (depicted as median values [Q25, Q75]).

Results: Initial IL-6 was 5000 ng/l [908,5000] and could be reduced to 302 ng/l [99,891]. PCT was non-significantly reduced from 27μg/l [11, 66] to 20μg/l [6,40]. Vasopressor-dependency could be reduced in 19 patients from 60 μg/´ [50,66] to 24 μg/´ [7,34]. Initial IL-6 in patients with catecholamine-reduction through adsorption was non-significantly different to those with no reduction (2376 ng/l [838, 5000] vs. 5000 ng/l [3488, 5000]). Mortality did not differ significantly between the groups (71% vs 79%). Length of intensive care unit stay (LOS) did differ significantly (13 days [4,24] vs 22 days [19,29]).

Conclusions: IL-6 can be reduced with adsorption. Patients with catecholamine-reduction did not differ in regard to their initial IL-6. LOS was shorter for patients treated with adsorption. According to our experience adsorption can be taken into consideration when CRSS is beginning.


1. Hinz B, et al. Int J Artif Organs. 2015;38:461-4.

P050 Determination of the started period of expanded application to the contact precautions among the scheduled surgical patients with prolonged stay in the mixed intensive care unit

K Ishii1, H Hidaka1, Y Koyama1, K Abo1, M Arai1, M Kosaka1, S Takenaka1, C Yokoo1, Y Yajima1, Y Kimura1, H Ohmura1, K Morinobu1, I Shimada1, Y Kimura2, M Sato3, H Morimatsu4

1Fukuyama City Hospital, Department of Anesthesiology and Oncological Pain Medicine, Fukuyama, Japan; 2Fukuyama City Hospital, Central Surgery Department Intensive Care Unit, Fukuyama, Japan; 3Fukuyama City Hospital, Medical department (medical information engineer), Fukuyama, Japan; 4Okayama University Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Department of Anesthesiology and Resuscitology, Okayama University Graduate School of Medicine, Okayama, Japan

Introduction: In our Intensive Care Unit (ICU), we have already started expanded application to the contact precautions. Applied patients are; 1) Emergency admission, 2) Patients who had already had bacteria* that are required to contact precautions, 3) Scheduled surgical patients with prolonged ICU stay, although we have not yet decided the started period of expanded application exactly. *Detected Bacteria(DB);MRSA, CD, MDRP, ESBL, Pseudomonas A, PISP, PRSP, VRSA. The aim of this study was to determine the adequate starting period of expanded application to the contact precautions in the scheduled surgical patients in the mixed ICU.

Methods: We performed retrospective observational study on 4221 patients who were admitted to our ICU after planed surgery from May 2013 to Dec. 2017. We detected the patients who acquired BD newly and investigated the relation to the length of ICU stay. The relationship between detection rate and categorized date was also analyzed using logistic regression adjusted for age, gender, APACHE2, and SOFA score. Using Youden´s index and ROC curve, we also calculated cutoff point of the duration of ICU stay related to detection rate. Finally, we made the logistic regression model of each cutoff day(day1 to 7) and compared Odds Ratio(OR) and AUC of each models using stata.

Results: Category day 2 or more, especially day 4 or more had significantly higher detection rate of DB compared to day 1 (Table 1). Similar results were observed in OR according to logistic regression. Day 2, 3 or 4 (Youden´s index) and day 3 (ROC curve) were recommended as the cutoff point. According to each cutoff day models of logistic regression, the day 4 model had the highest OR (51.3) and AUC (0.935).

Conclusions: Day 3 or day 4 in mixed ICU stay may be the most adequate period to start expanded application to the contact precautions if the scheduled surgical patient´s stay is prolonged.

Table 1 (abstract P050). Relationship between duration of ICU stay and detection of bacteria and toxin that required contact precautions

P051 Prevention of urinary tract infection in neurocritical patients - a reality

P Travassos, L Silva, G Costa, S Chaves, W Costa, R Vale, V Veiga, S Rojas

Hospital Bp - A Beneficência Portuguesa De São Paulo, Neurocritical Care Unit, Sao Paulo, Brazil

Introduction: The objective of this study was to evaluate the incidence density of urinary tract infection associated with bladder catheter in neurological intensive care unit and identification of actions that were related to low prevalence.

Methods: A retrospective analysis of the hospitalized patients from December 2014 to January 2017 was carried out, considering the patients who used the bladder catheter and the cases of urinary tract infection, correlating with improvement actions implemented in the period.

Results: In the analyzed period, 3837 patients were hospitalized in the unit, with mean age of 63.77 years, SAPS 3 of 44.58 points, SMR 0.59 and residence time of 4.79 days, with 50.92% using a catheter bladder of delay. Of these, 27 had a urinary tract infection, which represented 1.37% of the patients. During the analyzed period, urological physiotherapy was monitored, daily check of the urinary tract infection prevention bundle, analysis of all cases of infection with search of barriers breaking through Ishikawa methodology, feedback to the multiprofessional team of indicators related to the presence of invasive device, monthly monitoring of the mean time of bladder catheter with established goals.

Conclusions: It is possible to guarantee low prevalence of urinary tract infection, in a complex profile of patients, through a multiprofessional approach, accompanied by structured management of data analysis and monitoring

P052 Risk factors of surgical site infections after thoracic and lumbar surgery: a 6-year single centre prospective cohort study

V Spatenkova1, O Bradac2, Z Jindrisek1, J Hradil3, P Suchomel3, D Fackova4, M Halacova5

1Regional Hospital, Neurocenter, NICU, Liberec, Czech Republic; 2Military University Hospital and First Medical School, Charles University, Department of Neurosurgery, Prague, Czech Republic; 3Regional Hospital, Neurocenter, Department of Neurosurgery, Liberec, Czech Republic; 4Regional Hospital, Department of Clinical microbiology and immunology, Antibiotic Centre, Liberec, Czech Republic; 5Na Homolce Hospital, Department of Clinical Pharmacology, Prague, Czech Republic

Introduction: Surgical site infection (SSI) is a risk in every operation wound, as it negatively impacts patient morbidity and mortality, and also increases financial demands, such as prolonged hospital stay, further antibiotics and surgical procedures. The aim of this study was to analyse SSI and its risk factors after thoracic and lumbar surgery.

Methods: A six-year monocentric observation prospective cohort study monitored the incidence of SSI, wound complications and further risk factors in 274 consecutive patients after planned thoracic and lumbar surgery for degenerative disease, trauma and tumour. All patients received short antibiotic prophylaxis (before and during long operations). All wound complications and SSI were monitored up to 30 days and 1 year after operations. We searched for risk factors for SSI in multivariate logistic regression analysis.

Results: We recorded 22 incidences of SSI (8.03%; superficial 5.84%, deep 1.82%, organ 0.36%). There were no differences between these two groups in age (p=0.906), gender (p=0.545), body mass index (p=0.858), spine diagnoses (p=0.745), number of vertebrae (p=0.815), spine localization (p=0.808), use of metal (p=0.428), American Society of Anesthesiologists, ASA Score (p=0.766), urine catheters (p=0.206), drainage (p=0.498), corticoids (p=0.409), transfusions (p=0.262), ulcer prophylaxis (p=0.409) and diabetes mellitus (p=0.811), but SSI had longer hospital stay (p=0.003) and hospital wound complication (p<0.001). Predictor of SSI in multivariate logistic regression analysis was hospital wound complications (OR 20.40, 95% CI 7.32-56.85, (p<0.001) and warm season (OR 2.92, 95% CI 1.03-8.27, p=0.044).

Conclusions: Contrary to the prevailing literature, our study did not identify corticoids, diabetes mellitus or transfusions as risk factors for the development of SSI, but only wound complications and warm seasons.

P053 Measuring the impact of verbal immediate feedback at the end of observational session, on hand hygiene compliance among health care workers (HCWS) in a general intensive care unit (GICU), preliminary results

I Livshiz-riven1, L Koyfman2, A Borer1, A Gushanski1, S Askira1, V Habar1, E Ivanov3, M Klein2, A Danziger2, O Azulay3, R Nativ1, E Brotfain4

1Ben Gurion University of the Negev, Infection Control Unit, Soroka Medical Center. Faculty of Health Science, Ben-Gurion University of the Negev, Beer Sheva, Israel, Beer Sheva, Israel; 2Ben Gurion University of the Negev, Department of Anesthesiology and Critical Care, General Intensive Care Unit, Soroka Medical Center. Faculty of Health Science, Ben-Gurion University of the Negev, Beer Sheva, Israel, Beer Sheva, Israel; 3Ben Gurion University of the Negev, Department of Internal Medicine A, Soroka Medical Center. Faculty of Health Science, Ben-Gurion University of the Negev, Beer Sheva, Israel, Beer Sheva, Israel; 4Ben Gurion University of the Negev, Department of Anesthesiology and Critical Care, Beer Sheva, Israel

Introduction: Health care associated infections (HCAI) are a major problem for patient safety in intensive care units (ICUs). There are different education measures (written material with reminders, continuous feedback, interventions involving novel equipment) on performance of hand hygiene. In the present study, we assessed the impact of immediate verbal feedback on performance of hand hygiene by health care workers using a new Continuous Closed-circuit Television Monitoring (CCTV) method and direct, overt analog observation method.

Methods: This is an interventional study. We conducted overt – direct observations and covert - CCTV observational sessions to measure hand hygiene compliance before and after interventional measures of health care workers (HCWs) in our ICU. As interventional measures, we used personal verbal immediate feedback at the end of the overt observational session, performed by infection control nurse.

Results: Overall, 2500 opportunities to perform hand hygiene. The compliance rate at the beginning four months of the study, was 190/319 (59.6%), then, it increased to 293/309 (77%) when measured during sessions with feedback, and 112/167 (66%) when measured by overt observations without immediate feedback. The measurements dropped again to 487/990 (49.2%) in overt observations without immediate feedback, while the compliance rate using the feedback sessions continued being the highest measurement (250/403, 62%). Covert, CCTV sessions increased only from 130/533 (24.3%) (before intervention) to 35/120 (27.1%) after intervention measures.

Conclusions: We consider that our findings reflecting the inefficacy of the “verbal feedback on performance” method. We believe that it needs additional scrutiny and combining additional intervention strategies to improve hand hygiene compliance.

P054 Cytomegalovirus infection in immunocompetent intensive care unit patients

W Sellami, I Ben mrad, Z Hajjej, M Chniti, I Labbene, M Ferjani

Department of Critical Care Medicine and Anesthesiology, Military Hospital, Tunis, Tunisia

Introduction: Cytomegalovirus (CMV) has been recognized as an important pathogen in immunocompromised individuals for as long time. In recent years, some studies have focused on CMV infection among immunocompetent intensive care patients. The results are inconsistent and the impact of this virus on the prognosis of these patients is not solved. Our purpose were to determine the prevalence, the risk factors and the consequence of CMV infection in immunocompetent intensive care unit patients.

Methods: Observational retrospective case-control study comparing two groups of intensive care patients: CMV-positive and CMV-negative. Patients suspected of developing CMV infection were included. Clinical, demographic and biological characteristics and patient´s care were pointed out to identify risk factors. CMV impact on prognosis was judged by the complications developed and mortality. Another comparison among infected patients between the deceased and the living was carried out in order to determine CMV morbidity and mortality factors.

Results: CMV prevalence was 21% in immunocompetent patients suspected of having CMV infection. No significant differences in age, sex, comorbidities, severity, ventilation, use of amines and corticosteroids were found. Transfusion history (p=0.003) ans sepsis (p=0.013) were identified as risk factors in the univariate analysis. In the multivariate analysis, only transfusion history was a risk factor (p=0.004). CMV was not associated with significant morbidity and mortality. Severity score (IGS II) (p=0.02), use of corticosteroids (p=0.002), mechanical ventilation p=0.023), and bacteremia (p=0.001) were associated with mortality in the comparison between the deceased and the living.

Conclusions: CMV infection is common in immunocompetent intensive care patients. Transfusion history is a risk factor of infection. CMV is a marker of the severity of the underlying disease of patients rather than a cause of morbidity and mortality

P055 Necrotising soft tissue infections: the first 24h on ICU. Impact of inflammation and organ dysfunction on mortality

A Ogica, C Burdelski, S Kluge, G De Heer

Uniklinikum Eppendorf, Klinik für Intensivmedizin, Hamburg, Germany

Introduction: Necrotizing soft tissue infections (NSTI) are characterised by extensive tissue necrosis, triggering an overwhelming inflammatory response like sepsis or septic shock [1]. The mortality rate is high and the search for predicting factors has brought conflicting results. We hypothesized that inflammation parameters and organ dysfunctions in the first 24h may correlate with mortality on the intensive care unit (ICU).

Methods: We analysed retrospectively electronic data from patients who were admitted to our University Hospital during 2009-2017. For the statistical analysis we used SPSS, version 25.0.

Results: 59 patients with NSTI were admitted during the study period. There were 41 males (69.5%) and the median age was 53 years. 15 patients (25.4%) died while on ICU, all of them presented a septic shock at admission. In the univariate analysis just the SOFA Score correlated significantly to mortality (p=0.007). Plotting a receiver operator characteristic curve for the SOFA score against mortality, we obtained an area under the curve of 0.755. A SOFA score ≥7 showed 86% sensitivity and 64% specificity in predicting ICU mortality. Both kidney and liver dysfunction were significantly linked to a higher risk of mortality. An association of four or more organ dysfunctions increased the risk of death by a factor of 8.7.

Conclusions: SOFA score and presence of liver or kidney dysfunction respectively in the first 24h correlated well with an increased risk of death. The different inflammatory markers showed no predicting value towards the risk of mortality.


1. Jabbour G et al. World J Emerg Surg 2016; 11: 40.

P056 Ventilator-associated pneumonia and PaO2/FiO2 value accuracy: a new paradigm?

C Dominedò1, T Sequeira2, C Cillòniz3, I Martin-Loeches4, M Ferrer3, A Torres3

1Policlinico Agostino Gemelli - Università Cattolica del Sacro Cuore, Department of Intensive Care and Anestesiology, Rome, Italy; 2Hospital Prof. Doutor Fernando Fonseca, EPE, Amadora, Pulmonology Department, Lisbon, Portugal; 3Hospital Clinic of Barcelona; August Pi i Sunyer Biomedical Research Institute - IDIBAPS, University of Barcelona; Biomedical Research Networking Centres in Respiratory Diseases (Ciberes), Department of Pneumology, Barcelona, Spain; 4St. James’s Hospital, Multidisciplinary Intensive Care Research Organization (MICRO), Dublin, Ireland

Introduction: Ventilator-associated pneumonia (VAP) is one of the leading infection in critically ill patients. Lacking definitive diagnostic criteria, worsening gas exchange assessed by PaO2/FiO2 ≤240 mmHg has been proposed as a diagnostic criterion. The aim of the study was to assess the adequacy of PaO2/FIO2 ≤240 mmHg to diagnose VAP.

Methods: We prospectively included 255 adult patients admitted between 2007 and 2017 to the ICUs of the Hospital Clinic of Barcelona with a clinical diagnosis of VAP. Patients were divided according to PaO2/FIO2 ≤240 mmHg (group 1) or >240 mmHg (group 2) at pneumonia onset. The study was approved by the Ethics Committee of our institution. Patients next of kin provided written informed consent.

Results: PaO2/FiO2 was lower than 240 mmHg in 171 (67%) patients. Compared to patients in group 2, patients in group 1 were less severely ill at admission but presented a higher SOFA and CPIS score and a greater incidence of ARDS and shock at pneumonia onset (Fig 1). 117 (69%) patients in group 1 had a microbiological diagnosis of pneumonia, compared to 71 patients (85%) in group 2 (p=0.007). PaO2/FIO2 ≤240 mmHg was associated with less probability of having microbiological diagnosis of pneumonia (OR 0.40, 95% CI 0.21 to 0.79, p=0.008). When adjusted for other variables significantly associated with positive microbiology, PaO2/FIO2 ≤240 mmHg remained significantly associated with less probability of a microbiological diagnosis (adjusted OR 0.34, 95% CI 0.12 to 0.94, p=0.038). Hospital mortality was significantly higher in patients in group 1 compared to group 2 (42% vs 29%, p=0.044). However, no difference was found in non-response to treatment, ICU and hospital stay, ICU mortality (Table 1) and 90-days survival (Fig 2).

Conclusions: A significant higher number of patients with VAP didn’t have a definitive etiological diagnosis when using the proposed threshold criteria of PaO2/FIO2 ≤240 mmHg. PaO2/FiO2 ratio does not seem a good predictor of etiology in patients with VAP.

Table 1 (abstract P056). Outcome variables
Fig. 1 (abstract P056).

See text for description

Fig. 2 (abstract P056).

90-days survival

P057 The relationship between ex-vivo stimulated TNF-alpha levels and the development of nosocomial infections in critically ill mechanically ventilated patients

G Levin, J Boyd, D Maslove, S Sibley, P Norman, A Day, J Muscedere

Queen´s University, Department of Critical Care Medicine, Kingston, Canada

Introduction: Immunological dysfunction is common in critically ill patients but the optimal method to measure it and its clinical significance are unknown. Levels of tumor necrosis factor alpha (TNF-α) after ex-vivo whole blood stimulation with lipopolysaccharide has been proposed as a possible method to quantitate immunological function. We hypothesized that patients with a lower post-stimulation TNF-α level would have increased rates of nosocomial infections (NIs) and worse clinical outcomes.

Methods: A secondary analysis of a phase 2 randomized, multi-centre, double-blinded placebo controlled trial [1]. There were no differences in allocation groups; all the patients were analyzed as one cohort. On enrolment, whole blood was incubated with LPS ex-vivo and TNF-α level was measured. Patients were grouped in tertiles according to delta and peak TNF-α level. The primary outcome was the development of NIs; secondary outcomes included 90-day mortality.

Results: Data was available for 201 patients. Baseline characteristics and outcomes are reported in Tables 1 and 2. Patients in the highest tertile for post LPS stimulation delta TNF-α compared to the lowest tertile were younger, had a lower acuity of illness and had lower baseline TNF-α. When grouped according to peak post-stimulation TNF-α levels, patients in the highest tertile had higher serum TNF-α at baseline. Both comparisons showed no difference between NIs and clinical outcomes between tertiles. In multi-variate analysis peak or delta TNF-α were not associated with the occurrence of NIs.

Conclusions: Admission ex-vivo stimulated TNF-a level is not associated with the occurrence of NIs or clinical outcomes. Further study is required to evaluate the ability of this assay to quantify immune function over the course of critical illness.


1. Muscedere J et al. Crit Care Med 46:1450, 2018

Table 1 (abstract P057). Characteristics and outcomes of patients based on admission peak TNF-α levels post LPS stimulation (Age, APACHE II, baseline TNF-α, peak TNF-α post LPS stimulation: mean ± SD; sex, patients with an NI, 90-day mortality: n (%); TNF- a measured in pg/mL)
Table 2 (abstract P057). Characteristics and outcomes of patients based on admission change in TNF-α levels post LPS stimulation (Age, APACHE II, baseline TNF-α, peak TNF-α post LPS stimulation: mean ± SD; sex, patients with an NI, 90-day mortality: n (%); TNF-α measured in pg/mL)

P058 Focused study on ventilator associated condition (VAC) with infection-related ventilator associated complication (IVAC) in a mixed adult unit

D Terzi, K Qamar, N Parekh

Queen Elizabeth Hospital, Department of Critical Care, Birmingham, United Kingdom

Introduction: We believe traditional ventilator associated pneumonia (VAP) is limited by its complexity, subjectivity and marginal attributable mortality. It generates debate but not a matrix. The new paradigm VAC, designed by CDC in 2013 broadens the focus of surveillance, is simple, objective & automatable [1].

Methods: Inclusion Criteria: All patients intubated for at least 48 hours. Exclusion Criteria: All elective post-cardiac surgery. Data Capture: Specifically designed excel spreadsheet allowing classification of events into VAC, IVAC & use of antimicrobials. Follow Up: Extubation or death.

Results: A total of 133 patients were enrolled between 3rd September to 20th October in 2018. Sixteen patients were confirmed with IVAC (12%) (Fig 1). Our past focused studies based on same tool showed IVAC rate of 17% in 2014 & 14% in 2015 (Table 1). The major reason for this reduction is decrease in percentage of ventilated patients (45 vs 40) as well as slight reduction in length of stay on ventilator (3 vs 2.6 days) (Fig 2).

Conclusions: We believe nominal reduction in IVAC rate is due to decrease in % of ventilated patients since 2011 & a slight reduction in ventilator days. We continue to explore possibility of automation of VAC & IVAC data [2].


1. Klompas M, NEJM 368:1472-5, 2013

2. VAE calculator

Table 1 (abstract P058). Trends (%) in VAC, IVAC and associated microbial usage
Fig. 1 (abstract P058).

Data analysis of 133 patients in current study

Fig. 2 (abstract P058).

Percentage ventilated and LOS (days) on ventilator

P059 Risk factors for sepsis in very old patients with community-acquired pneumonia

C Dominedò1, C Cillóniz2, A Ielpo3, M Ferrer2, A Gabarrús2, D Battaglini4, J Bermejo-Martin5, A Meli6, C García-Vidal7, A Liapikou8, A Torres2

1Policlinico Agostino Gemelli - Università Cattolica del Sacro Cuore, Department of Intensive Care and Anestesiology, Rome, Italy; 2Hospital Clinic of Barcelona; August Pi i Sunyer Biomedical Research Institute - IDIBAPS, University of Barcelona; Biomedical Research Networking Centres in Respiratory Diseases (Ciberes), Department of Pneumology, Barcelona, Spain; 3University of Parma, Departments of Medicine and Surgery, Respiratory Disease and Lung Function Unit, Parma, Italy; 4Policlinico San Martino, University of Genova, Department of Surgical Sciences and Integrated Diagnostic, Genova, Italy; 5Hospital Clínico Universitario de Valladolid/IECSCYL, Group for Biomedical Research in Sepsis (Bio Sepsis), Valladolid, Spain; 6University of Milan, Department of Anesthesiology and Intensive Care Medicine, Milan, Italy; 7Hospital Clinic of Barcelona, Infectious Diseases Department, Barcelona, Spain; 8Sotiria Chest Diseases Hospital, Mesogion, Respiratory Department, Athens, Greece

Introduction: There is limited information about sepsis in very old patients hospitalized with community-acquired pneumonia (CAP).

Methods: We conducted a retrospective study using data that were prospectively collected at the Hospital Clinic of Barcelona. We included all very old patients (>80 years), with no severe immunosuppression, hospitalized with CAP between 2005 and 2017. We aimed to investigate the prevalence, etiology, risk factors and clinical outcomes of this population, comparing patients with and without sepsis (defined according to SEPSIS-3 criteria). The study was approved by the Ethics Committee of our institution (register: 2009/5451). Written informed consent was waived because of the non-interventional study design.

Results: Among 4,190 patients hospitalized with CAP, 1,136 (27%) were very old (>80 years). The incidence of sepsis in very old patients was 70% (n = 795). There was no significant difference in the distribution of pathogens in patients with and without sepsis (Figure 1). Male sex (OR 1.86 [95% CI 1.21 to 3.25]) and chronic renal disease (OR 2.01 [95% CI 1.22 to 3.29]) were independent risk factors for sepsis in the multivariable analysis, while prior antibiotic therapy before admission (OR 0.71 [95% CI 0.53 to 0.95]) was independently associated to a lower risk of sepsis (Figure 2). One-year mortality was higher in very old patients with sepsis compared with those without sepsis (Table 1). A propensity-adjusted multivariable analysis showed that risk factors for 30-day mortality in septic patients were chronic renal disease (OR 2.62 [95% CI 1.36 to 5.03]) and neurological disease (OR 2.56 [95% CI 1.47 to 4.46]), while diabetes mellitus (OR 0.43 [95% CI 0.21 to 0.87]) was a protective factor.

Conclusions: In very old patients hospitalized with CAP, antibiotic therapy before admission was associated with a decreased risk of sepsis, whereas diabetes mellitus was associated with a decreased risk of 30-day mortality.

Table 1 (abstract P059). Clinical outcomes according to sepsis
Fig. 1 (abstract P059).

Distribution of microbial aetiology according to sepsis

Fig. 2 (abstract P059).

Significant univariate and multivariable logistic regression analyses for sepsis (N = 1,136)

P060 Can we predict Legionnaire´s disease in patients presenting with community-acquired pneumonia to the emergency department by clinical parameters?

R Bolliger, O Neeser

Kantonsspital Aarau, Innere Medizin, Aarau, Switzerland

Introduction: Legionella species may cause life-threatening pneumonia and thus need early treatment. Differentiating Legionella pneumoniae (LP) from other types of pneumonia including Mycoplasma pneumoniae (MP), Steptococcus pneumoniae (SP) and viral types of community-acquired pneumonia (CAP) has important implications regarding antibiotic therapy. Current testing options for LP infection have limited sensitivity leading to time delays in treatment and to usage of empirical broad-spectrum antibiotics. Recently, a Legionella Scoring system based on six parameters has been proposed. We aimed to independently validate this score and investigate whether additional clinical and laboratory parameters would further improve its accuracy.

Methods: We analyzed patients hospitalized in a tertiary care hospital between 2013 and 2017 with CAP and a defined etiology. Association and discrimination were assessed using logistic regression analysis and area under the receiver operator characteristic curve (ROC AUC).

Results: We included 713 patients with CAP including 33 (5%) with LP; 56 (8%) with MP; 164 (23%) with SP and 460 (64%) with viruses. All clinical predictors included in the Legionella score were associated with LP CAP. The score overall showed high discrimination between LP and other CAP etiologies with an AUC of 0.83 (95% CI 0.76 to 0.90), which was better than each parameter alone. Results were similar for subgroups based on each of the different CAP types. Additionally, we found that a history of nausea further improves the diagnostic accuracy of the legionella score to an AUC of 0.84 (95% CI 0.77 to 0.91).

Conclusions: In patients hospitalized with CAP, a high Legionella score on admission strongly predicts LP infection and thereby can optimize the empiric antibiotic management. A clinical history of nausea further improves diagnosis. Systematic use of this scoring system in conjunction with other diagnostic tests may improve the diagnostic and therapeutic management of patients presenting with CAP.

P061 Ventilator-associated pneumonia due to multidrug-resistant Acinetobacter baumannii: risk factors and mortality relation with resistance profiles

A Ciginskiene1, A Dambrauskiene2, D Adukauskiene3, V Pilvinis3, I Zubaviciute3

1Hospital of Lithuanian University of Health Sciences Kauno Klinikos, Department of Intensive Care, Kaunas, Lithuania; 2Hospital of Lithuanian University of Health Sciences Kauno Klinikos, Infection Control Service, Kaunas, Lithuania; 3Hospital of Lithuanian University of Health Sciences Kauno Klinikos, Department of Intensive Care, Kaunas, Lithuania

Introduction: Acinetobacter baumannii (AcB) remains one of the most prevalent ventilator-associate pneumonia (VAP) causing pathogen. In recent years, share of drug resistant AcB strains across Europe was found to be steadily increasing. Consequently, in 2017, AcB was included in the WHO global priority list of drug-resistant bacteria to highlight the need for the research development. The aim of this study was to identify the relation of risk factors for ventilator-associated pneumonia (VAP) and mortality with drug resistance profiles of AcB.

Methods: A retrospective cohort study of patients treated in medical-surgical ICUs with drug-resistant strains of AcB as pathogens of VAPover a 2-year period was carried out.

Results: The data of 60 medical-surgical ICUs patients with VAP due to drug-resistant AcB were analysed. The proportions of multidrug-resistant (MDR), extensively drug-resistant (XDR), and potentially pandrug-resistant (pPDR) AcB were 13.3%, 68.3%, and 18.3%, respectively (p<0.05). The SAPS II scores on ICU admission were 42.6, 48.7, and 49 (p<0.048); hospital length of stay prior to ICU was 0, 1, and 2 days (p<0.036), prior to mechanical ventilation - 0, 0, and 3 days (p<0.013), and carbapenem use prior to VAP - 50%, 29.3%, and 18.2% (p <0.036) in MDR, XDR, pPDR AcB VAP groups respectively. The overall in-hospital mortality rate was 63.3%. In MDR, XDR, and pPDR AcB VAP groups it was 62.5%, 61.3%, and 72.7%, respectively (p = 0.772).

Conclusions: The VAP risk factors: higher SAPS II score, increased hospital length of stay prior to ICU and mechanical ventilation were related to higher resistance profile of AcB. Carbapenem use was found to be associated with the risk of MDR AcB VAP. Mortality due to drug-resistant AcB VAP was high, but not associated with AcB drug resistance profile. Thus, timely mechanical ventilation and ICU treatment may reduce the risk of VAP due to higher drug-resistant AcB, especially in more severely ill patients.

P062 Nosocomial wound infections in intensive care unit review of microbiology over 5 years period (2013-2017)

A Al Bshabshe1, M Joseph2, A Assiri3, I Asiri 4, M Hamid1, Y Alqarni 4, K Sinnah4

1King Khalid University, Medicine/Critical Care, Abha, Saudi Arabia; 2King Khalid University, Department of Microbiology, Abha, Saudi Arabia; 3KKUMC, Critical Care, Abha, Saudi Arabia; 4Critical care department, ACH, Abha, Saudi Arabia

Introduction: sepsis following traumatic and surgical intervention increases morbidity, mortality, cost and length of patient stay in hospital. The aim of this study was to identify the major pathogens associated with wounds infection and to review their antimicrobial reactions.

Methods: A 5-year review of nosocomial wound infection and colonization in patients admitted to the intensive care unit of tertiary care Hospital southern region of Saudi Arabia from Jan. 2013 to Aug. 2017. Patients of all ages and gender who required ICU attention at some point and defined as nosocomial infection using standard CDC criteria and presented with various degrees of wound and bed sore infections were included in the study. Data on bacterial isolates (n= 536) and reactions to antimicrobials (n= 51) were analyzed .

Results: There were 379 episodes of wound and 157episodes of bedsore infections. The most common organisms Klebsiella pneumoniae (22.8%) followed by Proteus mirabilis (15.1%); Acinetobacter baumannii (12.7%); Escherichia coli (10.8%); Pseudomonas aeruginosa (10.1%); Morganella morgani (7.6%); Providencia stuartii (3.4%); Staphylococcus aureus (3.0%); Enterobacter aerogenes (1.5%) and Methicillin-resistant Staphylococcus aureus (MRSA) (1.3%) (Fig 1). The percentage sensitivity of the 536 organisms to the 51 antimicrobial agents was 39.2%; intermediate sensitivity was 3.3% and resistant was 57.5% (Fig 2).

Conclusions: Data from this and other studies supports the hypothesis that high incidence of gram negative bacilli (91.4%) in particular Klebsiella pneumoniae, Proteus mirabilis and Acinetobacter baumannii are more common in tropical regions compared to gram positive bacteria (8.6%) mainly Staphylococcus aureus. This requires strong infection control actions to enhance patient care.

Fig. 1 (abstract P062).

Major organisms (%) causing wound infections in ICU (2013-2017)

Fig. 2 (abstract P062).

sensitivity pattern of the major organism causing wound infection in ICU

P063 Predicting performance of risk factors for MDR pneumonia according to the 2017 international ERS/ESICM/ESCMID/ALAT guidelines

C Dominedò1, A Ceccato2, M Ferrer2, A Gabarrús2, A Torres2

1Policlinico Agostino Gemelli - Università Cattolica del Sacro Cuore, Department of Intensive Care and Anestesiology, Rome, Italy; 2Hospital Clinic of Barcelona; August Pi i Sunyer Biomedical Research Institute - IDIBAPS, University of Barcelona; Biomedical Research Networking Centres in Respiratory Diseases (Ciberes), Department of Pneumology, Barcelona, Spain

Introduction: After the 2017 International ERS/ESICM/ESCMID/ALAT guidelines publication [1], no study has evaluated the identified risk factors for multi drug resistant pathogens (MDRP). We aimed to assess the predictive performance of these risk factors.

Methods: We analyzed a cohort of patients prospectively collected between 2004 and 2017 in six ICUs of the Hospital Clinic of Barcelona, Spain. Adult patients with no severe immunosuppression and a diagnosis of hospital-acquired pneumonia and ventilator-associated pneumonia with confirmed microbiology were enrolled. The study was approved by the Ethics Committee of our institution. Patients or their next of kin provided written informed consent.

Results: Among the 314 patients included in the study, 312 were at high MDR risk: 158 (51%) had no septic shock (group 1) whereas 154 (49%) had septic shock (group 2) (Table 1). Despite no difference in the microbiology, at pneumonia diagnosis, compared to group 1, patients in group 2 presented a higher SOFA [5 (4;7) and 9 (8;12), p<0.001] and ARDS [12 (8) and 26 (17), p 0.014]. Admission to hospital settings with high rate of MDRP and prior antibiotic use showed the highest prevalence in the overall population (81% and 79% respectively), with a high sensitivity (91% and 84% respectively) and negative predictive value (NPV) (83% and 75% respectively). Previous isolation of MDRP was the only risk factor presenting a specificity and a positive predictive value (PPV) of 100% and a high accuracy [AUC 0.80 (0.68 to 0.93)] (Fig 1). Despite the combination hospital settings with high rates of MDRP/previous antibiotic use had the highest prevalence (64%) in the overall population, only the combinations including previous isolation of MDRP showed a specificity and a PPV of 100%, with an AUC > 0.6 (Fig 2).

Conclusions: The guidelines categorized a higher proportion of patients with high risk of MDRP. Previous isolation of MDRP showed the best performance for MDR prediction.


Torres A et al. Eur Respir J. 50(3); 2017

Table 1 (abstract P063). Patients baseline characteristics at ICU admission
Fig. 1 (abstract P063).

Predictive performance of 2017 International ERS/ESICM/ESCMID/ALAT guidelines single risk factors for MDR pathogens

Fig. 2 (abstract P063).

Predictive performance of combinations of risk factors for MDR pathogens

P064 Monobacteremia of hospital-acquired gram-negative rods in Lithuanian university’s hospital intensive care unit

D Adukauskiene1, D Valanciene2, A Dambrauskiene3, A Adukauskaite4, L Jazokaite5, V Micpovilyte5, A Vitkauskiene6

1Hospital of Lithuanian University of Health Sciences Kaunas Clinics, Intensive Care Unit, Kaunas, Lithuania; 2Klaipeda Seamen´s Hospital, Anaesthesiology and intensive care department, Klaipeda, Lithuania; 3Hospital of Lithuanian University of Health Sciences Kaunas Clinics, Infection Control Service, Kaunas, Lithuania; 4Medical University of Innsbruck, Cardiology, Innsbruck, Austria; 5Hospital of Lithuanian University of Health Sciences Kaunas Clinics, Intensive Care Clinic, Kaunas, Lithuania; 6Hospital of Lithuanian University of Health Sciences Kaunas Clinics, Laboratory Medicine Clinic, Kaunas, Lithuania

Introduction: Growing antimicrobial resistance among Gram-negative rod (GNR) strains is a worldwide issue. Flora monitoring is associated with right first choice of antimicrobial treatment. The aim of study was to analyze sensitivity of hospital-acquired (HA) GNR monobacteremia strains to antimicrobial drugs and risk factors for mortality in Intensive Care Unit (ICU).

Methods: Ongoing retrospective cohort study of patients treated in ICUs of Kaunas Clinic’s with positive blood culture for GNR taken after 72 hrs of hospitalisation during 7yrs period was carried out.

Results: We’ve found 196 cases of HA bacteremia due to GNR: 72 (36.7%) caused by Acinetobacter spp. (P=.03), 38 (19.4%) by Escherichia coli, 34 (17.3%) by Klebsiella spp., 19 (9.69%) by Serratia spp. and just few others. Sensitivity to carbapenems (n=182, 92.9%), amikacin (n=131, 66.8%) cefoperazon/sulbactam (n=116, 59.2%), and piperacillin/tazobactam (n=93, 47.4%) was found. There were found 151 (77%) multi-drug-resistant (MDR) and among them 64 (42.4%) – extra-drug-resistant strains. Among Acinetobacter spp. there were found 61 (84.7%) MDR strain (P=.04). In general it was related with male gender (n=90/118, P=.04, OR=4.56, CI95%=2.4-15.4), elderly (P=.001, RR=13.3), acute kidney injury (AKI) (n=98/128, P=.03, OR=0.84, CI95%=0.35-0.92). In multi-variate analysis lethal outcome 73.5% (n=144) was associated with male gender (P=.04, OR=2.92, CI95%=1.98-8.67), mechanical ventilation (P=.04, OR=4.125, CI95%=2.70-6.34), septic shock (P=.001, OR=8.21, CI95%=1.33-6.26), AKI (P=.02, OR=5.67, CI95%=1.18-27.25), MDR strain (P=.01, OR=3.47, CI95%=2.46-7.56).

Conclusions: 7yrs study revealed Acinetobacter spp. as predominant, mostly MDR pathogen of HA GNR bacteremia in ICUs. HA MDR GNR bacteremia was related with male gender, elderly and AKI. High sensitivity of GNR to carbapenems was found. High rate of mortality 74% was associated with male gender, mechanical ventilation, septic shock, AKI and MDR strain.

P065 Risk factors and prognosis of critically ill patients colonized with KPC-producing enterobactereacea in Brazil

E Bogossian1, D Salgado1, S Nouér2

1Universidade Federal do Rio de Janeiro, Intensive Care department, Rio de Janeiro, Brazil; 2Universidade Federal do Rio de Janeiro, Infectious Disease Department, Rio de Janeiro, Brazil

Introduction: Of great concern is the dissemination in health care environments of multi-drug resistant (MDR) bacteria, especially Klebsiella pneumoniae carbapenemase (KPC)-producing enterobactereaceae.

Methods: Our objective is to identify the differences in risk factors and outcomes between patients who did and did not acquire KPC- producing bacteria during their stay in the ICU of an university hospital in Rio de Janeiro, Brazil. We designed a nested case – control study of a retrospective cohort from May 2014 to June 2015. Three hundred and thirty nine patients were admitted to the ICU. 36 cases (patients infected or colonized by KPC-producing bacteria and 292 ‘controls’ (all other patients who did not have MDR isolation). The two groups were compared according to demographic clinical and microbiological data.

Results: There was no significant differences in gender, age, severity scores between the two groups. Patients with KPC- producing bacteria had longer ICU and hospital stays than control patients and they required more life-support such as mechanical ventilation, vasopressor use and dialysis. In hospital mortality rate was higher in KPC- producing bacteria group (63.9% vs 35.3% p=0.001). A multivariate analysis showed that mechanical ventilation (OR: 2.92; CI 1.02-8.32) and previous use of antibiotics (OR: 13.39; CI 1.62-111.20) were risk factors for acquiring KPC.Factors associated with in-hospital mortality were: male gender, (OR: 3.29; CI 1.54-7.05), Charlson score (OR: 1.26; CI 1.09-1,45) and mechanical ventilation (OR: 16.63; CI 4.79-57.70).

Conclusions: Colonization was associated with longer ICU and hospital LOS, more requirements of life-support and higher in-hospital mortality rate. Previous antibiotic use and mechanical ventilation were the most important risk factors for patients becoming colonized. Factors associated with mortality were male gender, Charlson score and mechanical ventilation.

P066 Infection with ESBL-E in the ICU: analysis of antibiotic therapy

S Cuypers, L De Bus, J De Waele, P Depuydt

UZ Gent, Intensive Care, Gent, Belgium

Introduction: Carbapenems are usually the preferred antibiotics for the treatment of severe infections caused by extended-spectrum Β -lactamase-producing Enterobacteriaceae (ESBL-E), even in the presence of documented susceptibility for other antibiotic classes. Ecological arguments, on the other hand, provide rationale to limit carbapenem use as much as possible.

Methods: We retrospectively evaluated carbapenem and non-carbapenem antibiotic treatment for ESBL-E infections in the intensive care unit (ICU) of Ghent University Hospital during a 4-year period (2013-2016). Clinical cure was defined as stop of antibiotics together with resolution of clinical signs of infection.

Results: Empirical antibiotics were appropriate in 98 of 132 episodes (74%) and consisted of carbapenem and non-carbapenem monotherapy in 37% and 63%, respectively. Non-carbapenem antibiotics were appropriate in 37%. Appropriate empirical therapy (n=98) consisted of carbapenem and non-carbapenem antibiotics in 50.0% of episodes each. Carbapenems were de-escalated in 10/49 episodes (20%); appropriate non-carbapenem antibiotics were changed in 18/49 episodes (36%); in 1, respectively 3 episodes, this was due to clinical failure or emerging resistance. Clinical cure was observed in 86% and 90% of episodes appropriately treated with carbapenem or non-carbapenem antibiotics, respectively; in this subgroup, a bivariate analysis showed absence of clinical cure to be associated with septic shock (OR 1.8; 1.1-3.2), but not with carbapenem antibiotics (OR 1.3; 0.5-3.6).

Conclusions: In our ICU, non-carbapenem empirical therapy in ESBL-E infection carried a high risk for inappropriateness. However, no clear relationship between the use of carbapenem versus non-carbapenem antibiotics and clinical cure was found.

P067 The favorable outcome of Burkholderia cenocepacia blood stream infection (BSI) in cancer patients receiving chemotherapy is associated with early administration of antibiotic therapy

A Pronina, I Kurmukov, O Sekhina, S Kashia, S Menshikova

“N.N. Blokhin National Medical Research Center of Oncology” of the Ministry of Health of the Russian Federation, Moscow, Russia

Introduction: B. cenocepacia BSI is a rare and dangerous complication in cancer patients (pts) receiving chemotherapy. We described the causes of B. cenocepacia associated bacteremia, characteristics of clinical presentation and outcomes in patients receiving chemotherapy for solid tumors.

Methods: 11 adult patients (median age 53 years, 2 males, 9 with permanent vascular access) receiving chemotherapy with B. cenocepacia BSI confirmed by a culture-based study (identification by MALDI-TOF with a confidence index over 2,300).

Results: Sanitary and epidemiological examination revealed the connection between infection and intravenous infusion of dexamethasone performed concurrently with chemotherapy. In 5 patients fever with chills and hypertension developed within 2 hours after infusion of the infected drug; empirical intravenous antibiotic therapy started immediately after collecting blood culture. In 6 patients fever appeared after 2-4 days outpatiently, so they received antibiotics per os. All these patients had permanent vascular access, and BSI was detected either the next chemotherapy course when fever reappeared (3 pts) while using vascular access, or as a result of a specific examination (3 pts). In all cases empirical antibiotic therapy started on the first day of fever, drug correction was performed in 6 patients according to results of bacteriological research. Septic shock developed in 1 patient, pneumonia in 3 patients. Permanent vascular access was preserved only in 1 case. All patients were cured and continued to receive antitumor treatment.

Conclusions: Detection of more than 1 case of B. cenocepacia BSI should be the reason for sanitary and epidemiological examination. A favorable outcome of BSI treatment is associated with the early start of antibiotic therapy and its correction after microbiological examination.

P068 Emerging infections due to Shewanella spp.: a case series in Hong Kong

WS Ng, HP Shum, WW Yan

Pamela Youde Nethersole Eastern Hospital, Department of Intensive Care, Chai Wan, Hong Kong

Introduction: Shewanella species are emerging opportunistic pathogens that can cause severe soft-tissue, respiratory, hepatobiliary, gastrointestinal infections, and bacteraemia. Most of the published information for this organism are limited to isolated case reports and small case series. Here we report the largest case series of infections caused by Shewanella species.

Methods: Patients admitted to a regional hospital in Hong Kong with Shewanella species infection from 1st Apr, 2010 to 30th Sep, 2018 were included. Demographics, antibiotics, microbiology and outcomes were retrospectively analyzed.

Results: In an 8.5-year period, we identified 71 patients with Shewanella species infection and 65% of them were male. Their median age was 76 (IQR 66-84). Among them, 27% had diabetes mellitus, 20% had chronic kidney disease and 31% had an underlying malignancy. Hepatobiliary sepsis (45%) was the most common presentation, while soft tissue and respiratory tract infection accounted for 28 and 17% of infection respectively. Among 93 collected specimens, 48% were S. algae, 9% were S. putrefaciens and 43% could be identified up to species only. The identified organisms were usually sensitive to Ceftazidime (98%), ciprofloxacin (95%), gentamicin (96%), piperacillin (88%) and sulperazon (89%). Seven patients (9.9%) required ICU care and their APACHE IV predicted risk of death was 0.34 (IQR 0.14-0.77). Their median ICU length of stay was 5 days (IQR 4-8) and all of them could be discharged from ICU. The median hospital length of stay for all patients was 6 days (IQR 3-16) and the hospital mortality was 13%.

Conclusions: This large case series suggested that Shewanella infections are commonly associated with underlying comorbidities, especially with malignancy. Antibiotic resistant isolates are uncommon.

P069 Multidrug-resistant bacteria in burns: incidence, and risk factors

L Debbiche1, A Mokline1, H Fraj1, M Ben Saad1, I Rahmani1, K El Feleh1, L Thabet2, A Messadi1

1Trauma and Biurn Center, Intensive Burn Care Department, Tunis, Tunisia; 2Trauma and Burn Center, Laborabory of Biology, Tunis, Tunisia

Introduction: Despite improvements in early treatment, survival following burn injury remains challenged by sepsis and emergence of Multidrug-Resistant Bacteria (MRB). The objective of our study was to assess epidemiological aspects and bacterial resistance patterns of bacteria isolated from intensive burn care unit and to identify risk factors.

Methods: A prospective, monocentric study was conducted from May to September 2018 in a burn unit in Tunisia. In all patients, a search for BMR carriage (skin, rectal, bronchial, urinary, and catheter samples) and for carbapenemases (VIM, NDM and OXA 48) by polymerase chain reaction (PCR) were carried out initially at admission then weekly.

Results: During the study period, 31 patients were included. The mean age was 35 years. They were 17 men and 14 women. The average burned surface area was 39%. Patients were transferred from another hospital structure in 80% of cases with a delay of 36 hours. Isolated MRB were: Enterobacteriaceae producing extended-spectrum Β -lactamases (ESBLs) (49%), Carbapenemase-producing Enterobacteriaceae (CPE) (87%), Methicillin- Resistant Staphylococcus aureus Strains (19%) and glycopeptide-resistant enterococci (GRE) (13%). Carbapenemase-encoding genes were detected in 22 patients: 12 at admission and 10 at the first week post admission. The genes detected were New Delhi metallo-Β -lactamase (92%), Verona integron-encoded metallo-Β -lactamase (58%), and oxacillinase-48 (58%). Univariate analysis identified risk factors associated with acquiring ESBLs and CPE were: secondary transfert; indwelling device, antibiotic use of aminosids and amox-clavulanic acid and non-compliance with hygiene measures.

Conclusions: Detection of Multidrug-Resistant Bacteria in our study was higher: ESBLs in 49% of cases and CPE in 87% .So, detection and isolation of these patients and strengthen infection control measures allows us to improve therapeutic efficacy and improves their prognosis.

P070 Incidence of developing repeated hospital acquired infection in intensive care unit; an emerging challenge in resource limited settings

E Shimber

Hawassa University Comprehensive Specialized Hospital, Emergency and Critical Care Medicine, Hawassa, Ethiopia

Introduction: Repeated incidence of acquiring infection in Critical care unit of a hospital is one of the burdens for the countries’ health system. Despite various researches have been done regarding hospital acquired infections, developing repeated attack still needs further investigation. This research tries to magnify the threats and suggest a way forward.

Methods: A retrospective survey was conducted on all patients admitted to an ICU of Hawassa University Comprehensive Specialized Hospital from July 2018 to October 2018. Repeated ICU-acquired infections were defined as reinfection with different strain of organism or relapse of previous infection with the same pathogen during patient’s stay in ICU after being admitted for more than 48 hours and being treated for initial infection appropriately. Infection diagnosis was made based on Centers for Disease Control criteria. Multivariate analysis was used. Finally, ICU-mortality rate was tasted among patients who developed the repeated ICU-acquired infection.

Results: A total of ninety-four patients were admitted to ICU. The rate of developing ICU acquired infection was 22% with repeated rate of infection being 8.5%. Overall mortality rate was 44%. Out of these, patients with ICU acquired repeated infection took only 4% of the proportion. Those of who stayed more than two weeks develop repeated attack at a rate of 87.5%. All Patients were having indwelling device. 50 % of them had chronic co morbid medical disease. Duration of ICU stay and indwelling device use had positive association with a strong Pearson correlation of 1. Drug sensitivity test revealed around 62.5% of them have developed resistance or intermediate sensitivity for initial antibiotics of choice.

Conclusions: Despite the low rate of repeated infection, multi-drug resistance pattern is high. The research reveals the need for proper antibiotic stewardship and application of standard infection control measures in ICUs.

P071 Changes of pathogens and clinical outcomes of ICU-acquired bacteremia as results of 5-year antimicrobial stewardship implementation

M Dementienko, V Gusarov, D Kamyshova, M Zamyatin

Federal State Public Institution “National Medical and Surgical Center named after N.I.Pirogov” of the Ministry of Healthcare of the Russian Federation, ICU, Moscow, Russia

Introduction: Conflicting results have been found regarding impact of the antibiotic stewardship program (ASP) on pathogens resistance and clinical outcomes of intensive care unit (ICU)-acquired bacteremia. The aim of the study was to evaluate the effect of ASP implementation on the incidence of ESKAPE-bacteremia and candidemia in ICU patients, level of antimicrobial resistance among causative microorganisms, as well as the hospital length of stay (LOS) and mortality of ICU patients with bacteremia.

Methods: Prospective interventional single-center study in the period Jan 2011 to Dec 2017. Intervention onset: 2013. The intervention: ASP, including multidisciplinary team building, antimicrobial therapy and prophylaxis protocols, infection control measures, education, internal audit. The data from 2012 (pre-intervention) and 2017 (intervention periods) reports were compared. We assessed incidence of ESKAPE-bacteremia and candidemia. The hospital LOS and mortality of ICU’s patients with bacteremia were secondary end-points.

Results: 4960 blood samples were analyzed, 917 (18.5%) positive blood cultures from 374 IÑU’s patients were obtained. Changes of the pathogens, causing ICU-acquired bacteremia are presented in the Table 1. The overall proportion of ESKAPE-bacteremia and candidemia decreased from 69 (53.9%) to 41 (29.3%), p <0.001.

Conclusions: Implementation of ASP in hospital allows to decrease incidence of ESKAPE-bacteremia and candidemia, which may lead to improved clinical outcomes in ICU’s patients (Fig 1).

Table 1 (abstract P071). Changes of the pathogens, causing ICU-acquired bacteremia
Fig. 1 (abstract P071).

Clinical outcomes of the patients with ICU-acquired bacteremia

P072 Association of multi-drug resistant (MDR), extended-drug resistant (XDR) and pan-drug resistant (PDR) gram negative bacteria and mortality in an intensive care unit(icu)

S Chatterjee1, S Sinha1, A Bhakta2, T Bera3, T Chatterjee3, S Todi1

1AMRI Hospitals, Critical Care, Kolkata, India; 2AMRI Hospitals, Microbiology, Kolkata, India; 3Jadavpore University, Clinical Research, Kolkata, India

Introduction: The study assessed mortality associated with non-MDR, MDR, XDR and PDR gram-negative bacterial (GNB) infections in the ICU.

Methods: Retrospective cohort study conducted in the ICU of a tertiary-care hospital in India between July2017 to January2018. All consecutive infectious episodes with positive culture for GNB and treated with antibiotic therapy within 24 hours of sample collection were included. Variable information collected included patient demography, APACHE IV score, culture report, drug resistance pattern of organism isolated(non-MDR, MDR, XDR/PDR). Primary outcomes were all cause ICU mortality which was compared between three groups-non-MDR, MDR, and XDR/PDR. All analysis were done after adjusting for disease severity.

Results: Of 323 eligible ICU patients 59.4% were males. Mean age significantly differed between non-MDR, MDR and XDR/PDR groups(68.2±13.6, 67.3±16.2 and 62.6±19.1 years respectively, p=0.003). Mean APACHE IV score was similar between groups(67.1±28.2, 69.7±24.9 and 75±27.7 respectively, p=0.07). Commonest isolated GNB was Klebsiella pneumoniae(35%). 39%(n=126) of patients had non-MDR organism, 26%(n=84) were infected with MDR and 35%(n=113) had XDR/PDR organisms in their bio-specimen. Commonest isolated MDR GNB was E.Coli(31%) while commonest XDR/PDR GNB was Klebsiella pneumoniae(55.8%). Overall ICU mortality did not significantly differ between non-MDR and MDR groups(19.4%vs.21.4%; p=0.7); there was significant difference in mortality between MDR and XDR/PDR groups(21.4%vs.41.6%; p =0.003). Risk-adjusted analysis showed higher odds of mortality for MDR infected patients compared to non-MDR group but this was not statistically significant(OR 1.14 95%CI 0.6-2.8; p=0.7). However, compared to MDR, XDR/PDR infected patients were significantly more likely to die in adjusted analysis(OR 2.5 95% CI 1.4-4.9, p=0.004).

Conclusions: Prevalence of XDR/PDR GNB and its associated mortality is high in our ICU compared to MDR GNB.

P073 First outbreak of colistin resistant Klebsiella pneumoniae in an adult intensive unit in South India

J Gopaldas1, G Raju1, B Hosdurg2, B Malavalli2

1Manipal Hospital, Bangalore, India, Critical care medicine, Bangalore, India; 2Manipal Hospital, Bangalore, India, Laboratory Medicine, Bangalore, India

Introduction: Colistin-resistant Klebsiella pneumoniae (CR-KP) is increasingly reported around the world. It is worrying to note emergence of resistance to last line of defence against MDR gram negative infections in regions endemic to Carbapenem resistance. We report the first outbreak of CR-KP co-producing carbapenemases in an adult Intensive Care Unit (ICU) from South India.

Methods: Retrospective analysis of all patients with Carbapenem resistant Klebsiella pneumoniae blood stream infection (BSI) was done between January 2017 and December 2017. Microbiological and clinical variables along with outcomes were analysed.

Results: Seven patients had CR-KP with no prior exposure to Colistin. All seven were modified Hodge test (MHT) negative making probability of blaKPC unlikely. In resource limited setting, analysis beyond MHT could only be performed for CR-KP samples. 2/7 samples belonging to CR-KP isolates produced the blaNDM-1 whilst 5/7 CR-KP isolates did not produce either blaKPC or blaNDM carbapenemases prompting hypothesis of blaOXA-48 or blaVIM as the causative factor. Compared to Carbapenem resistance only group, CR-KP group had higher APACHE II, ICU length of stay and mechanical ventilation duration. 28 day mortality was noted to be 56.3% for Carbapenem resistant and 43% for CR-KP groups. Aggressive infection control measures were undertaken with successful containment of CR-KP strains along with reduction in overall BSI.

Conclusions: Infection control measures form the backbone of patient care in centres showing endemicity for Carbapenem resistant Klebsiella to prevent Colistin resistance and also to reduce occurrence of overall blood stream infections.

P074 Rapid diagnosis of carbapenem resistance: experience of a tertiary care cancer center with multiplex PCR

S Mukherjee

Tata Medical Center, Critical Care Medicine, Kolkata, India

Introduction: Sepsis due to carbapenem resistant organisms has high mortality; inappropriate empirical antibiotic is one of the main causes of this poor outcome. On the contrary, “Too much” broad spectrum empiric antibiotics will increase drug resistance, even in community, because of selection pressure. So, early diagnosis of resistance pattern (carbapenemase genes) is crucial. Aim of this study is to compare rapid diagnostic test like polymerase chain reaction (PCR) with conventional culture sensitivity (C/S) to identify carbapenem resistance.

Methods: This is a prospective observational study done in Tata Medical Center, Kolkata, India. Real time multiplex PCR technique has been developed “in house” in our microbiology lab and can identify NDM, NDM1, KPC, OXA - 23, OXA - 48, OXA - 58 & VIM carbapenemase genes. Blood cultures were sent as per clinical & laboratory diagnosis of sepsis in ICU patients. Culture positive samples had been used for conventional C/S by VITEK 2 system along with PCR study to identify carbapenemase genes. Result of PCR technique was been compared with conventional C/S method.

Results: Multiplex PCR results were available within 3-4 hours of positive blood culture compared to conventional C/S method that takes 2 -3 days. Among 392 positive blood cultures, 146 samples were positive for carbapenemase genes. Most common gene identified was OXA - 48 (43%), followed by NDM1 (25%). Our PCR technique has very high sensitivity, specificity, positive & negative predictive value (99.35%, 94.65%, 93.1% & 99.43% respectively) while comparing with final C/S report by VITEK 2 system (Table 1). There was only one false negative diagnosis for carbapenem resistance.

Conclusions: Real time multiplex PCR for carbapenemase gene can be helpful for early diagnosis of carbapenem resistance and can help us to choose / modify antibiotics or to use ‘targeted therapy’. It is more practical to “rule - in” infection rather than “rule - out” by this technique.

Table 1 (abstract P074). Comparison of positive Carbapenemase Genes by bottle PCR with detection of carbapenem resistance by VITEK 2 method

P075 Carbapenemase producing enterobacteriaceae colonization in an ICU: risk factors and clinical outcomes

M Miranda, JP Baptista, J Janeiro, P Martins

Centro Hospitalar e Universitário de Coimbra, Intensive Care Unit, Coimbra, Portugal

Introduction: Carbapenemase-producing Enterobacteriaceae (CPE) colonization has been increasingly reported in Intensive Care Units (ICUs) since their first identification more than 20 years ago. Colonization with CPE seems to constitute a risk factor for mortality. The aim of our study was to identify associated risk factors and clinical outcomes among patients with fecal colonization by CPE admitted to a Portuguese tertiary hospital ICU.

Methods: A 2-year retrospective study was performed in patients with previous unknown CPE status (colonization or infection), admitted to our ICU. Rectal swabs were performed and analyzed using real-time polymerase chain reaction testing. Clinical records were reviewed to obtain demographic and clinical data.

Results: Of 903 patients admitted, 38 (4.2%) harbored CPE, 15 (39.5%) were colonized at admission and 23 (60.5%) acquired CPE colonization during ICU stay. The most frequent carbapenemase genes detected were KPC (87.5%) and VIM (12.5%). CPE carriers had high rates of hospitalization (previous or ongoing), invasive procedures (mainly intraabdominal surgery), malignancy (hematopoietic or solid tumor), antibiotic intake as well as slightly higher severity scores (SOFA, SAPS II and APACHE II). Nine patients developed CPE-related infections during ICU stay (4 pneumonia, 3 urinary tract infections, 1 peritonitis and 1 surgical wound infection). CPE colonization was associated with an increased length of stay (26.3 versus 14.3 days in the overall population, p=0.01) but not with ICU or in-hospital mortality (26.3% and 42% in CPE colonized patients versus 28.2 and 36.5%).

Conclusions: Hospitalization was the most important risk factor for CPE colonization. In this study and when compared with overall ICU population, patients colonized with CPE had significantly increased ICU length of stay. There seems to be a slightly higher in-hospital mortality rate, even though with no statistical significance.

P076 Indian experience with ceftazidime-avibactam used in treatment of serious infections in ICU setting: subset analysis from the REPROVE and RECLAIM trials

M Naik, C Adhav, P Gupta

Pfizer, Medical Affairs, Mumbai, India

Introduction: Gram-negative pathogens—particularly Pseudomonas aeruginosa and Enterobacteriaceae—predominate in nosocomial pneumonia (NP) and cIAI both. These infections are becoming difficult to treat with available treatment options due to growing antimicrobial resistance in India. Ceftazidime-avibactam has in-vitro activity against Gram-negative organisms producing class A, class C and some class D beta-lactamases. We carried out a qualitative analysis to assess the safety and efficacy outcomes of the Indian population cohorts involved in the REPROVE and RECLAIM trials.

Methods: In line with the global REPROVE protocol, Indian patients enrolled in the study with NP, were randomly assigned (1:1) to 2000 mg ceftazidime and 500 mg avibactam or 1000 mg meropenem. In the RECLAIM study, Indian patients with a diagnosis of cIAI were enrolled in the study and were randomly assigned (1:1) to receive either ceftazidime-avibactam (2000 mg of ceftazidime and 500 mg of avibactam) followed by metronidazole (500 mg); or meropenem (1000 mg). The primary efficacy outcome measure in the REPROVE and RECLAIM studies was clinical cure rate of CAZ-AVI compared with that of meropenem at TOC (test-of-cure) visit in pre-defined analysis sets. In both studies, Non-inferiority was concluded if the lower limit of the two-sided 95% CI for the treatment difference was greater than –12·5% in the primary analysis sets. As the Indian subset study was not statistically powered to detect a difference in the sub-group, we descriptively analysed the efficacy results in the Indian population and compared them with the overall results in the global trial. In addition, the study also analysed the safety of CAZ-AVI in the Indian patients by monitoring the number and severity of adverse events.

Results: In the REPROVE study, 90 patients across 4 centres from India while the RECLAIM study comprised of 142 patients from 8 centers in India. The efficacy results from the Indian subset analysis are summarized in Table 1. The safety analysis set in the REPROVE study included 78 Indian patients and in the RECLAIM study included 125 Indian patients, results summarized in Table 2. Overall, efficacy and safety outcomes were comparable with the global study.

Conclusions: The safety and efficacy outcomes in Indian population treated with CAZ-AVI as part of the REPROVE and RECLAIM studies were in line with the global trials. Although there are inherent limitations to this analysis due to the subgroup size, it provides information on clinical efficacy and safety relevant for the intended use of CAZ-AVI in Indian patients. CAZ-AVI showed to be an effective option for critically patients with HAP and cIAI and may be considered a good alternative to carbapenems in the ICU setting for the treatment of resistant pathogens.

Table 1 (abstract P076). Clinical efficacy in Indian subset from REPROVE and RECLAIM studies
Table 2 (abstract P076). Summary of adverse events (AE) upto last follow up visit (LFU) in REPROVE and RECLAIM studies

P077 Is there a difference in vancomycin level in patients admitted to medical compared to surgical ICU: a single center retrospective study in a mixed ICU setting

H Farhat1, R Jaafar1, R Farsakoury1, A Kassab1, D Ghaziri2, A Hallal1

1American University of Beirut Medical Center, Surgery, Beirut, Lebanon; 2American University of Beirut Medical Center, Pharmacy, Beirut, Lebanon

Introduction: Adequate utilization of vancomycin is essential for achieving therapeutic targets while avoiding clinical failure and development of antimicrobial resistance.Our aim is to determine whether there is a difference in vancomycin therapeutic blood level between surgical and medical patients in intensive care unit (ICU) post-therapeutic dose.

Methods: A retrospective study was carried out in ICU at a tertiary health care center in Lebanon.Electronic health records of patients admitted to ICU who have received vancomycin between 2012 and 2017were reviewed.The sample size was calculated to detect 30%difference in sub-therapeutic level of vancomycin in patients admitted to medical compared to surgical ICU.Our power analysis estimated the sample to be 40 patients per group for ß=0.8 and a=0.05.Trough vancomycin levels below 15mg/l before the 4th dose were reported as sub-therapeutic. Descriptive analyses were conducted using number and percent.The Fisher exact test was used to compare outcomes across the different ICU services.Statistical analyses were performed using Statistical Analysis Software version 9.4.Statistical significance was set at a two-sided p-value of 0.05.

Results: A total of 44 surgical ICU patients with mean age 54.5+/-20.5 years and 41 medical ICU patients with mean age 60.6+/-17.93 years were obtained.Sixteen out of 44 surgical patients (36.3%) as compared to 8 out of 41 medical patients (19.5%) had sub-therapeutic levels (p= 0.068). Eight out of 44 surgical patients (18.1%) as compared to 17 out of 41 medical patients (41.4%) had trough level >20 mg/L (p=0.017). Twenty out of 44 surgical patients (45.4%) and 16 out of 41medical patients (39.02%) had therapeutic levels.

Conclusions: Our study showed that a higher percentage of surgical ICU patients have a sub-therapeutic level. We also noted that a statistically significant higher percentage of medical ICU patients had a supra-therapeutic vancomycin trough levels. Further studies are needed to reach optimal dosage of vancomycin in surgical ICU patient.

P078 Piperacillin-tazobactam (TZP) dosing in septic patients: does obesity weight against efficiency?

P Alexandrzak1, C Lu1, N Van Grunderbeeck2, O Pouly3, F Lambiotte4, C Vinsonneau5, B Hennart6, D Thévenin7

1CH Lens, Pharmacie Médicaments, Lens, France; 2CH Lens, Réanimation Polyvalente / USC, Lens, France; 3CHRU Lille, Réanimation Médicale, Lille, France; 4CH Valenciennes, Réanimation Polyvalente, Valenciennes, France; 5CH Beuvry, Réanimation Polyvalente, Béthune, France; 6CHRU Lille, Laboratoire de Toxicologie, Lille, France; 7CH Lens, Réanimation Polyvalente/USC, Lens, France

Introduction: Obese ICU patients display various disorders likely to impact drugs pharmacokinetic/pharmacodynamic (PK/PD) parameters. These phenomena make it hard to predict antibiotics PK/PD behavior, and thus to choose the appropriate dosing [1, 2]. Our study aims to determine whether the doses of TZP used to treat obese critically ill patients in ICU allow reaching an effective concentration, at the early phase of sepsis (24 hours after initiation).

Methods: A prospective observational study was performed on 12 obese (BMI >30 kg/m²) patients treated with TZP in 5 ICU. TZP maximal and minimal concentrations and concentration at 50% of the time between 2 administrations were determined and compared with targets of 8 and 4 Minimal Inhibitory Concentration (MIC) of Pseudomonas aeruginosa. Data such as treatment modalities, sequential organ failure assessment score, renal clearance and albumin were collected.

Results: Zero patient achieved the target of 8 MIC during 50% of the dosing interval. Four patients achieved the target of 4 MIC. The results are presented in Table 1. Several patients were treated using standard dosing or without loading dose. Continuous infusion was frequent.

Conclusions: TZP doses used in obese ICU patients do not reach targets to treat infections with resistant bacteria. It may exist room for improvement in this setting. Considering the risks of underdosing or toxicity, it seems preferable to use therapeutic drug monitoring to adapt TZP dosing in this population. Higher SOFA was associated with correct exposition, but doses used and renal function were different in these patients.


1) Roberts JA et al. Lancet Infect Dis. 2014 (6):498–509.

2) Taccone FS et al. Crit Care. 2010;14(4):R126.

Table 1 (abstract P078). impact of factors likely to influence PK/PD parameters on target achievement

P079 Pharmacokinetics of meropenem during continuous renal replacement therapy

N Hattori1, N Takahashi1, T Nakada1, Y Niibe2, T Suzuki2, S Yamazaki2, T Suzuki2, I Ishii2, S Oda1

1Chiba university, Graduate school of medicine, Department of Emergency and Critical Care Medicine, Chiba-city, Japan; 2Chiba University Hospital, Department of Pharmacy, Chiba-city, Japan

Introduction: Early administration of effective intravenous antimicrobials is recommended for the management of the patients with sepsis. Although Meropenem (MEPM) is one of the first-line drugs in patients with sepsis because of its broad spectrum, the optimal dose in the critical care settings especially during continuous renal replacement therapy (CRRT) has not been established since therapeutic drug monitoring of MEPM has not been popular.

Methods: Eighteen critically ill patients who received CRRT were enrolled in this study. One gram of MEPM was administered over 1 hour, every 12 hours, and blood samples at 1, 2, 6, 9 and 12 hours after administration were collected on day 1, 2 and 5. All samples were stored at -80°C until analysis. The measurement of the blood concentration of MEPM was performed using high performance liquid chromatography with ultraviolet detection (HPLC-UV).

Results: The patients were 66.2±15.0 years old, 8 were male, their mean body weight was 71.7±20.8 kg, and their mean estimated creatinine clearance was 36.2±20.0 mL/min. CRRT was provided at 22.5±6.0 ml/kg/h. The peak concentration of the first administration of MEPM was 51.0±14.8 μ g/mL. All patients achieved time above minimum inhibitory concentration (TAM)> 50%, if minimum inhibition concentration (MIC) ≤4μ g/mL. Seventy eight % and 60% of the patients achieved TAM>50%, if MIC ≤8μ g/mL and ≤16μ g/mL, respectively.

Conclusions: One gram of MEPM every 12 hours was an appropriate dose for the patients who received CRRT.

P080 Augmented renal clearance in critically ill patients: a single centre cohort study of creatinine clearance in intensive care.

B Johnston1, M Habgood2, D Perry3, M Joshi4, A Krige3

1Royal Liverpool and Broadgreen University Hospital Trust, Critical Care, Liverpool, United Kingdom; 2East Lancashire Hospital Trust, Anaesthesia, Blackburn, United Kingdom; 3East Lancashire Hospital Trust, Critical Care, Blackburn, United Kingdom; 4University of Central Lancashire, Statistics, Preston, United Kingdom

Introduction: Augmented renal clearance (ARC) is defined as a creatinine clearance (CrCl) greater than 130ml/min/1.73m2. In patients with ARC, dose escalation may be appropriate to avoid sub-therapeutic plasma concentration of many drugs. [1]

The East Lancashire Hospital Trust ICU retrospectively analysed admissions for the development of ARC with the aim of characterising those patients who develop ARC.

Methods: We conducted a single centre retrospective cohort study between 2014 - 2016. Daily CrCl was recorded. Patients with renal impairment (serum creatinine levels >110mmol/l) or patients that underwent renal replacement therapy were excluded. The prevalence of ARC was determined, and multiple logistic regression was used to identify risk factors for developing ARC.

Results: ARC was observed in 624/1337 (47%) admissions. ARC was more common in males 56.9% compared to females 35.5% (p<0.001). ARC was significantly associated with younger age (p<0.001). There was no threshold phenomenon with age before which ARC was more likely. ARC was associated with lower APACHE II scores (p<0.001).

Conclusions: ARC is common in patients admitted to ITU. Our study is consistent with previous reports that suggest ARC affects between 35%-65% of admissions. [1] ARC in ITU patients can potentially lead to increased renal clearance of drugs and has been related to sub-therapeutic levels of antimicrobials, anticoagulants and anti-epileptic drugs. Research is needed to define the impact of ARC on drug clearance and dosage in ITU.[2]


[1] Bilbao-Meseguer I, et al. Clin. Pharmacokinet. 57:1107, 2018

[2] Udy AA, et al. Intensive Care Med;39:2070–2082, 2013

P081 Concentration of meropenem in cerebrospinal fluid and serum in patients with meningoencephalitis

A Kuzovlev1, A Shabanov2, A Goloubev1, I Chernenkaya3, S Petrikov3, A Grechko1

1Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, Moscow, Russia; 2Federal Research and Clinical Center of Intensive Care Medicine and Rehabilitology, N.V. Sklifosofsky research center of emergency medicine, Moscow, Russia; 3N.V. Sklifosofsky research institute of emergency medicine, Moscow, Russia

Introduction: Meningitis is one of the complications of severe traumatic brain injury, and it is often associated with encephalitis (incidence from 1.3–4.8% to 10–20%). The aim of the investigation was to study the dynamics of the concentration of meropenem in serum and cerebrospinal fluid (CSF) with intravenous and intrathecal administration of meropenem.

Methods: In eight patients with bacterial meningoencephalitis blood serum and CSF were studied prior to the administration of meropenem and 5-10 min, 1, 2.5 and 5 hrs after it. Antibiotic regimen: 2000 mg of vancomycin (1000 mg BID) and meropenem (2000 mg TID diluted in 100 ml of saline IV + 20 mg BID diluted in 5 ml of saline bolus slowly intrathecally). Meropenem infusion was carried out for 30 minutes, 5 mins after it 5 ml of blood and 1 ml of CSF were sampled. Prior to antibiotics administration blood and CSF were taken for microbiological examination. To determine the concentration of antibiotics iquid chromatography/mass spectrometry was used. The samples were analyzed on an Agilent 1260 Infinity liquid chromatograph coupled to a Sciex QTrap 6500 mass detector (Sciex, US).

Results: Microbial growth was not detected in blood samples. Acinetobacter spp. was isolated in CSF (MIC of meropenem ≤8 mg/ml). Dynamics of meropenem concentrations: time 0 1.9 mg/ml (significantly lower than MIC – 8 mg/ml); directly after the infusion concentration of meropenem increased to 55.3 mg/ml and then gradually decreased reaching 5.4 mg/ml at 6 hrs. Concentration of meropenem in the CSF in all patients and at all time points significantly exceeded the MIC for Acinetobacter spp.

Conclusions: Combined (intravenous+intrathecal) administration of meropenem allows to achieve and maintain a constant concentration of meropenem in CSF, significantly exceeding the MIC of the antibiotic.

P082 Carbapenems for the treatment of nosocomial pneumonia: a systematic review and meta-analysis

M Howatt1, J Muscedere1, A Kalil2, M Klompas3, M Metersky4

1Queen´s University, Department of Critical Care, Kingston, Canada; 2University of Nebraska Medical Center, Department of Medicine, Omaha, United States; 3Harvard Medical School and Brigham & Women´s Hospital, Department of Medicine, Boston, United States;4University of Connecticut, Department of Medicine, Omaha, United States

Introduction: Inadequate empiric therapy for Hospital Acquired Pneumonia (HAP) and Ventilator Associated Pneumonia (VAP) is associated with increased morbidity and mortality. Carbapenems are an attractive choice for empiric therapy but there are concerns about overuse and development of resistance. To better understand the benefits and risks of using carbapenems in empiric treatment regimens for HAP and VAP, we conducted a systematic review and meta-analysis.

Methods: This review was conducted according to PRISMA guidelines and was registered (CRD42018093602). We included randomized controlled trials comparing carbapenem versus non-carbapenem regimens. Data were abstracted in duplicate. The primary outcome was all-cause mortality. Secondary outcomes included mortality by proportion of VAP and overall clinical response. We determined risk ratios (RR) for each outcome using a random-effects model.

Results: Of 14,369 unique references, 20 trials enrolling a total of 5534 patients were included. For the primary outcome of mortality, carbapenem use had a RR of 0.84 (95% CI 0.74 – 0.96, p = 0.01) (Figure 1). When stratified by proportion of VAP (<33%, 33-66%, >66%) RRs for mortality were 0.95 (0.77 – 1.17, p = 0.66), 0.78 (0.57 – 1.07, p = .13), and 0.81 (0.66 – 0.99, p = 0.04), respectively (not shown). Overall clinical response had a RR of 1.04 (0.99 – 1.09, p = 0.12) (Figure 2). Significant statistical heterogeneity was not found.

Conclusions: Carbapenem use for the treatment of HAP and VAP may be associated with a mortality benefit, particularly in populations with a high prevalence of VAP. However, there were no differences in clinical response rates. Limitations of available data include short-term patient follow up, high levels of potential bias, and heterogenous definitions of clinical response. Additional high-quality studies are needed to determine when the inclusion of carbapenems in empiric regimens is warranted.

Fig. 1 (abstract P082).

Patient mortality in carbapenem vs non-carbapenem regimens

Fig. 2 (abstract P082).

Overall clinical response

P083 Impact of piperacillin/ tazobactam versus amoxicillin/ clavulanate as empirical treatment for Klebsiella pneumoniae bacteremia: a propensity score matched analysis

CY Chan, HP Shum, WW Yan, MY Man

Pamela Youde Nethersole Eastern Hospital, ICU, Chai Wan, Hong Kong

Introduction: Klebsiella pneumoniae (KP) bacteraemia is associated with high short-term mortality. As Extended-spectrum β-lactamases (ESBL) producing KP is less common in our locality, either Piperacillin/ Tazobactam (PIP-TAZO) or Amoxicillin/ Clavulanate (AMC) was commonly given as empirical treatment. However, whether PIP-TAZO could achieve a better clinical outcome as compared with AMC is not clear.

Methods: Adults admitted into a regional hospital in Hong Kong with KP bacteraemia from January 2009 to June 2017 were retrospectively reviewed. Demographics, antibiotics, microbiology and outcomes were identified and analyzed. Isolates resistant to either PIP-TAZO or AMC were excluded. Cox regression analysis and propensity score matching methods were used to determine predictors for 30-day mortality.

Results: A total of 322 patients with KP bacteremia were identified. Forty-nine patients (15.2%) required critical care and 56 (17.4%) died within 30 days of hospital admission (Table 1). Among all patients, 82 (25.5%) received PIP-TAZO while 240 (74.5%) received AMC. The unadjusted 30-day mortality was higher in PIP-TAZO group (25.6%) as compared with AMC group (14.6%) but patients in PIP-TAZO group were sicker. Propensity score matching yielded 61 matched pairs (122 patients). Their demographic and disease severity were well balanced (Table 2). No significant difference in 30-day mortality was noted between the matched cohorts (Fig 1). Cox regression analysis identified only respiratory tract and gastrointestinal tract infection as independent predictors for 30-day mortality while the use of PIP-TAZO was insignificant.

Conclusions: With the help of propensity score matching analysis, the apparently higher 30-day mortality among KP bacteremic patients received PIP-TAZO could be explained by their higher disease severity. This implies that the choice of empirical antibiotic, is affected by disease severity, and does not affect 30-day mortality.

Table 1 (abstract P083). Clinical characteristics and outcome parameters of all recruited patients
Table 2 (abstract P083). Clinical characteristics and outcome parameters of matched cohort
Fig. 1 (abstract P083).

Kaplan-Meier Survival Plot showing probability of survival between two patient groups (matched cohort)

P084 Ceftolozane-tazobactam in vitro activity in Pseudomonas aeruginosa and enterobacterales isolates recovered in intensive care units in Portugal. STEP study

M García-castillo1, S García-Fernández1, J Melo Cristino2, C Chaves3, H Ramos4, M Ribeiro5, M Feijó Pinto6, J Diogo7, E Gonçalves8, L Sancho9, R Ferreira10, V Alves11, E Ramalheira12, J Romano13, L Pássaro13, R Cantón1

1Hospital Universitario Ramón y Cajal (IRYCIS), Madrid, Spain, Microbiology Department, Madrid, Spain; 2Centro Hospitalar Lisboa Norte – Hospital de Santa Maria, Lisboa, Portugal, Microbiology Department, Lisboa, Portugal; 3Centro Hospitalar Universitário de Coimbra, Coimbra, Portugal, Microbiology Department, Coimbra, Portugal; 4Centro Hospitalar do Porto, Porto, Portugal, Microbiology Department, Porto, Portugal; 5Centro Hospitalar de São João, Porto, Portugal, Microbiology Department, Porto, Portugal; 6Centro Hospitalar Lisboa Central, Lisboa, Portugal, Microbiology Department, Lisboa, Portugal; 7Hospital Garcia de Orta, Almada, Portugal, Microbiology Department, Almada, Portugal; 8Centro Hospitalar Lisboa Ocidental, Lisboa, Portugal, Microbiology Department, Lisboa, Portugal; 9Hospital Prof. Dr. Fernando da Fonseca, Amadora-Sintra, Portugal, Microbiology Department, Sintra, Portugal; 10Centro Hospitalar Universitário do Algarve, Portimão, Portugal, Microbiology Department, Portimao, Portugal; 11Unidade Local de Saúde de Matosinhos, Matosinhos, Portugal, Microbiology Department, Matosinhos, Portugal; 12Centro Hospitalar Baixo Vouga, Aveiro, Portugal, Microbiology Department, Aveiro, Portugal; 13MSD Portugal, MSD Portugal, Lisboa, Portugal

Introduction: Ceftolozane-tazobactam (C/T) is approved for complicated urinary tract infections (cUTIs), including pyelonephritis, and in combination with metronidazole for complicated intra-abdominal infections (cIAIs). The aim of the STEP study was to assess the in-vitro activity of C/T against clinical isolates prospectively collected from patients with cUTI, cIAIs and lower respiratory tract infections (LRTI) admitted ICUs in Portugal.

Methods: 396 Pseudomonas aeruginosa and 427 Enterobacterales isolates collected in 11 Portuguese hospitals (June 2017-July 2018) from patients with cUTIs, cIAIs and LRTI admitted at ICUs were included. Antimicrobial susceptibility was evaluated by standard ISO-broth microdilution for C/T and 16 other antimicrobials and interpreted using EUCAST guidelines.

Results: Isolates were recovered from cUTI (22.7%), cIAI (20.2%) and LRTI (57.1%). Activity of C/T and comparators by resistance phenotypes and source of infection in P. aeruginosa is shown in Table 1. The agents more active in P. aeruginosa were tobramycin, amikacin and C/T (88.9/88.9/86.4% susceptible, respectively). Activity of C/T in major species of Enterobacterales is shown in Table 2. The agents more active in E. coli were meropenem, colistin and ertapenem (98.3/98.3/98.3% susceptible), while in Klebsiella spp. were colistin, amikacin and meropenem (91.7/91.6/84.9% susceptible). Activity of C/T against ESBL-E. coli and -Klebsiella spp. was 76.3/21.3% susceptible. As expected, C/T had scarce activity against carbapenemase-phenotype.

Conclusions: C/T exhibited good overall activity against E. coli, although it might be affected in other Enterobacterales by local epidemiology, particularly due to carbapenemase producers. It was one of the most potent agents against P. aeruginosa, in which the activity was maintained regardless of the resistant phenotype. These results reinforce that C/T represents a therapeutic option in ICU patients with cUTIs, cIAIs and LRTI.

Table 1 (abstract P084). Activity of ceftolozane-tazobactam and comparators by resistance phenotypes in P. aeruginosa
Table 2 (abstract P084). Antimicromial activity of ceftolozane-tazobactam broken down by species and source of infection

P085 A high C-reactive protein/procalcitonin ratio predicts Mycoplasma pneumoniae infection

O Neeser1, T Vukajlovic1, L Felder1, S Haubitz1, A Hammerer-Lercher2, C Ottiger2, B Müller1, P Schütz1, C Fux1

1Kantonsspital Aarau, Internal Medicine, Aarau, Switzerland; 2Kantonsspital Aarau, Laboratory Medicine, Aarau, Switzerland

Introduction: Differentiating Mycoplasma pneumoniae (MP) from Streptococcus pneumoniae (SP) and viral etiologies of community-acquired pneumonia (CAP) has important implications regarding empiric antibiotic therapy. We investigated parameters upon hospital admission to predict MP infection.

Methods: All patients hospitalized in a tertiary care hospital 2013 – 2017 for CAP with confirmed etiology were analyzed using logistic regression analyses and area under the receiver operator characteristics curves for associations between demographic, clinical and laboratory features and the causative pathogen.

Results: We analyzed 568 patients with CAP, including 47 (8%) MP; 152 (27%) with SP and 369 (65%) with influenza or other viruses. Comparing MP and SP, younger age (OR 0.56 per 10 years, 95% CI 0.42-0.73), lower neutrophil/lymphocyte ratio (OR 0.9, 0.82-0.99) and elevated CRP/PCT ratio (OR 15.04 (5.23-43.26) for a 400mg/g cut-off) independently predicted MP. With a ROC AUC of 0.91 (0.80 for the >400mg/g cutoff), the CRP/PCT ratio was the strongest predictor of MP versus SP. The CRP/PCT ratio also provided good discrimination between MP and viral infections (AUC 0.83; OR 5.55 for the >400mg/g cutoff, 2.26-13.64).

Conclusions: In patients hospitalized with CAP, a high admission CRP/PCT ratio predicts Mycoplasma pneumoniae infection and may improve empiric management.

P086 Probiotics for the prevention of VAP and ICU-acquired infections in multi-trauma patients: a preliminary study

G Tsaousi, G Stavrou, Z Aidoni, K Fotiadis, K Kotzampassi

Department of Surgery, Aristotle University of Thessaloniki, Thessaloniki, Greece

Introduction: The prophylactic use of probiotics has emerged as a promising alternative to current strategies viewing to control nosocomial infections in a critically-ill setting. However, their beneficial role in VAP prevention remains inconclusive. Our aim was to delineate the efficacy of probiotics for both VAP prophylaxis and restriction of ICU-acquired infections in multi-trauma patients.

Methods: Randomized, placebo-controlled study enrolling 58 multi-trauma patients, requiring mechanical ventilation for >10 days. Participants were randomly assigned to receive either probiotic (n=28) or placebo (n=30) treatment. A four-probiotic formula was applied and each patient received two capsules per day from Day1 to Day15 post ICU admission. The content of one capsule was given as an aqueous suspension by nasogastric tube, while the other one was spread to the oropharynx after being mixed up with water-based lubricant. The follow-up period was 30 days, while ICU stay and mortality were also assessed.

Results: The use of probiotics reduced notably the incidence of VAP [32.1% vs 53.3%; p=0.001; RR placebo=3 (95%CI 1.6-5.4)], central venous line infection [21.4% vs 40%; p=0.031; RR placebo=4 (95%CI 2-7.9)], trauma-related infection [10.7% vs 26.6%; p=0.014; RR placebo=5.4 (95%CI 2.4-11.9)] or sepsis [14.3% vs 30%; p=0.005; RR placebo=5.2 (95%CI 2.3-11.4)] compared to placebo. Furthermore, fewer patients in probiotics group presented an ICU stay > 30 days [7.1% vs 40%; p=0.002; RR placebo=2.8 (95%CI 1.7-4.1)], while no difference in 30-day mortality rate was identified between groups (10.7% probiotics vs 6.7% placebo).

Conclusions: The prophylactic administration of probiotics exerted a positive effect on the incidence of VAP or other ICU- acquired infections and ICU stay in a critically-ill subpopulation being notorious for its high susceptibility to infections, namely multi-trauma patients.

P087 Use of a C-reactive protein-based protocol to guide the duration of antibiotic therapy in critically ill patients: a randomized controlled trial

I Borges1, M Santana1, A Lana1, L Martins2, E Colosimo2, C Oliveira3, S Saturnino3, M Andrade3, C Ravetti1, V Nobre1

1Faculdade de Medicina da Universidade Federal de Minas Gerais, Núcleo Interdisciplinar de Investigação em Medicina Intensiva - NIIMI, Belo Horizonte, Brazil; 2Faculdade de Medicina da Universidade Federal de Minas Gerais, Departamento de Estatística, Belo Horizonte, Brazil; 3Faculdade de Medicina da Universidade Federal de Minas Gerais, Departamento de Clínica Médica, Belo Horizonte, Brazil

Introduction: The rational use of antibiotics is one of the main strategies to limit the development of bacterial resistance. In this study we aimed to evaluate the effectiveness of a C reactive protein (CRP) based protocol in reducing antibiotic treatment time in critically ill patients.

Methods: An open randomized clinical trial was conducted in two adult intensive care units of a university hospital in Brazil ( NCT02987790). Patients were randomly allocated to: i) intervention - duration of antibiotic therapy guided by CRP levels, and ii) control - duration of therapy based on best evidences for rational use of antibiotics. In the CRP group, antibiotic suspension was recommended after five full days of antibiotic therapy when CRP levels had fell > 50% of peak value (if peak > 100mg/L) or after three full days of antibiotic therapy when absolute values had reached values < 35mg/L (if peak <100mg/L). The primary outcome was days of antibiotic therapy in the index infection episode.

Results: 130 patients were included: 64 in the CRP group and 66 in the control group. The median (Q1-Q3) age was 61 (51-68) years, with SAPS 3 of 59 (50-70), without difference between groups. In the intention to treat analysis, the median (Q1-Q3) duration of antibiotic therapy for the index infection episode was 7.0 (5.0-8.8) days in the CRP group and 7.0 (7.0-11.3) days in the control group (p=0.011). In the cumulative suspension curve of antibiotics, a significant difference in the exposure time between the two groups was identified, with less exposure in the CRP group (p=0.007). In the pre-specified per protocol analysis, with 59 patients allocated in each group, the median duration of antibiotics was 6.0 (5.0-8.0) days in the CRP group and 7.0 (7.0- 10.0) days in the control group (p=0.011). Mortality and relapse rates were similar between groups.

Conclusions: Daily levels of CRP may aid in reducing the time of antibiotic therapy in critically ill patients, even in a scenario of judicious use of these drugs.

P088 Hemophagocytic lymphohistiocytosis in the University Hospital of Infectious Diseases Cluj Napoca

L Herbel1, D Miclaus1, A Dicea1, A Muntean1, L Urian2, L Damian3, D Dima2, M Lupse4

1University Hospital of Infectious Deseases, ICU, Cluj Napoca, Romania; 2Oncology Institute, Hematology, Cluj Napoca, Romania; 3University of Medecine and Pharmacy, Rhumatology, Cluj Napoca, Romania; 4University of Medecine and Pharmacy, ICU, Cluj Napoca, Romania

Introduction: The Macrophage Activation Syndrome (MAS) or hemophagocytic lymphohistiocytosis(HLH) is a life threatening complication characterized by pancytopenia, liver failure, coagulopathy and neurologic symptoms and is thought to be caused by the activation and uncontrolled proliferation of T lymphocytes and well differentiated macrophages, leading to widespread hemophagocytosis and cytokine overproduction [1,2].The etiology is unknown, but is considered to have an infectious trigger.The aim of our study is to evaluate the impact of HLH in our 9 beds infectious diseases ICU, during 83 months period (2012-2018).

Methods: A retrospective study based on electronic databases, including all patients admitted in our ICU, that have matched at least 5 out of 8 criteria for HLH diagnosis (1):fever; hepato-splenomegaly; >2 cytopenia (Hb <9 g/dl, PLT</dl, PMN -N<1000/mmc); hypertriglyceridemia>265mg/dl, fibrinogen<150mg/dl; hemophagocytosis- bone marrow, spleen, and/or lymphnodes; NK activity reduced/absent; ferritin level>500UI/L; CD 25>2400. We have evaluated the etiology established with cultures, serology, and molecular methods, treatment with corticosteroids, IV immunoglobuline, cyclosporine, etoposide and outcome (2).

Results: 2112 patients were admitted to ICU, 15 patients(0.71%) met the criteria for HLH. The average length of stay in ICU was 15 days; 6 patients died (40%) without relation with the followed treatment.

Conclusions: HLH is not a rare condition in infectious diseases ICU. The etiology is more frequent established compared with literature data. Treatment (corticosteroids, immunoglobuline, cyclosporine, etoposide) is not associated with increased survival


(1) Henter JL et al: Pediatr Blood Cancer 2007;48(2)124-31

(2) Stuart J.Carter et al: Rheumatology review, Oxford Univ Press 22 feb 2018

P089 Forecasting hemorrhagic shock using patterns of physiologic response to routine pre-operative blood draws

X Li1, M Pinsky2, G Clermont2, A Dubrawski1

1Carnegie Mellon University, Auton Lab, Pittsburgh, United States; 2University of Pittsburgh, School of Medicine, Pittsburgh, United States

Introduction: Irreversible hemorrhagic shock (IHS), a critical condition associated with significant blood loss and poor response to fluid resuscitation, can induce multiple organ failures and rapid death [1]. Determining the patients who are likely to develop IHS in surgeries could greatly help pre-operative assessment of patient outcomes and allocation of clinical resources.

Methods: machine learning model of IHS is developed and validated via porcine induced bleed experiment. 36 healthy sedated Yorkshire pigs first had one 20mL rapid blood draw during a stable period, and then were bled at 20mL/min to mean arterial pressure (MAP) of 30 mmHg. 10 subjects had IHS defined as MAP<20mmHg. Arterial, central venous and airway pressures collected at 250 Hz during the blood draw [Fig 1] were used to extract characteristic sequential patterns using Graphs of Temporal Constraints (GTC) methodology [2], and a decision forest (DF) model was trained on these patterns to determine subjects at high risk of impending IHS.

Results: In a leave-one-subject-out cross-validation, our method confidently identifies 30% (95% CI [15.6%, 44.4%]) of the subjects who are likely to experience IHS when subject to substantial bleeding, while only giving on average 1 false alarm in 10,000 such predictions. This method outperforms logistic regression and random forest models trained on statistically featurized data [Tab 1, Fig 2].

Conclusions: Our results suggest that by leveraging sequential patterns in hemodynamic waveform data observed in pre-operative blood draws, it is possible to predict who are prone to develop IHS resulting from blood loss in the course of surgery. Future work includes validating the proposed method on data collected from human subjects, and developing a clinically useful screening tool with our investigations.

Work partially funded by NIH GM117622.


[1] Gutierrez et al. Critical care. Crit Care, 8(5):373, 2004.

[2] Guillame-Bert et al. JMLR, 18(121):1–34, 2017.

Table 1 (abstract P089). Comparison of model performances with 95% confidence intervals in square brackets
Fig. 1 (abstract P089).

An example 5-second segment of three hemodynamic variables used in our models (Arterial Pressure (ART), Central Venous Pressure (CVP), Airway Pressure (AIR)) extracted from one subject

Fig. 2 (abstract P089).

ROC curves of logistic regression (orange), random forest (green) and GTC-DF (blue). False Positive Rates (FPR) and False Negative Rates (FNR) in (b) and (c) are logarithmically scaled to emphasize performance at clinically relevant low FPR and FNR settings, respectively. FPR and TPR corresponding to 50% sensitivity threshold for each model is marked with a red dot

P090 Oxytocin and hydrogen sulfide expression in porcine brain after hemorrhagic shock

N Denoix1, T Merz2, M Wepler2, H Guendel1, C Waller3, P Radermacher2, O McCook2

1Ulm University Medical Center, Clinic for Psychosomatic Medicine and Psychotherapy, Ulm, Germany; 2Ulm University Medical Center, Institute for Anesthesiological Pathophysiology and Process Engineering, Ulm, Germany; 3Universitätsklinik der Paracelsus Medizinischen Privatuniversität, Clinic for Psychosomatic Medicine and Psychotherapy, Nuernberg, Germany

Introduction: The H2S and oxytocin(Oxy) systems are reported to interact with one another [1]. H2S plays a major role in the hypothalamic control of Oxy release during hemorrhage [2]. There is scarce information about Oxy receptor(OxyR) expression in the brain in general and what is there is ambivalent. OxyR has been immunohistochemically(IHC) detected in the human hypothalamus but not in the hippocampus, in contrast to rodents [3], which underscores the need for additional characterization in relevant animal models. Thus the aim of this study is to map the expression of the Oxy and H2S systems in the porcine brain in a clinically relevant model of hemorrhagic shock (HS).

Methods: Anesthesized atherosclerotic pigs (n=9) underwent 3h of HS (MAP 40+/-5mmHg) [4], followed by 72h resuscitation. IHC detection of Oxy, OxyR, the H2S producing enzymes cystathionine-γ -lyase (CSE) and cystathionine-β -synthase(CBS) was performed on formalin fixed brain paraffin sections.

Results: Oxy, OxyR, CSE and CBS were localized in the porcine brain. Proteins were differentially expressed in the hypothalamus (Fig 2), parietal cortex and cerebellum (Fig 1). Cell types positively identified were: magnocellular neurons of the hypothalamus, cerebellar Purkinje cells and granular neurons, and hippocampal pyramidal and granular neurons of the dentate fascia. Arteries and microvasculature were also positive for OxyR and CSE.

Conclusions: Our results confirm the presence of Oxy and OxyR in the hypothalamus similarly to the human brain. Novel findings were: OxyR in the cerebellum and CSE expression in the hypothalamus and cerebellum. The coexpression of OxyR and CSE may link and help better understand neurochemical systems and physiological coping in hemorrhagic shock. Funding: CRC1149


[1] Merz T et al. Intensive Care Med Exp 6:41, 2018

[2] Ciosek J et al. J Physiol Pharmacol 54, 2, 233-246, 2003

[3] Boccia ML et al. Neuroscience 253, 155–164,2013

[4] Hartmann C et al. Crit Care Med 45(12):e1270-e1279, 2017

Fig. 1 (abstract P090).

See text for description

Fig. 2 (abstract P090).

See text for description

P091 Impact of mean arterial pressure and norepinephrine administration during the first 48 hours in septic shock patients

F Torrini, A Dell´Anna, S Muscolino, T Taccheri, S Carelli, R Maviglia, M Antonelli

Fondazione Policlinico Universitario A. Gemelli-IRCCS, Dept. of Anesthesia and Intensive Care Medicine, Rome, Italy

Introduction: Septic shock is one of the main causes of intensive care unit (ICU) admission, leading to mortality up to 50% of patients. Acute kidney injury (AKI) frequently occurs and is associated to great morbidity and mortality. Hemodynamic optimization may reduce the incidence of AKI, but the use of vasopressors to increase mean arterial pressure (MAP) could have deleterious effect on renal perfusion. We aimed at investigating the effect of MAP and norepinephrine (NE) on the incidence of AKI in septic shock patients

Methods: Retrospective study based on prospectively collected data on digital medical records (Digistat) at our ICU. Adult patients with a diagnosis of septic shock surviving at least 48 hours between 01/2010 and 01/2015 were included. We collected data on hemodynamic parameters, doses of vasoactive/inotropic drugs hourly in the first 48h and clinical scores. Hemodynamic data were analyzed individually as well as in their average, maximum and minimum value and the AUCs were calculated.Patients were categorized on the basis of NE dose (NE<0.5,0.5<NE<1,NE>1 mcg/kg/min). AKI was defined according to KDIGO guidelines.

Results: 220 patients were included. Among these, 131 (59.5%) died at day 60 and 134 (60.9%) developed AKI. Average MAP was lower in patients who developed AKI (79 vs 83 mmHg, p= .005), even after adjustment for the average dose of NE (mean difference 2.3 mmHg [95% CI 0.3-4.4]; p= .026). Conversely, different NE doses were not associated to a significant difference in the occurrence of AKI, while MAP was significantly lower at higher doses of NE (83 vs 82 vs 77 mmHg, p< .001). The average MAP had a low predictive value on the occurrence of AKI, but was significantly related to an increased risk of AKI even after adjustment for other variables (OR 0.96, 95% CI 0.93-0.99)

Conclusions: Average MAP in the first 48 hours in septic shock patients is significantly related to the occurrence of AKI, regardless of the average dose of norepinephrine administrated

P092 Achievement of an international consensus target mean arterial pressure following initiation of angiotensin II in patients with catecholamine-resistant distributive shock

L Forni1, MT McCurdy2, AK Khanna3, LW Busse4, L Chawla5, J Hästbacka6

1Royal Surrey County Hospital, Guildford, United Kingdom; 2University of Maryland School of Medicine, Baltimore, United States; 3Cleveland Clinic Foundation, Cleveland, United States; 4Emory University, Atlanta, United States; 5La Jolla Pharmaceutical Company, San Diego, United States; 6University of Helsinki and Helsinki University Hospital, Helsinki, Finland

Introduction: In patients with distributive shock, increasing mean arterial pressure (MAP) to a target of >65 mmHg can improve tissue perfusion. Patients unable to achieve the target MAP of >65 mmHg despite adequate fluid resuscitation as well as catecholamines and vasopressin standard care (SC), may benefit from the non-catecholamine vasopressor angiotensin II to increase MAP. This post-hoc analysis examined whether patients from the ATHOS-3 study with a baseline (BL) MAP <65 mmHg and treated with SC plus either angiotensin II (Ang II) or placebo achieved a MAP of >65 mmHg for 3 consecutive hours, without increasing the dose of SC therapy.

Methods: Patients were assigned in a 1:1 ratio to receive Ang II or placebo, plus SC. Randomization was stratified according to MAP (<65 or >65 mmHg) at screening. In patients with BL MAP <65 mmHg, we evaluated whether patients achieved a MAP of >65 mmHg for the first 3 hours after initiation (MAP measurements taken at hours 1, 2, and 3), without an increase in the dose of SC.

Results: Among 321 treated patients, 102 had BL MAP <65 mmHg (Ang II, 52; placebo, 50). Median BL MAP (IQR) was 61 (57-63) and 62 (59-64) mmHg for placebo and Ang II groups, respectively. Patients with BL MAP <65 mmHg who were treated with Ang II were more likely to achieve MAP ≥65 mmHg for 3 consecutive hours after initiation without an increase in SC dose (67%, 95%CI 53-80), compared with placebo-treated patients (24%, 95%CI 13-38, OR=6.52, p<0.0001).

Conclusions: In this post-hoc analysis of patients with BL MAP <65 mmHg, patients receiving Ang II plus SC were significantly more likely to achieve a MAP >65 mmHg for the first 3 consecutive hours after initiation than patients receiving SC only. This suggests that administering Ang II may help patients with catecholamine-resistant distributive shock to achieve the consensus standard target MAP.

P093 Norepinephrine synergistically increases the efficacy of volume expansion on venous return in septic shock

I Adda, C Lai, JL Teboul, L Guerin, F Gavelli, C Richard, X Monnet

Hôpitaux universitaires Paris-Sud, Hôpital de Bicêtre, APHP, Service de médecine intensive-réanimation, Le Kremlin-Bicêtre, France

Introduction: Through reduction in venous capacitance, norepinephrine (NE) increases the mean systemic pressure (Psm) and increases cardiac preload. This effect may be added to the ones of fluids when both are administered in septic shock. Nevertheless, it could be imagined that NE potentiates in a synergetic way the efficacy of volume expansion on venous return by reducing venous capacitance, reducing the distribution volume of fluids and enhancing the induced increase in stressed blood volume. The purpose of this study was to test if the increase in Psm induced by a preload challenge were enhanced by NE.

Methods: This prospective study had included 30 septic shock adults. To reversibly reproduce a volume expansion and preload increase at different doses of NE, we mimicked fluid infusion through a passive leg raising (PLR). In patients in which the decrease of NE was planned, we estimated Psm (using respiratory occlusions) at baseline and during a PLR test (PLRHigh). The dose of NE was then decreased and Psm was estimated again before and during a second PLR (PLRLow).

Results: NE dose decreased from 0.32[0.18-0.62] to 0.26[0.13-0.50] μ g/kg/min (p<0.001). The increase in Psm induced by PLRHigh (Δ Psm) at the highest dose of NE was significantly greater than the Δ Psm induced by PLRLow (35[20-43]% vs 14[8.8-20]%, p<0.001). The increase in cardiac index induced by PLRLow was significantly greater than that induced by PLRHigh (p<0.001). Δ PsmHigh - Δ PsmLow was moderately correlated with the diastolic arterial pressure at BaselineHigh (p=0.03, r=0.41) and with the NE-induced change in mean arterial pressure (p=0.004, r=0.57).

Conclusions: NE enhances the increase in Psm induced by a PLR, which mimics a fluid infusion. This suggests that it may potentiate the effects of fluid in a synergetic way in septic shock patients. This may decrease the amount of administered fluids and contribute to decrease the cumulative fluid balance.

P094 Predictors of the response to vasopressin in ICU patients with norepinephrine-resistant hypotension: a single-center, retrospective study

Y Yasuda, D Kasugai, M Nishikimi, M Higashi, T Yamamoto, M Ozaki, A Numaguchi, N Matsuda

Nagoya University Graduate School of Medicine, Department of Emergency and Critical Care, Nagoya, Aichi, Japan

Introduction: Arginine vasopressin (AVP) can be used in addition to norepinephrine (NE) for NE-resistant septic shock. However, a subgroup who will response to AVP is unknown. The purpose of this study was to determine factors which could predict the response to AVP in patients with NE-resistant hypotension.

Methods: This was a single-center, retrospective analysis of patients who administered AVP for NE-resistant hypotension in our intensive care units (ICUs). Eligible patients were adult patients who administered AVP in addition to NE due to hypotension (mean arterial pressure (MAP) < 65) in our ICUs between August 2014 and December 2017. We divided all patients into two groups by response to AVP; responders and non-responders. The responders were defined as an increase of MAP ≥ 10 mmHg at 1h after AVP initiation. We conducted univariate and multivariate logistic regression analysis to evaluate the effect of variables on AVP response.

Results: A total of 163 patients were included; 107 responders (66%), 56 non-responders (34%). There was no significant difference for MAP at the time of AVP initiation (51 vs 52 mmHg; p = 0.40), initiation dose of AVP (0.028 vs 0.031 U/min; p = 0.67), and dose of NE at the time of AVP initiation (0.18 vs 0.18 μ g/kg/min; p = 0.95). MAP at 1h after AVP initiation was significantly higher in responders than non-responders (74 vs 55 mmHg; p < 0.01). Responders were older (69 vs 66; p = 0.02) and had lower heart rate (HR) (99 vs. 108; p = 0.01) and lactate (3.2 vs. 4.8 mmol/L; p = 0.02) at the time of AVP initiation. The multivariate logistic analysis revealed that HR ≤ 103 (OR 2.56, 95% CI 1.29-5.10, p < 0.01), lactate ≤ 3 (OR 3.11, 95% CI 1.56-6.19, p < 0.01) and age ≥ 63 (OR 2.16, 95% CI 1.05-4.45, p = 0.04) were significantly associated with the response to AVP.

Conclusions: HR, lactate levels and age before AVP initiation can predict the response to AVP in ICU patients with NE-resistant hypotension.

P095 The maximum norepinephrine dosage of initial 24 hours predicts early death in septic shock

D Kasugai1, A Hirakawa2, N Jinguji3, K Uenishi3

1Nagoya university Gtaduate School of Medicine, Department of Emergency and Critical Care, Nagoya, Aichi, Japan; 2Fujita Health University, Department of Disaster and Traumatology, Fujita Health University, Toyoake, Japan; 3Fujita Health University Hospital, Department of Emergency and General Internal Medicine, Fujita Health University Hospital, Toyoake, Japan

Introduction: The mortality of septic shock refractory to norepinephrine remains high. To improve the management of this subgroup, the knowledge of early indicator is needed. We hypothesize that maximum norepinephrine dosage on the initial day of treatment is useful to predict early death in septic shock.

Methods: In this retrospective single-center observational study, septic shock patients admitted to the emergency intensive care unit (ICU) of an academic medical center between April 2011 and March 2017 were included. Cardiac arrest before ICU admission and those with do-not-resuscitate orders before admission were excluded. The maximum dosage of norepinephrine initial 24 hours of ICU admission (MD24) was used to assess 7-day mortality.

Results: One-hundred-fifty-two patients were included in this study. Median SOFA score was 11 (9-13), and median MD 24 was 0.24 (0.16-0.38) mcg/kg/min. Vasopressin and steroid were administered in 58 (38 %) and 67 (44 %) cases. Nineteen patients (13 %) died within a week. Non-survivors had higher MD24, higher SOFA score, and higher rate of vasopressin use. The higher MD24 predicted 7-day mortality (area under curve 0.797, threshold 0.60 mcg/kg/min, sensitivity 50 %, specificity 91%). After adjustment of inverse probability of treatment weighing method using propensity scoring, MD24 higher than 0.6 mcg/kg/min was independently associated with 7-day mortality (OR: 9.96, 95 %CI: 2.56-38.8, p < 0.001).

Conclusions: The maximum dosage of norepinephrine higher than 0.6 mcg/kg/min initial 24 hours was significantly associated with 7-day mortality in septic shock, and may be useful in the selection of higher severity subgroup.

P096 The impact of norepinephrine on right ventricular function and pulmonary haemodynamics in patients with septic shock - a strain echocardiography study

K Dalla

Sahlgrenska University Hospital Mölndal, Göteborg, Sweden

Introduction: Septic shock is characterized by myocardial depression and severe vasoplegia. Right ventricle performance could be impaired in sepsis. The effects of norepinephrine on RV performance and afterload in septic shock are not immediately evident. The aim of the present study was to investigate the effects of norepinephrine on RV systolic function, RV afterload and pulmonary haemodynamics.

Methods: Eleven, volume-resuscitated and mechanically ventilated patients with norepinephrine-dependent septic shock were included. Infusion of norepinephrine was randomly and sequentially titrated to target mean arterial pressures (MAP) of 60, 75 and 90 mmHg. At each target MAP, strain- and conventional echocardiographic were performed. The pulmonary haemodynamic variables were measured by using a pulmonary artery thermodilution catheter. The RV afterload was assessed by calculating the effective pulmonary arterial elastance (Epa) and pulmonary vascular resistance index (PVRI).

Results: The norepinephrine-induced elevation of MAP increased central venous pressure (38%, p<0.001), stroke volume index (7%, p<001), mean pulmonary artery pressure (19%, p<0.001) and RV stroke work (20%, p=0.045), while neither pulmonary vascular resistance index nor Epa was affected. Increasing doses of norepinephrine improved RV free wall strain from -19% to -25% (32%, p=0.003), tricuspid annular plane systolic excursion (22%, p=0.010) and tricuspid annular systolic velocity (17%, p=0.029). There was a trend for an increase in cardiac index assessed by both thermodilution (p=0.079) and echocardiography (p=0.054).

Conclusions: The RV function was improved by increasing doses of norepinephrine, as assessed both by strain- and conventional echocardiography. This is explained by an increase of RV preload. Pulmonary vascular resistance is not affected by increased doses of norepinephrine.

P097 Peripheral perfusion versus lactate-targeted fluid resuscitation in septic shock: the ANDROMEDA SHOCK physiology study. preliminary report

G Hernandez1, R Castro1, L Alegría1, S Bravo1, D Soto1, E Valenzuela1, M Vera1, V Oviedo1, C Santis2, G Ferri2, M Cid2, B Astudillo2, P Riquelme2, R Pairumani2, G Ospina-Tascón3, J Bakker1

1Pontificia Universidad Católica de Chile, Departamento de Medicina Intensiva, Santiago, Chile; 2Hospital Barros Luco-Trudeau, Unidad de Pacientes Críticos, Santiago, Chile; 3Fundación Valle del Lili, Universidad ICESI, Department of Intensive Care Medicine, Cali, Colombia

Introduction: The potential role of peripheral perfusion as a target for fluid resuscitation (FR) in septic shock is unknown. We aimed to determine if capillary refill time-targeted FR (CRT) is superior to a lactate-targeted fluid resuscitation (L) regarding fluid balances and evolution of tissue-perfusion variables.

Methods: Clinical RCT in 2 ICUs (sample size of 46 pts). Fluid responsive septic shock patients were randomized into 2 groups with FR aimed at normalizing CRT or at normalizing or decreasing lactate at rates >20%/ 2h during a 6h period. Hemodynamic and perfusion variables were assessed at 2, 6 and 24h, as well as SOFA scores and fluid balances up to 72h. Tissue perfusion variables were assessed in parallel (NIRS, sublingual microcirculation, LiMON, lactate/pyruvate and central venous-arterial pCO2 to arteriovenous O2 content difference ratio)

Results: Eighteen patients (mean age 66 [42,74], APACHE II 23 [14,30], SOFA 10 [7,12] have been included until now (8 in L group and 10 in CRT group). Baseline lactate was 3.6 [3.1,5.0] vs 3.8 [3.0, 4.7] mmol/l; and CRT was 4.5 [2.5, 6.5] vs 4.5 [3.0, 6.0] secs, in L and CRT groups, respectively. No differences between groups were observed in macrohemodynamics at baseline. Resuscitation fluids at 6h were not different 1.00 [0.50,4.25] vs 1.13 [0.25,2.00] liters in L and CRT groups, respectively, with a trend to less positive 24h fluid balance in CRT group 3.25 [0.55,4.65] vs 2.25 [0.81,4.58] liters. StO2 and ICG/PDR were not different. The results of some perfusion related parameters are shown in Table 1.

Conclusions: This preliminary results suggest that using CRT as a target for FR in septic shock appears to be feasible, and not associated with impairment of tissue perfusion-related parameters as compared to lactate-targeted FR.

Grant FONDECYT Chile 1170043

Table 1 (abstract P097). Hemodynamics and Tissue perfusion variables

P098 Efficacy comparison of angiotensin II and vasopressin in a norepinephrine-resistant hypotension pig model

S Li1, J Wilkie1, H Jin1, D Merrill2, M Frierson1, L Johnson1, D Zhou1, R Marsden1, S Chen1, A Seacat1, L Chawla1, G Tidmarsh1, J Rolke1

1La Jolla Pharmaceutical Company, San Diego, United States; 2Da Vinci Biomedical Research Products, Inc., South Lancaster, United States

Introduction: Shock patients often become resistant to catecholamines which often require the addition of a non-catecholamine vasopressor. Preclinical studies suggest that in the presence of a-adrenoceptor antagonism, the renin-angiotensin aldosterone system exerts the major vasopressor influence. We sought to determine the effects of AngII or Lypressin (LYP [porcine vasopressin]) on blood pressure in a norepinephrine (NE)-resistant hypotension pig model.

Methods: Phentolamine (PHN), a reversible α-blocker that antagonizes the vasoconstriction by NE, was continuously infused to induce hypotension. After NE-resistant hypotension was established, LYP or AngII was then co-infused with PHN. Mean arterial pressure (MAP) and heart rate were continuously recorded (Fig. 1).

Results: As shown in Fig. 2, PHN reduced MAP to 75.3±9.2% from the baseline (100±13.8%) and blunted the MAP increase induced by 0.2μg/kg/min NE from 193±14.4% to 130.2±15.4%. Additional 0.3μg/kg/min NE didn’t further increase MAP (130±17.6%), indicating a steady state of NE resistance. AngII at 80 (1x highest label dose), 160 (2x) and 320 (4x) ng/kg/min all significantly increased the MAP over baseline (143.6±18.3, 158.3±13.5 and 168.5±14.4%, respectively) in a dose-dependent fashion. Prior to PHN infusion, a 2.1-unit bolus of LYP was administered which confirmed an increase in MAP. After PHN infusion, LYP was administered at 0.07 (1x highest label dose), 0.21 (3x), 0.7 (10x) and 2.1 (30x) U/min resulting in no increase over baseline MAP (76.9±14.7, 85.4±13.9, 98.8±14.1 and 93±9.7%, respectively).

Conclusions: In a background of α-adrenoceptor blockade, at clinically comparable doses, the vasopressor effect of Ang II was maintained while those of NE and LYP were attenuated. These data suggest that the blood pressure effect of vasopressin-like peptides may require a functioning α-adrenoceptor. Patients with shock who are resistant to increasing doses of catecholamines may also have vasopressin resistance potentially making angiotensin II a preferred vasopressor for these patients.

Fig. 1 (abstract P098).

Study design and representative individual data

Fig. 2 (abstract P098).

Average MAP of the last minute during each treatment

P099 Use of a low-dose heparin anticoagulation protocol for Impella CP® devices in cardiogenic shock

A Newsome, AT Taylor, SG Garner

The University of Georgia College of Pharmacy, Pharmacy, North Augusta, United States

Introduction: Impella CP® is a percutaneous left ventricular assist device (LVAD) that requires a heparin-dextrose purge solution to decrease risk of device thrombosis. The manufacturer recommends heparin 50 units per mL, but supra-therapeutic anticoagulation has been observed with this concentration. The purpose of this evaluation was to observe the efficacy and safety of a low-dose heparin-based purge solution (25 units per mL in dextrose 20%) as part of an Impella CP® protocol.

Methods: A single site retrospective review was conducted for all adults with the Impella CP® from January 2015 to December 2017.The primary outcome evaluated percentage of activated clotting times (ACT) below therapeutic range, and safety was evaluated by bleeding events, as defined by the Bleeding Academic Research Consortium (BARC) bleeding criteria. Secondary objectives included evaluating the incidence of device thrombosis and rate of heparin-induced thrombocytopenia (HIT). Platelet trends following Impella CP® implantation were characterized.

Results: A total of 18 patients were included. The percentage of activated clotting times (ACT) readings within goal of 160 to 200 seconds was 49%, and 38% of readings were sub-therapeutic. No device thrombosis was observed. Per BARC bleeding criteria, 22% (n=4) patients experienced class 3a bleeding and 39% (n=7) experienced class 2 bleeding. Though four (22%) of patients were tested for HIT, no patients were positive. Patients showed universal reductions in platelet counts.

Conclusions: The use of a lower dose heparin concentration purge solution was not associated with increased device thrombosis although bleeding still occurred. The use of a low-dose anticoagulation protocol of heparin 25 units per mL in dextrose 20% with supplemental systemic anticoagulation as needed may be a safe and effective alternative and warrants further evaluation.

P100 Use of resuscitative endovascular balloon occlusion of the aorta for massive gastrointestinal bleeding in ICU

Y Kato, A Kuriyama, M Onodera

Kurashiki Central Hospital, Department of Emergency Medicine, Kurashiki Okayama, Japan

Introduction: Resuscitative Endovascular Balloon Occlusion of the Aorta (REBOA) has been increasingly used for the management of both traumatic and non-traumatic hemorrhagic shock. However, there is limited evidence for its use in gastrointestinal bleeding (GIB), especially in the ICU setting. We successfully treated a patient with massive GIB using REBOA in the ICU. We will discuss the difficulty performing the procedure and its countermeasure.

Methods: A case report.

Results: An 83-year-old woman was transferred to our hospital with shock. Coffee grounds material was found in a nasogastric aspirate after intubation and upper gastrointestinal endoscopy identified a pulsating large duodenum ulcer without active bleeding, for which an elective procedure was planned. She was admitted to our ICU, responded to initial resuscitation, and thereafter extubated. Her systolic blood pressure (SBP) suddenly dropped to 40mmHg with massive hematochezia at that night, and did not increase despite resuscitation with blood products, crystalloid and norepinephrine. To buy time until measures for stop bleeding, we planned to place REBOA in the ICU. Following the placement of a sheath in the left femoral artery, we tried to place a 7 Fr intra-aortic balloon occlusion catheter, which unintentionally and repeatedly went into the right common iliac artery because her left femoral artery was tortuous. After compressing the right lower abdomen, we managed to introduce REBOA in Zone 1. It took approximately 60 minutes to successfully place the catheter. The patient’s SBP increased immediately after the balloon inflation and bleeding was endoscopically controlled. Inflation time was 36 minutes and no complications were observed.

Conclusions: Although REBOA could be a useful procedure for massive GIB, it could be difficult to place the catheter without fluoroscopy. Caution should be exercised with placement when REBOA is performed as a bedside technique in ICU.

Consent: Informed consent to publish has been obtained from the patient

P101 Modulation of endothelial glycocalyx and microcirculation during anaerobic exercise by supplementation with pomegranate extract

Z Pranskuniene1, E Belousoviene2, N Baranauskiene3, N Eimantas3, M Brazaitis3, A Pranskunas2

1Lithuanian university of health sciences, Department of Drug Technology and Social Pharmacy, Kaunas, Lithuaniadevices in cardiogenic; 2Lithuanian university of health sciences, Department of Intensive Care Medicine, Kaunas, Lithuania; 3Lithuanian Sports University, Sports Science and Innovation Institute, Kaunas, Lithuania

Introduction: The natural components of the pomegranate fruit may provide additional benefits for endothelial function and microcirculation. We hypothesized that chronic supplementation with pomegranate extract might improve glycocalyx properties and microcirculation during anaerobic condition.

Methods: Eighteen healthy and physically active male volunteers aged 22–28 years were recruited randomly to the pomegranate and control groups (9 in each group). The pomegranate group was supplemented with pomegranate extract for two weeks. At the beginning and end of the experiment, the participants completed a high intensity sprint interval cycling-exercise (anaerobic exercise) protocol. The systemic hemodynamics, microcirculation flow and density parameters, glycocalyx markers, and lactate and glucose levels were evaluated before and after the two exercise bouts.

Results: No significant differences in the microcirculation or glycocalyx were found over the course of the study. The lactate levels were significantly higher in both groups after the first and repeated exercise bouts, and were significantly higher in the pomegranate group relative to the control group after the repeated bout: 13.2 (11.9–14.8) vs. 10.3 (9.3–12.7) mmol/L, p = 0.017.

Conclusions: Chronic supplementation with pomegranate extract has no impact on changes to the microcirculation and glycocalyx during anaerobic exercise, although an unexplained increase in blood lactate concentration was observed.

P102 Identification of cells associated with VA-ECLS mortality by single cell expression profiling

E Kort, M Weiland, E Eugster, E Grins, M Leacche, S Fitch, T Boeve, G Marco, M Dickinson, P Wilton, S Jovinge

Fredrik Meijer Heart And Vasc Inst, Spectrum Health/Van Andel Inst, DeVos Cardiovascular Research Program, Grand Rapids, United States

Introduction: Extracorporeal membrane oxygenation in adults in accompanied by high mortality. Our ability to predict who will benefit from ECMO based on currently available clinical and laboratory measures is limited. The advent of single cell sequencing approaches has created the opportunity to identify cell populations and pathophysiological pathways that are associated with mortality without bias from a priori cell type classifications. Identification of such cell populations would provide both an important prognostic markers and key insight into immune response mechanisms and therefore a possibility for advanced drug matching that may impact clinical response to ECMO in these patients.

Methods: Whole genome transcriptomic profiles were generated from a total of 40,935 peripheral blood monocytes obtained from 37 patients at the time of cannulation for ECMO (Fig 1). Differential gene expression analysis was performed with the Monocle package for the R statistical analysis framework. Time-to-event data were analyzed in a survival analysis with a log-rank test for differences.

Results: Genes encoding several members of the heat shock family of proteins were up-regulated in cells from non-survivors. Notably, these genes were expressed by a small fraction of cells (2.4% on average). Nevertheless, the proportion of cells expressing these genes was a significant predictor of survival to 30 days (p = 0.02 by log rank test), with a particularly pronounced effect in the first 5 days after initiation of ECMO support (Fig 2).

Conclusions: The proportion of cells expressing genes encoding members of the heat shock proteins is predictive of survival on ECMO.

Fig. 1 (abstract P102).

Project methodology - single cell whole transcriptome profiling

Fig. 2 (abstract P102).

Kaplan Meier curves for patients on VA-ECLS stratified by proportion of HSP positive cells at time of cannulation as measured by single cell RNASeq

P103 Clinical features and management of patients with high-risk pulmonary embolism in a Moroccan hospital: analysis of 6 years of data

A El Khaoudi, R Benmalek, S Ballali, E Bennouna, L Azzouzi, R Habbal

CHU iBN Rochd, Intensive care, Cardiology department, Casablanca, Morocco

Introduction: High-risk pulmonary embolism is a life-threatening disorder associated with fatal outcomes. The estimated mortality of PE complicated with cardiogenic shock is between 25% and 30%. The diagnosis of PE in hemodynamically unstable patients remains a challenge for both intensivists and cardiologists since the clinical presentation is variable and non specific. The aim of this study is to assess the clinical features, electrocardiographic (EKG) and echocardiographic findings associated with high risk PE, and to describe the management and prognostic of this disease.

Methods: a single-center, cross-sectional descriptive study including all patients admitted for PE confirmed by transthoracic echocardiography or Computed tomography angiography complicated with cardiogenic shock in the cardiology intensive care unit in the Casablanca university hospital, from September 2011 to July 2017. During the 6 years study period, 69 patients were included.

Results: Mean age was 55.5 years (+/- 8.3), females represented 71.3%. All the patients were hemodynamycally unstable at admission, presenting with dyspnea stage III (41.7%) to IV (52.6%), and/or acute chest pain (23.2%). Tachypnea, hypoxia, and signs of deep vein thrombosis were respectively present in 59, 41, and 24 %. EKG abnormalities were represented by sinus tachycardia in 62.5%, S1Q3 pattern in 27%, RBBB in 19,6% and atrial fibrillation in 17.4%. Echocardiographic aspect of acute cor pulmonale was found in all the patients, right atrial thrombus in 15 patients (21.7%), CT angiography was performed in 39 patients after stabilization. Thrombolytic therapy was performed in 46 patients (66.6%). The mortality rate of our serie was 41.8%.

Conclusions: High-risk PE is a severe disorder associated with high mortality and morbidity, thus a multidisciplinary coordination between cardiologists, radiologists and intensivists is crucial for the management of patients with high-risk PE.

P104 Thrombolytic therapy for submassive pulmonary embolism – single centre experience

N Maric, M Mackovic, N Udiljak

Clinical hospital Sveti Duh, Intensive Care, Zagreb, Croatia

Introduction: Pulmonary embolism (PE) is one of the major causes of mortality, morbidity and hospitalization in Europe. Haemodynamic benefits of thrombolysis in high risk PE are undeniable, yet use of thrombolytic agents in intermediate risk category is controversial due to lack of evidence and fear of complications.

Methods: Single-centre retrospective study of 80 PE patients (pt) treated with alteplase (rtPA) in a medical ICU from January 2015 to November 2018 was performed. General data, risk category, PESI and Qanadli score, TnI, ECG and ABG were obtained. Primary outcome was the complication rate, especially bleeding and secondary the survival rate.

Results: PE comprised 8.7% (214/2457) of all admissions. 80 of them (37.4%) received rtPA. F:M ratio was 51:29, mean age of 68.9y (26-90). 40% of pt were high risk, 54% intermediate (45% high, 9% low) and 6% were in low risk category. Mean PESI was 139.3 suggesting overall high mortality risk. Mean Qanadli score was 21.97 representing high clot burden. Fibrinolytic therapy was applied during CPR in 11 pt with survival rate of 45%. 83% of pt had elevated TnI levels, and 62.5% had newly diagnosed DVT. Majority of pt (43%) had no known predisposing conditions, followed by immobility (21%) and cancer (15%). In ECG analysis tachycardia and V1-V3 T wave inversion were the most common findings whereas hypoxemia± hypocapnia were the most prominent features in ABG analysis. 8 pt (10%) had bleeding complications (none intracranial), 3 (3.7%) during rtPA, 5 (6.3%) in the first 36h and only 3 pt required transfusion. Mortality rate was 12%: 6% directly due to PE (all during CPR) and 6% due to late complications (newly diagnosed cancer and infections).

Conclusions: In our experience, fibrinolytic therapy is safe and effective but in submassive PE should be applied after thorough assessment of risks and benefits on individual basis aiming to patient tailored precision medicine.

P105 Our experience of emergent cases: the surgical treatment of thrombo-embolus in transit

M Obeid1, A Abdurakhmanov1, O Mashrapov1, S Muminov2, N Rakhimov1, I Abdukhalimo1

1Health Ministry Of The Republic of Uzbekistan Republic Research Center Of Emergency Medicine Of Uzbekistan, Cardiosurgery, Tashkent, Uzbekistan; 2Health Ministry Of The Republic Of Uzbekistan Republic Research Center Of Emergency Medicine Of Uzbekistan, Emergency Vascular Surgery, Tashkent, Uzbekistan

Introduction: Blood clots of the main veins of the lower extremities are the main cause of pulmonary thromboembolism (PE). Pulmonary thromboembolism is recorded in 2-15% of autopsies. Floating blood clots in the path from the legs to the pulmonary arteries are a severe form of venous thromboembolism with high early mortality.

Aim of the study: Our surgical experience and prophylaxis of transient pulmonary thromboembolism are demonstrated in the study.

Methods: During the period from 2015 to 2018, 6 patients with pulmonary thromboembolism have been on surgery at the RSCEMP. All patients were female. The average age of patients was 56.2 ± 3.4 years old. All patients came to us at an early stage with deep vein thrombosis, mainly after the removal of uterine fibroids and inadequate hormone therapy. In all patients, hypoechoic loose thromb were observed during the examination of the lower limb CDS, and the floating atrial right thrombus was visualized on the echocardiography. All operations were performed under the conditions of artificial blood circulation with sternotomy access. After connecting the CPB pump, as a rule, the surgery is ended with a thromboembolectomy from the right atrium. On the next stage, inferior vena cava clipping was performed through the retroperitoneal access (using the original clip).

Results: Mortality rate was 0% during the hospital stay. In the long-term period of 18 months, mortality was also not noted. The inferior vena cava syndrome was developed in 2 patients. In 4 patients, postthrombotic syndrome was developed during long-term period.

Conclusions: Thrombo-embolectomy from the right atrium and inferior vena cava clipping prevents massive pulmonary embolism and prolongs the life of the patients. However it can aggravate the development and course of chronic venous insufficiency.

P106 New onset versus pre-existing atrial fibrillation in critically illness: Do patient populations differ?

B Johnston, N Miller, A Hampden-Martin, I Welters

Royal Liverpool University Hospital, Intensive Care Unit, Liverpool, United Kingdom

Introduction: Atrial fibrillation (AF) is the commonest arrhythmia in critical illness and is associated with mortality. Systemic inflammation and infection have been identified as triggers for fast heart rate and development of new-onset AF (NOAF) during critical illness. It is unknown, however, if patients with pre-existing AF (PEAF) who develop a fast heart rate differ from patients with NOAF with regard to organ function, biometric characteristics and disease severity.

Methods: This study was performed at a large inner-city University Hospital between January 1st and December 31st 2017. Patients who had fast AF during their critical illness were prospectively identified by the research team. Fast AF was defined as patients with AF and a heart rate above 110/minute. Biometric data, routine blood results, previous medical history, treatment of AF and duration of fast AF were collected. Patients with pre-existing AF (PEAF) were compared with those who had no history of arrhythmias documented.

Results: Atrial fibrillation was new-onset in 49 patients (50.5%), and pre-existing in the remaining 48 (49.5%). The APACHE II score on admission was higher for NOAF patients compared to PEAF patients (17.0 (13.5 – 21.5) vs 14.0 (11.0 - 17.0), p=0.005). In addition, a greater number of NOAF patients were admitted with sepsis, though these differences were not significant. Patient with NOAF were significantly younger than those with PEAF (67.0 (59.0 - 77.0 years vs 74.5 (66.5 - 79.0 years), p=0.008). Chronic cardiovascular comorbidities were more frequent in patients with PEAF. Kidney function was worse in patients with NOAF, platelet counts was lower and prothrombin times higher.

Conclusions: As expected, patients with PEAF suffered more often from cardiovascular comorbidities and were older. In contrast NOAF patients were sicker and had worse clotting profiles and kidney function. Our results indicate that patients with PEAF differ with regard to organ function and co-morbidities. Consequences for treatment remain unclear at this stage.

P107 Dobutamine after cardiac surgery: a randomised controlled clinical trial

R Franco1, JL Vincent2, J Almeida3, J Fukushima1, G Oliveira1, S Rizk1, C Park1, M Mourão1, G Landoni4, L Hajjar1

1Heart Institute, Cardiology, Sao Paulo, Brazil; 2Dept of Intensive Care, Erasme Hospital (Université libre de Bruxelles), Brussels, Belgium; 3Heart Institute, Surgical ICU, Sao Paulo, Brazil; 4IRCCS San Raffaele Scientific Institute, Milan, Italy

Introduction: The purpose of the study was to evaluate whether a restrictive strategy regarding dobutamine use was non-inferior to a liberal strategy in patients undergoing cardiac surgery.

Methods: Clinical trial, randomised, unicentric, controlled, parallel-group, non-inferiority trial. Patients were randomly assigned preoperatively to two distinct dobutamine strategies: a liberal strategy, in which all patients would receive dobutamine after weaning from CPB; or a restrictive strategy, in which the use of dobutamine after CPB weaning would be guided by hemodynamic evidence of low cardiac output. The primary outcome was composite endpoint of arrhythmias (ventricular or supraventricular tachyarrhythmias), acute myocardial infarction, stroke, and death from all causes within 30 days after cardiac surgery.

Results: A total of 160 patients were included in the final analysis; 80 assigned to the restrictive strategy and 80 to the liberal strategy. The use of dobutamine was lower in the restrictive group (67.1 vs. 100%, P <0.001). The primary outcome occurred in 31.3% of the restrictive group and 33.8% of the liberal group (P = 0.736). There were no significant differences between the restrictive and liberal strategies regarding the incidence of supraventricular or ventricular tachyarrhythmias (23.8% vs. 27.5%, P = 0.587), cardiogenic shock (11.3% vs. 13.8%, P = 0.633), low cardiac output syndrome (13.8% vs. 15%, P = 0.822), acute myocardial infarction (1.3% vs. 2.5%, P = 1,000), stroke, 5% vs. 0%, P = 0.497) and death from all causes (2.5% vs. 6.3%, P = .443).

Conclusions: The use of a restrictive strategy regarding the use of dobutamine, based on a clinical scenario of reduction of the cardiac index associated with signs of tissue hypoperfusion, is non-inferior to a liberal strategy in patients undergoing cardiac surgery.

P108 Preconditioning with levosimendan is cost-effective

A Alvarez1, J Ferrer2, J Jiménez-Rivera1, J Iribarren1, J Montoto3, J Lacalzada-Almeida4, P Jorge4, MJ García-González5, R Martinez-Sanz3, M Mora1

1Hospital Universitario de Canarias, Intensive Care, La Laguna, Spain; 2Hospital Universitario de Canarias, Health Economist. University La Laguna, La Laguna, Spain; 3Hospital Universitario de Canarias, Cardiovascular Surgery, La Laguna, Spain; 4Hospital Universitario de Canarias, Cardiology Department, La Laguna, Spain; 5Hospital Universitario de Canarias, La Laguna, Spain

Introduction: Coronary Artery Bypass Grafting (CABG) plays an important role in patients with left ventricular dysfunction and major vessels disease, in this context it becomes necessary to design cost-effective strategies like preconditioning with levosimendan.

Methods: Two strategies are compared in terms of cost-effectiveness: Preconditioning with levosimendan (n=15), compared to standard care (n=117). Both strategies were used in elective CABG with preoperative Left Ventricular Ejection Fraction less than 40%. The adverse effects studied included: postoperative de novo atrial fibrillation, low cardiac output (LCO), renal failure and prolonged mechanical ventilation. The costs of adverse effects were evaluated. Deterministic and probabilistic sensitivity analysis was performed, and Monte Carlo simulations.

Results: Average cost on levosimendan group was €14,601.72 while the average cost per patient without levosimendan was €18,401.71. Mean ICU stay per patient was estimated at 2.47 and 5.18 days with and without levosimendan use respectively. Patients with no complications represented 46.7% of the total in the levosimendan arm, as compared to 27.4% in the non-levosimendan arm. Figure 1 shows, in the cost-effectiveness plane, the cost and effectiveness pairs obtained as a result of each Monte Carlo simulation. In all simulations, use of levosimendan was less expensive and more effective. The variable incidence of LCO does have an impact on the potential complications resulting from it. Cost per patient in the base case was €14,601.73 ranging from €13,264.27 and €14,103.12 with administration of levosimendan, as compared to €18,401.71 for the base case, ranging from €16,899.89 and €17612.64. Low Cardiac Output, represents 6%-18% in the intervention arm vs 25%-35% in the control arm.

Conclusions: Preconditioning with levosimendan, is a cost-effective strategy preventing postoperative LCO in high-risk patients undergoing elective CABG.

Fig. 1 (abstract P108).

Cost-effectiveness plane

P109 Analysis of the LEVO-CTS and LICORN trials on pre-operative levosimendan treatment in isolated CABG surgery

M Heringlake1, G Chatellier2, T Caruba2, P Pollesello3, M Kivikko4, T Sarapohja5, W Toller6

1Universitätsklinikum Schleswig-Holstein, Klinik für Anästhesiologie und Intensivmedizin, Lübeck, Germany; 2Université Paris Descartes, Sorbonne Paris Cité, Paris, France; 3Orion Pharma, Hospital Care, Espoo, Finland; 4Orion Pharma, Global Medical Affairs, Espoo, Finland; 5Orion Pharma, Statistics, R&D, Espoo, Finland; 6Medical University, Universitätsklinik für Anästhesiologie und Intensivmedizin, Graz, Austria

Introduction: Both the LEVO-CTS[1] and LICORN[2] trials evaluated the role of levosimendan in preventing low cardiac output syndrome in patients undergoing cardiac surgery. The studies were similar in their design and recruited patients with preoperatively low LVEF undergoing either isolated CABG or valve surgery combined with CABG (Table 1). In both, a 24-hour levosimendan infusion was started at induction of anesthesia. Neither study met the primary efficacy composite enpoints, but both showed a clear tendency for better outcome in patients undergoing a CABG compared to a valve procedure. We are currently evaluating the solidity of a co-analysis based on shared end-points.

Methods: We are planning a shared analysed of the data related to the CABG settings and analyze the aggregated mortality data for both studies at 1 and 3 months by Cochran-Mantel-Haenszel odds ratio. Data from individual studies would be analysed as fixed effect and Breslow-Day test was used to evaluate homogeneity of the odds ratios

Results: In the placebo groups of the two studies, the mortality is similar; 7.9% (22/279) in LEVO-CTS and 7.3% (9/123) in LICORN, corroborating the working hypothesis that the two studies can be co-analysed. In a preliminary combined analysis (Fig 1), 90-day mortality was 7.7% (31/402) in the placebo group and 2.9% (12/407) in the levosimendan group. Odds ratio was significantly in favor of levosimendan (0.36; 95% confidence interval 0.18-0.72; p=0.0026, Fig. 1)

Conclusions: The LEVO-CTS and LICORN trials can be co-analysed in their sub-setting of patients requiring isolated CABG surgery for mortality at 1 and 3 months. A preliminary analysis on mortality reinforce the hypothesis that, in isolated CABG surgery, levosimendan lowers post-operative mortality significantly both at 1 and 3 months, when started at the induction of anesthesia


[1] Mehta RH et al. N Engl J Med. 2017;376(21):2032-2042.

[2] Cholley B et al. JAMA. 2017;318(6):548-556

Table 1 (abstract P109). The designs of LEVO-CTS and LICORN studies
Fig. 1 (abstract P109).

LEVO-CTS and LICORN CABG all-cause mortality at 3 months (combined n=809)

P110 Relevance of mobile intensive care unit during pre-hospital phase treatment in ST elevation myocardial infarction

R Viejo Moreno1, A Cabrejas Aparicio2, M Gálvez Marco3, N Arriero Fernández4, Z Eguileor Marín4, J Balaguer Recena5, C Carriedo Scher6, C Marian Crespo4

1GUETS - SESCAM, MICU Guadalajara, Guadalajara, Spain; 2MICU Azuqueca de Henares; GUETS - SESCAM, Guadalajara, Spain; 3MICU-Guadalajara; GUETS - SESCAM, Guadalajara, Spain; 4Intensive Care Unit - H.U Guadajara, Guadalajara, Spain; 5Cardiology department - H.U. Guadajara, Guadalajara, Spain; 6GUETS - SESCAM, Guadalajara, Spain

Introduction: Emergency medical system (EMS) -based ST elevation myocardial infarction (STEMI) networks allows not only STEMI diagnosis in the pre-hospital phase but also reduces treatment delays; treat your fatal complications and the immediate activation of the catheterization laboratory. The aim of study was to investigate the effect of out-of-Hospital by Mobile Intensive Care (MICU) versus Hospital beginning treatment in hospitalization length and survival of patients with STEMI diagnosis

Methods: Observational retrospective study on STEMI patients managed by prehospital MICU and the Emergency Department (ED) of Hospital Guadalajara (Spain) from January 2015-September 2018. Items compared were demographic variables, cardiovascular risk factors, treatment, time to first medical contact to percutaneous coronary intervention (PCI), location of coronary lesions, GRACE score, length in ICU and hospital stay and survival to discharge and to 30 days.

Results: 360 were evaluated by STEMI, 219 (60.8%) by MICU. 280 (77,8%) male, mean age was 61.0 (RIC: 53.0-71.0) years old (Fig.1). There were differences in P2Y12 inhibitor administration. Ticagrelor was more frequent in patients admitted directly in hospital 94 (66.7%) Vs 82 (37.4%) (p=0.00) instead of Prasugrel by MICU physicians 13 (9.2%) Vs 112 (51.2%) (p=0.00). Morphine was more administered in MICU 49 (34.8%) VS 121 (55.3%) patients (p=0.00). Initial attention of patients with STEMI by MICU reduced time to PCI, stay length in ICU: 76.8 (37.4-191.0) Vs 44.8 (36.6-53.1) hours (p=0.02), and was associates lower mortality at hospital discharge 13 (9.2%) Vs 8 (3.7%) (p=0.02) and to 30 days compared to those arriving directly the hospital by themselves 15 (10.6%) VS 10 (4.6%) (p=0.02).

Conclusions: Around 40% STEMI patients still come to hospital by their own. A STEMI out-and-in of Hospital network and MICU based EMS with presence of physician allows to reduce the time to PCI and decreases the mortality of patients with SCACEST in Guadalajara.

Fig. 1 (abstract P110).

Demographic, clinical and prognosis characteristics of acute coronary syndrome groups. MICU: mobile intensive care unit. CRF: cardiovascular risk factor; ORB: oxygen reservoig bag; ET: endotracheal tube; PCI: percutaneous coronary intervention; LMCA: left main coronary artery; LADA; left anterior descending; LCA: left circumflex artery; RCA: right coronay artery. ICU: intensive care unit.

P111 Contrast induced nephropathy after primary percutaneous coronary intervention in STEMI patients

M Trajkovic1, M Kovacevic1, A Vulin1, M Femic2, M Petrovic1, S Dimic1, M Jarakovic1, B Crnomarkovic1, S Keca1, I Srdanovic1

1Insitute of Cardiovascular Diseases of Vojvodina, ICCU, Novi Sad, Serbia; 2Clinical Center of Vojvodina, ICU, Novi Sad, Serbia

Introduction: Contrast induced nephropathy (CIN) is a complex acute renal failure syndrome, which can occur after primary percutaneous coronary intervention (PCI) and is an important cause of morbidity and mortality in this subgroup of patients. The aim of our study was to establish the incidence and predictors of CIN after primary PCI.

Methods: We performed a retrospective analysis of STEMI patients treated with primary PCI in the period from January until September of 2017. CIN was defined as an absolute increase in baseline serum creatinine of ≥0.5 mg/dL (44 μmol/l) or >25% relative rise within 72 hours after primary PCI. We analyzed demographic characteristics, risk factors, clinical status at hospital admission, laboratory parameters, left ventricle ejection fraction and data regarding PCI procedure.

Results: The study included 729 patients, with an average age of 62.33 ± 11.78 years, 66.1% of the patients were males. An average of 175.67 ± 91.77ml of contrast medium per patient was utilized. CIN developed in 11(1.5%) patients and overall intra-hospital mortality was 8.4 %. In multivariate analysis, the independent predictors of CIN were age>75 years (OR 6.434; 95% CI (1.174 - 35.256); p=0.032), diabetes mellitus (OR 21.74; 95% CI (2.672 - 176.891); p=0.004), Killip class III (OR 17.394;95% CI (2.309 - 146.539); p=0.006), creatinine at admission (OR 1.011;95% CI (1.004-1.018);p=0.03), and hemoglobin level at admission (OR 0.960;95%CI (0.925 - 0.996); p=0.029). Patients with CIN had significantly longer length of hospitalization (10.63 ± 7.86 vs. 5.79 ± 5.89 days; p=0,007) and higher intra-hospital mortality (27.3% vs. 8,1%; p=0.023). CIN was independent predictor of mortality (OR 12.485; 95% CI (1.472-105.909); p=0.021).

Conclusions: In our study group of STEMI patients treating with primary PCI, the independent predictors of CIN were age >75 years, diabetes mellitus and higher Killip class at admission. CIN was the independent predictor of mortality.

P112 Early outcome of left main stem disease coronary bypass surgery in intensive care unit

A Omar, S Hanoura, P Sivadasan, S Sudarsanan, H Osman, Y Shouman, A AlKhulaifi

Hamad Medical Corporation, Heart Hospital-Cardiothoracic surgery, Doha, Qatar

Introduction: Left main coronary artery (LMCA) disease is a disease of the main coronary branch that gives more than 80% of blood supply to the left ventricle, it carries high mortality without surgical intervention; [1] however the influence of LMCA surgery on morbidity ICU measures needs to be explored. We aim to determine whether LMCA is definitive risk factor for prolonged ICU stay as a primary outcome and whether LMCA is definitive risk factor for early morbidity

Methods: Retrospective descriptive study with purposive sampling analyzing 398 patients underwent isolated coronary artery bypass surgeries (CABG). Patients were divided into 2 groups those with LMCA disease as group 1 (75 patients) and those with coronary arty disease requiring surgery but without LMCA disease as group 2 (324 patients) then we will correlate with ICU outcome parameters including ICU stay length, post-operative atrial fibrillation, acute kidney injury, re-exploration, perioperative myocardial infarction, post operative bleeding and early mortality.

Results: Patients with LMS had significantly higher diabetes prevalence (43.3% vs 29%, p=0.001). However, we did not find a statistical significant difference regarding ICU stay, or other morbidity and mortality outcome measures

Conclusions: Diabetes was more prevalent in patients with LMS. The latter group showed similar outcome as those without LMS in this study these findings may help in guiding decision making for future practice and stratifying the patients care.


1) Göl MK, et al. Journal of cardiac surgery. 2000;15:217-22.

P113 Impact of comorbidities and patient age on clinical outcomes in patients with myocardial infarction - a nationwide cohort study

C Bachli1, U Wagner2, B Müller1, P Schütz1, A Kutz1

1Kantonsspital Aarau, Aarau, Switzerland; 2Swiss Federal Office for Statistics, Neuchâtel, Switzerland

Introduction: Multimorbidity in patients admitted for acute myocardial infarction [AMI] is associated with higher risk for in-hospital mortality and adverse clinical outcomes. We investigated to what extent an increasing number of comorbidities affects the age-stratified excess risk of death and other clinical outcomes among patients with myocardial infarction.

Methods: We analyzed nationwide administrative data of 174`803 admissions for an acute myocardial infarction between 2006 and 2016. We calculated multivariate regression models to study the association of four comorbidities (chronic kidney disease [CKD], diabetes mellitus, heart failure [HF], and atrial fibrillation) and excess risk of in-hospital mortality, length of hospital stay [LOS], and 30-day readmission and stratified the analysis for different age categories.

Results: The incidence of admissions for AMI increased continuously during the observed decade without an increase in in-hospital mortality, LOS, and 30-day readmission. Among admitted patients with AMI, there was a stepwise increase in risk for adverse outcomes for each comorbidity. Compared to patients with no comorbidity, patients with 4 comorbidities had 6-fold increased risk for mortality (adjusted odds ratio [OR] 6.6, 95% confidence interval [CI] 5.6 to 7.7) and a similar risk for readmission (OR 1.0, CI 0.9 to 1.2). The LOS was 5.3 days (CI 5.1 to 5.5) in patients with no comorbidity and increased by 2.5 days (CI 2.4 to 2.5) with each additional comorbidity. These associations were stronger in younger compared to older patients. CKD was the strongest predictor of in-hospital mortality and LOS, while HF was the strongest predictor of 30-day readmission.

Conclusions: This study of nationwide admitted patients with AMI found a stepwise increase in the risk for adverse outcome with increasing number of comorbidities, particularly in the younger patient population. Younger, multimorbid patients may thus have the largest benefits from multidisciplinary treatments.

P114 Preferable quality of life on average 2 years after emergency ECLS therapy

C Lang1, F Schroth1, V Zotzmann1, T Wengenmayer1, B Schmid2, C Bode1, D Staudacher1

1University Heart Center Freiburg, Cardiology, Freiburg, Germany; 2University Emergency Center, Freiburg, Germany

Introduction: Certified cardiac arrest centers, sophisticated post cardiac arrest care and prehospital ECLS teams aim to increase survivor rates with a preferable neurological outcome after cardiac arrest. Centers also provide emergency ECLS and ECLS pick ups for cardiogenic shock patients before arresting. Few data answer the question of the long-term quality of life after ECLS therapy.

Methods: In a retrospective single center register we included patients after emergency ECLS (eCPR and cardiogenic shock) between 10/2010 and 10/2017 discharged alive and performed a follow-up after 2 years on average at 6/2018. In our center criteria to initiate ECLS therapy in cardiogenic shock or under cardiac arrest are an observed collaps, shockable rhythm, absence of frailty and severe comorbidities. All patients were requested to take part in a telephone interview. Thus, we analyzed survival, CPC scores and SF36 scores.

Results: 97 patients with hospital survival after ECLS were screened. 44% (N=43) had survived until 6/2018; 38 patients were not accessible; 16 had ceased. 33 survivors (mean±SD; min-max; 53±17; 19-78 years, 8 women) answered SF36 questionaires 29±15; 7-64 months after ECLS (45% cardiogenic shock, 55% eCPR with shockable rhythm in 89%). The participants´ CPC scores were in median 1. The results of the SF36 were physical functioning 84±21, physical role functioning 84±28, bodily pain 92±17, general health 68±23, vitality 65±26, social role functioning 83±28, emotional role functioning 96±14 and mental health 81±22. Survivors who did not take part at the SF36 had a CPC score of in median 2 (N=10, 5 personally signed refusals, 3 language barriers, 2 vegetative states).

Conclusions: After emergency ECLS therapy and hospital survival 44% of our patients survived the following 2 years up to over 5 years with a preferable neurological outcome and a general mentally and physically satisfactory quality of life. A vague outcome in 39% limits the results of our study.

P115 Cutaneous blood flow as a predictor of successful weaning from VA-ECMO

W Mongkolpun, P Bakos, L Peluso, F Annoni, JL Vincent, J Creteur

Erasme Hospital, Université libre de Bruxelles, Intensive Care Department, Brussels, Belgium

Introduction: Successful weaning from VA-ECMO requires the restoration of a sufficient cardiac function to ensure an adequate tissue perfusion. Skin blood flow (SBF) is among the first to deteriorate during circulatory shock and the last to be restored after resuscitation. SBF would be a good predictor of successful weaning from VA-ECMO.

Methods: Patients with VA-ECMO, who required a first weaning attempt, were included. Weaning procedure (WP) was performed by a reduction of VA-ECMO blood flow to 1 L/min for 10 minutes. The weaning criterion was an aortic velocity–time integral (VTI) > 10 cm. Successful weaning from VA-ECMO was defined as hemodynamic stabilization and without the need to increase the vasopressor dose during the next 24 hours. SBF, assessed by skin laser Doppler (Periflux5000, Perimed, right index finger); perfusion unit: PU), together with global hemodynamic parameters were obtained before and after 3 min of weaning. Receiver operating characteristic curves (ROC) were generated to assess the ability and reliability of baseline parameters to predict a successful weaning.

Results: We studied 22 WPs in patients with VA-ECMO for pulmonary embolism (n = 2), post cardiotomy (n = 3), acute coronary syndrome (n = 14), myocarditis (n = 3). These were successful (SW) in 12 and unsuccessful (NSW) in 10. At baseline, hemodynamic variables, lactate, ECMO blood flow were similar in both groups (Table1). SBF was greater in SW than NSW patients (Table1). During WP, CI rose from baseline and was similar in SW and NSW (p=0.1) (Table 2). VTIs were higher in SW than NSW (13 (12-14) vs 8 (6-8), respectively, p=0.03). SBF decreased in SW and remained low in NSW (Table 2). From the ROC curves analyses, baseline SBF had the highest area under the ROC curve with a cut off ≥ 34 PU (sensitivity 83%, specificity 92%) (Figure1).

Conclusions: SBF is a good predictor of successful weaning from VA-ECMO

Table 1 (abstract P115). Baseline clinical characteristic including SBF between successful and unsuccessful weaning groups
Table 2 (abstract P115). Hemodynamic parameters and SBF during WP between successful and unsuccessful weaning groups (*p<0.05 Before WP vs During WP)
Fig. 1 (abstract P115).

Baseline parameters and VTI during WP to predict successful weaning from VA-ECMO

P116 Postoperative cognitive dysfunction after endo-CABG: an observational prospective cohort study

I Gruyters1, J Dubois2, J Vandenbrande2, A Yilmaz3, M Van Tornout2, JP Ory2, B Stessel2

1Jessa Hasselt, Anesthesia, Hasselt, Belgium; 2Jessa Hasselt, Dept of anesthesia and intensive care, Hasselt, Belgium; 3Jessa Hasselt, Dept of cardiac-thoracic surgery, Hasselt, Belgium

Introduction: Postoperative cognitive dysfunction (POCD) is defined as a temporarily decline in cognition associated with surgery. Long-term POCD (3 months after surgery) occurs in 10-30% of cardiac patients and is associated with a higher morbidity and mortality. Endo-CABG is a new minimally invasive endoscopic coronary artery bypass grafting (CABG) technique that requires retrograde arterial perfusion which may be associated with a higher incidence of neurological complications. The aim of this study is to assess the incidence of POCD after endo-CABG.

Methods: Sixty consecutive patients undergoing an endo-CABG were enrolled. POCD was assessed following the recommendations of the “1995 statement of consensus on assessment of neurobehavioral outcomes after cardiac surgery”. A comparative group of 60 patients undergoing percutaneous coronary intervention (PCI) and a control group of 60 healthy volunteers were also enrolled. Additional tests included the digit span test and digit symbol-coding test. Patients were tested at baseline and at 3-month follow-up. POCD is defined as a Reliable Change Index (RCI) ≤ -1.645 (significance level 5%), or Z-score ≤ -1.645 in at least two different tests.

Results: After enrolling 60 patients in each group, respectively 46 in the Endo-CABG-group, 44 in the PCI-group and 48 healthy controls were analysed. Patients suffering from a CVA within three months after their procedure were automatically classified as having POCD (PCI: n= 1; Endo-CABG: n= 1). The total incidence of POCD was not different between groups (PCI: n= 7; Endo-CABG: n= 6, p=0.732).

Conclusions: Our results suggest that the risk of POCD after Endo-CABG is low and comparable with the risk of POCD after PCI.

P117 30-day clinical outcome following endo-CABG: an observational cohort study

F Polus1, A Yilmaz2, P Starinieri2, B Robic2, J Dubois1, B Stessel1

1Jessa Hospital, Anesthesiology & Intensive Care, Hasselt, Belgium; 2Jessa Hospital, Cardiothoracic Surgery, Hasselt, Belgium

Introduction: Traditional CABG via median sternotomy is associated with major disadvantages such as a prolonged recovery time, a poor cosmetic result, a significant risk of chronic post-sternotomy pain, sternal instability, delayed bone healing and wound infection. Therefore, a new minimally invasive endoscopic coronary artery bypass grafting (Endo-CABG) technique to treat patients with single- and multi-vessel coronary artery disease has been developed. Endo-CABG combines a thoracoscopic technique via 3 thoracic ports (5 mm) and a mini-thoracotomy utility port (3-4 cm) through the intercostal space. Hence, all walls of the heart can be easily grafted. The aim of this study is to discuss the short-term clinical results of endo-CABG.

Methods: From 01/2016 to 01/2018, data from 342 consecutive patients undergoing an Endo-CABG at the Jessa Hospital, Belgium, were prospectively entered into a customized database. This database was merged with data from the Belgian Association for Cardio-Thoracic Surgery (BACTS) and retrospectively reviewed. Subgroup analysis is performed based on the European system for cardiac operative risk evaluation (EuroSCORE). Data are presented as mean (SD) or number (%).

Results: Seven patients were excluded from the analyses because of missing data. Up to 85% of the patients presented with multivessel disease necessitating multivessel endo-CABG. A Y-graft anastomosis was performed in 40% of the patients. Respectively 78%, 18% and 2% of the patients had a low risk, a moderate risk and a high risk EuroSCORE. Mortality at 30 days was comparable with the predicted EuroSCORE: 0% in the low risk group, 6,5% in the moderate risk group and 25% in the high risk group. Graft failure was present in 1.2% of all cases. The overall incidence of in hospital revision was 7.3%.

Conclusions: Our results suggest that the Endo-CABG technique is a safe and effective procedure to treat single- and multi-vessel coronary artery disease without the need for a specific patient selection.

P118 Incidence and outcome of rhabdomyolysis after aortic dissection surgery

P Sivadasan1, AS Omar2, C Carr3, A Pattath4, S Hanoura4, S Sudarsanan4, H Ragab4, A Karmakar4

1Hamad Medical Corporation, Cardiac Anaesthesia & ICU Division, Department of Cardiothoracic Surgery, Doha, Qatar; 2Hamad Medical Corporation, Cardiothoracic intensive Care, Doha, Qatar; 3Hamad Medical Corporation, Cardiothoracic Surgery, doha, Qatar; 4Hamad Medical Corporation, Cardiac Anesthesia, Doha, Qatar

Introduction: Rhabdomyolysis ( RML) post aortic surgery probably affects the renal outcome adversely [1,2]. There is no robust data regarding the same in literature.

Methods: Retrospective single center data review; prior approval from Institutional review board. Patients were divided to two groups Group 1- with RML ( CK above cut off levels 1050 U/Litre) and Group 2 without RML. The determinants of RML and the impact of the same on outcome; predominantly renal function was evaluated. Chi-square tests are performed for categorical variables whereas, student t tests (un-paired ) are performed with continuous variables. Correlation is performed between Creatine kinase and creatinine rise. P value 0.05 (two tailed) is considered for statistical significant level.

Results: Out of 33 patients, 21 patients (63.64%) developed Rhabdomyolysis ( GROUP RML) and 12 did not( GROUP NON RML). Demographic and intraoperative factors had no significant impact on the incidence of RML. There was a significantly higher incidence of renal complications including new postoperative dialysis in the RML group. Other morbidity parameters were also higher in the RML group.

Conclusions: There is high prevalence of RML after aortic dissection surgery - Identification of risk factor and early intervention might help to mitigate the severity of renal failure


1. Omar AS, et al. BioMed Res Int, 2016;2016:7497936

2. Anthony DG, et al. Crit Care Med.2011; 39:1992-1994.

P119 Early postoperative leukocytosis as a marker of cardiac surgery outcome. A single center retrospective study

S Hanoura, A Omar, H Osman, S Aboulnaga, M Eissa, Y Shouman, R Singh, A AlKhulaifi

Hamad Medical Corporation, Cardiothoracic surgery/Anesthesia and Intensive Care, Doha, Qatar

Introduction: Despite of the modern advances in cardiac surgeries, cardiopulmonary bypass (CPB) is still widely used that is associated with variant degrees of acute systemic inflammatory reaction where leukocytosis emerges as an early sign [1]. We hypothesize that early post operative leukocytosis (within 36 hours) could be associated with post cardiac surgeries morbidity.

Methods: Single center retrospective observational study over three years. We examined 1145 patients, out of these we recruited 924 patients underwent cardiac surgeries under CPB machine. Patients with off pump cardiac surgeries, preoperative sepsis, recent steroid use, renal failure, preoperative pneumonia, and emergently operated were excluded from the study. Patients were divided into 2 groups based on early white cell count (WBCs), group I: Control WBCs <12000 cells/mL (304 patients) and group II with WBCs > 12000 cells/mL (620 patients).

Results: Patients at group II had significantly higher incidence of acute kidney injury and ventilator associated pneumonia (180 vs 70, 20 vs 0 p= 0.03 and 0.00). The Length of mechanical ventilation and the length of hospital stay were significantly higher in group II (4.9±38.1 vs 9.6±20 hours, and 10±11.9 vs 7.8±5.3 days with P-Value 0.00 and 0.000 respectively). Moreover, surgical re-exploration was significantly higher in group II (66 vs 20, P-Value 0.05).in the meanwhile patients in group II had significant prolonged bypass time and aortic cross clamp time (117±46.6 vs 110.±36.5 and 74±36 vs 70±31 P=0.002 and 0.007) respectively.

Conclusions: Asymptomatic high WBCs post cardiac surgery had a prevalence of 67% in our study and carries a significant increased risk of morbidity intensive care stay. Further studies are needed for validating and explaining these findings.


Lamm G, et al: J Cardiothorac Vasc Anesth 20:51-56, 2006

P120 Postoperative thrombocytopenia after isolated aortic valve replacement

F Ampatzidou, R Ioannidis, A Dimaki, G Kechagioglou, N Mihail, G Drossos

G.Papanikolaou General hospital Thessaloniki, Cardiac Surgery ICU, Thessaloniki, Greece

Introduction: Transient thrombocytopenia is not unusual after aortic valve replacement. The aim of the study was to investigate incidence and risk factors associated with this phenomenon

Methods: In this study we retrospectively analyzed platelet counts during the first five postoperative days, from patients who underwent isolated aortic valve replacement from January 2015 to November 2018 in our Cardiothoracic ICU department. Group A consisted of pts with nadir platelet count less than 60 x 109/L and Group B the rest of the cohort. The following factors were compared between the groups: age >75 years old, gender, pre-op platelet count less than 150x 109/L (and more than 80x 109/L), chronic renal failure(GFR < 60ml/min/1.73 m2), pre-op use of aspirin, pre-op use of clopidrogrel, cardiopulmonary bypass time (CPB)>120 min, use of sutureless prosthetic valve, and transfusion with more than 3 red blood cell units(RBC). Aspirin and clopidrogrel were stopped 5 days before the procedure. Chi square test was used for the statistical analysis

Results: A total of 198 patients underwent isolated aortic valve replacement, during the study period. Platelet count < 60 x 109/L was found in 28(14.1%) pts consisted group A mean aged 76.2±6.4 vs 68.8±11.1. No correlation was found with gender (p=0.85), aspirin use (p=0.06), clopidrogrel use (p=0.67) and CPB time>120 min(p=0.15) Statistical significant factors are shown in Table 1.

Conclusions: Postoperative thrombocytopenia after aortic valve replacement was associated with age >75, low pre-op platelets count, chronic renal failure, and sutureless aortic valve. All patients with thrombocytopenia were transfused with >3 RBC units

Table 1 (abstract P120). Statistical significant factors

P121 Characterization of multi-view hemodynamic data by learning mixtures of multi-output regressors

E Lei1, K Miller1, M Pinsky2, A Dubrawski1

1Carnegie Mellon University, Auton Lab, Pittsburgh, United States; 2University of Pittsburgh, Department of Critical Care Medicine, Pennsylvania, United States

Introduction: We investigate whether central venous pressure (CVP) pressure waveform signal can be informative in detection of slow bleeding in post-surgical patients. We apply a novel machine learning method to analyze CVP datasets to characterize bleeding in a porcine model of fixed rate blood loss.

Methods: Thirty-eight pigs were anesthetized, instrumented with catheters, kept stable for 30 minutes, and bled at a constant rate of 20ml/min to mean arterial pressure of 30 mmHg. CVP waveforms were extracted from inspiration and expiration phases of respiration and statistically featurized. The proposed machine learning method, Canonical Least Squares (CLS) clustering, identifies correlation structures that differ between subsets of observations. We extend it to supervised classification. Both clustering and classification methods yield human-interpretable models that reflect distinctive patterns of correlations within CVP waveforms.

Results: We conducted three experiments to discover structure in the physiological response to bleeding. First, we clustered respiration cycles with full knowledge of blood loss. The color-coded cluster assignments are shown in the Figure 1. They are consistent with escalation of bleeding. Second, we deployed clustering on only CVP features without blood loss. Temporal structure was complemented with some subject-specific clusters (Fig 2). Third, we ran CLS classification to decide whether an observation came from before or after the onset of bleeding (performance shown in the Table 1).

Conclusions: Our results show that the CVP waveforms carry information about the physiologic status of the subject and indicate the amount of blood lost. Clusters still correspond to bleeding status even when no information about bleeding was available to the algorithm. CLS clustering enables a detailed yet interpretable view of discovered structures in complex waveform data.

Work partially funded by DARPA FA8750-17-2-0130 and NIH GM117622.

Table 1 (abstract P121). Bleeding Classification Performance
Fig. 1 (abstract P121).

CLS cluster assignments with 4 clusters when blood loss is known

Fig. 2 (abstract P121).

CLS cluster assignments with 4 clusters when blood loss is unknown

P122 Minimally invasive thoracotomy is a ICU proven alternative approach for mitral valve surgery

E Van laeken1, T Bové2, K Vandewiele2, K Lapage3, H Peperstraete1

1Ghent University Hospital, Intensive Care, Gent, Belgium; 2Ghent University Hospital, Cardiac Surgery, Gent, Belgium; 3Ghent University Hospital, Anesthesiology, Gent, Belgium

Introduction: We aimed to evaluate impact of minimally invasive thoracotomy (MIVT) versus sternotomy on patients (pts) receiving mitral valve surgery.

Methods: A retrospective observational study. Between 18/01/05 and 04/05/17 and after propensity score matching we retained 340 pts (95 sternotomy and 245 MIVT) (Table 1). We used SPSS 25 for parametric and non-parametric tests (p<0.05 = statistically significant).

Results: Hospital stay was 14 (4) days in the sternotomy group and 13 (4) days in the MIVT group, p=0.016.

On day (D) 1, ALT was higher in the MIVT group 30 ±28.6U/L vs. 22.1 ±19.0U/L, LDH was higher in the sternotomy group 533.5 ±187.0 U/L vs. 356.8±143.6U/L. Sternotomy gave lower thrombocyte counts 144.8 ±46.3 vs. 157.6 ±49.8 10³/μL, p=0.031; lower PT 67.8±9.5 vs. 70.8 ±11.0 %; and longer aPTT 39.7±8.2 vs. 37.4 ±5.57sec. So, clotting is less activated in MIVT. Inflammation is activated, at D2 MIVT led to higher white blood cell counts 10.9 (4.7) 10³/μL vs. 9.5 (4.4) 10³/μL; p=0.049, but a lower CRP 16.3 (9.4) mg/L vs 165.0 (67.0)mg/L, p<0.001.

Urea levels on D1 and 2 were lower in the MIVT group, 0.4 ±0.15 vs. 0.4 ±0.17g/L and 0.4 ±0.20 vs. 0.5 ±0.25g/L; p=0.001 & p<0.001. In the MIVT group creatinine kinase was higher 701.0 (595.0) U/L vs 431.0 (237.0), p<0.001, probably due to femoral cannulation and possible leg ischemia. We found no difference in: cardiac biomarkers, creatinine, glomerular filtration rate need for transfusion or revision, bleeding, length of ICU stay, cerebrovascular accidents, wound infections, pulmonary complications, prolonged ventilation, atrial fibrillation or need for dialysis.

Conclusions: We found no important differences in ICU outcomes when we compared sternotomy and MIVT approaches for mitral valve surgery.

Table 1 (abstract P122). Patient characteristics

P123 Improving interstage mortality in hypoplastic left heart syndrome in an evolving congenital cardiac service in the UAE

V Sheward1, K Ramakrishnan1, L Kiraly2

1Sheikh Khalifa Medical City, PCICU, Abu Dhabi, United Arab Emirates; 2Sheikh Khalifa Medical City, Paediatric Cardiac Surgery, Abu Dhabi, United Arab Emirates

Introduction: The Sheikh Khalifa Medical City (SKMC) Hospital is an 800 bed hospital, which has offered palliative cardiac surgery to babies with hypoplastic left heart syndrome (HLHS) for the last decade. During this time, the service has developed, with increasing patient numbers, increasing use of ECMO by the service, and expansion of the cardiac ICU.

Methods: A retrospective notes review was conducted of all admissions to the cardiac ICU since 2006. Patients with HLHS, who underwent Norwood 1 surgery, were analysed. Population statistics, surgical interventions, interstage events, and interstage mortality were reviewed.

Results: Over the last decade, the number of patients with HLHS who underwent Norwood 1 has increased. Interstage mortality has decreased, and is currently 30-40%. Significant morbidity was not seen at a rate higher than in the international literature. Discharge planning, and community access to allied health professional services remained a concern.

Conclusions: The paediatric congenital cardiac surgical service in the United Arab Emirates is relatively new (compared to some services around the world). Interstage mortality in HLHS is improving as a result of programme development, surgical progress and postoperative care. In the interstage period, there is currently no home monitoring programme in place. Some patients were found to have had very extended hospital admissions. Improved community support may reduce interstage mortality further, as well as improve the social situation of many of these patients.

P124 Five year experience of off-pump surgical revascularization in patients with multi-vessel coronary disease

A Khadzhibaev1, A Abdurakhmanov1, M Obeid1, O Mashrapov1, N Rakhimov1, I Abdukhalimov1, U Ganiev2

1Health Ministry Of The Republic Of Uzbekistan Republic Research Center Of Emergency Medicine Of Uzbekistan, Cardiosurgery, Tashkent, Uzbekistan; 2Health Ministry Of The Republic Of Uzbekistan Republic Research Center Of Emergency Medicine Of Uzbekistan, Cardiology, Tashkent, Uzbekistan

Introduction: Surgical revascularization in patients with multi-vessel coronary artery disease still raises many questions. The aim of the study was to analyze the single center results of off pump CABG in patients with multivessel coronary disease.

Methods: This retrospective study analyzed 1020 patients with history of coronary artery disease who underwent off-pump CABG surgery in 2013-2018 years at the Republican Research Center for Emergency Medicine. The average age of the patients was 58.2±0.95 years. Women were 309 (30.3%) and men - 711(69.7%). Most patients(76.9%) had III Class NYHA, and 12.3% and 10.8% patients IV and II class respectively. 283 patients had concomitant diseases: diabetes mellitus and COPD. The majority of patients were diagnosed with unstable coronary artery disease (94.4 %); 5.6 % of patients underwent surgery based on emergency indications or with acute myocardial infarction.

Results: The mean surgery duration was 210±1.7 minutes. The mean amount of bypasses (bypass index) was 3.2±0.2, in 840 (82.3%) cases LIMA was used as a preferred graft for LAD bypass, in 30 (3%) patients with concomitant carotid and coronary artery disease were performed simultaneous interventions. In 95 (9.3%) cases due to the haemodynamical instability we turned on-pump. Postoperative complications were observed in 112(10.9%) patients. We lined out the prevalence of cardiac complications, such as heart failure and rhythm disturbances, observed in 87 (8.5%) and 53 (5.1%) patients respectively. Hospital mortality rate was 2.8% (29/1020). The cause of mortality in all cases was acute heart failure, due to the initial severity of the disease, and in 11(1.07%) cases an acute myocardial infarction was diagnosed. Duration of postoperative period was 7.8 ± 0.9 days.

Conclusions: Off-pump Coronary artery bypass grafting can be safely performed with relatively low incidence of mortality and postoperative morbidity.

P125 Prognostic value of mid-regional pro-adrenomedullin and mid-regional pro-atrial natriuretic peptide as predictors of multiple organ dysfunction development and ICU length of stay after cardiac surgery with cardiopulmonary bypass in adults

A Goncharov1, D Popov2, M Rybka1

1Bakulev National Medical Research Center of Cardiovascular Surgery, Department of Anaesthesiology, Moscow, Russia; 2Bakulev National Medical Research Center of Cardiovascular Surgery, Laboratory of Clinical Microbiology and Antimicrobial Therapy, Moscow, Russia

Introduction: One of the most harmful complications after cardiac surgery with cardiopulmonary bypass is a syndrome of multiple organ dysfunction (MODS). We consider that mid-regional pro-adrenomedullin (MR-proADM) and mid-regional pro-atrial natriuretic peptide (MR-proANP) plasma concentrations can be used as predictors of MODS development and LOS in ICU.

Methods: Thirty six adult patients (mean age 60 years, 27 male) with cardiovascular diseases undervent cardiac surgery with cardiopulmonary bypass (heart valve(s) replacement – 20 (55.6%) patients, aorta and it`s branch surgery – 14 (38.9%) patients, valvular surgery and coronary artery grafting – 2 (5.5%) patients). NYHA heart failure class II was in 3 (8.3%) patients, III – in 27 (75%) patients, IV – in 6 (16.7%) patients. In the dynamics levels of MR-proADM and MR-proANP were measured in the venous blood with the Kryptor compact plus analyzer (Thermo Fisher Scientific, Germany) before 1 day and on the 1st and 6th days after surgery. All patients were divided into subgroups according to the lengths of stay in the ICU and the development of MOD in the postoperative period. The data are shown as median and 25th and 75th percentiles. The data were compared by Mann-Whitney U-test, p-value of <0.05 was considered statistically significant.

Results: Levels of MR-proANP did not significantly change at the study stages and did not have a significant difference between subgroups. The levels of MR-proADM increased in the first postoperative day and remained elevated for 6 days. This increase was significantly higher in subgroups of increased LOS in ICU and with MODS. The data are shown in the Table 1.

Conclusions: MR-proADM can be used as predictor of MODS and LOS in the ICU for adult patients underwent cardiac surgery with cardiopulmonary bypass.

Table 1 (abstract P125). MODS development and LOS in ICU prediction

P126 Predictive factors of prolonged postoperative ICU stay aftercardiac surgery

F Ampatzidou, R Ioannidis, A Dimaki, N Mihail, G Kechagioglou, G Drossos

G.Papanikolaou General hospital Thessaloniki, Cardiac Surgery ICU, Thessaloniki, Greece

Introduction: Prolonged intensive care unit (ICU) stay after cardiac surgery is associated with increased mortality and cost .The aim of this study was to investigate factors influencing prolonged ICU stay.

Methods: Consecutive patients who underwent cardiac surgery from June 2012 to October 2018 in our Cardiothoracic department, were retrospectively investigated. Group A consisted of pts with prolonged stay defined as more than 3 days and group B the rest of the cohort. The following characteristics and perioperative factors were compared between the groups: smoking, diabetes, COPD, REDO(re-operation), ejection fraction (EF)<50%, emergent procedure, cardiopulmonary bypass time (CPB)>120 min, low cardiac output syndrome (LCOS), acute kidney injury(KDIGO) and mortalityChi square test was used for the statistical analysis.

Results: From a total of 3163 patients who underwent cardiac surgery, 247 pts consisted group A mean aged 66.8±10.1 vs 65.2±10.6 and Euroscore II 5.7±7.8 vs 2.1±3.1.No correlation was found with smoking (p=0.28) and diabetes(p=0.67) Statistical significant factors are shown in Table 1. Mortality was 20.2% for group A and 1.9% for group B(p<0.01).

Conclusions: History of COPD, pre-op EF less than 50 %, redo and emergent procedures, prolonged CPB time, postoperative AKI and LCOS have statistical significant correlation with prolonged ICU stay after cardiac surgery. Mortality in this group of patients is higher.

Table 1 (abstract P126). Statistical significant factors

P127 Identifying coagulation thresholds associated with bleeding complications in pediatric post-cardiotomy extracorporeal membrane oxygenation

L Thalji1, G Schears1, K Bohman1, D Kor1, J Stubbs2, M Nemergut3

1Mayo Clinic, Department of Anesthesiology and Perioperative Medicine, Rochester, United States; 2Mayo Clinic, Department of Laboratory Medicine and Pathology, Rochester, United States; 3Mayo Clinic, Department of Anesthesiology and Perioperative MedicineMedicine, Rochester, United States

Introduction: Hemorrhagic complications of extracorporeal membrane oxygenation (ECMO) pose a major morbidity and mortality. Optimal anticoagulation strategies balancing risks of bleeding and thrombosis in children are poorly understood. We aimed to identify factors associated with non-surgical bleeding in the first 12 ECMO hours.

Methods: We evaluated all pediatric (<18 yrs) post-cardiotomy patients requiring ECMO between Dec 2002–July 2017 stratifying them by presence/absence of surgical bleeding. Non-surgical bleeding was defined as chest tube output >3cc/kg/hr during the first 12-hours not requiring reoperation. Patient characteristics and coagulation parameters at various time points after ECMO initiation were compared between groups, and receiver operator characteristic (ROC) curves were constructed to identify models and thresholds with optimal predictive performance.

Results: Amongst 117 patients, 63 (53.8%) were male, median (IQR) age and weight were 47.4 days (7.3 – 394.2), and 4.0kg (3.0 – 8.5). The 30 patients (25.6%) who experienced non-surgical bleeding were younger (7.3 vs 109.5 days, p<0.001) and smaller (3.3 vs 4.8kg, p<0.001). Operative and ECMO durations, and anticoagulant choices were comparable. Coagulation parameter differences between groups are shown in Table 1. Univariate ROC analysis revealed baseline kaolin thromboelastography R-time to best predict bleeding [OR per 10min increase 1.21 (1.06 - 1.37), p=0.004, AUC 0.76]. This persisted across multivariable models, with adjustment for weight [OR per 1kg decrease 1.70 (1.11 - 2.59), p=0.014] and the presence of hypoplastic left heart syndrome [OR 5.09 (1.21 - 21.32), p=0.026] producing the best predictive performance [OR per 10min R-time increase 1.26 (1.06 - 1.48), p=0.007, AUC 0.89], Figure 1.

Conclusions: Deranged coagulation parameters, particularly kaolin R-time may predict non-operative bleeding in pediatric ECMO patients. These findings may guide therapeutic anticoagulation while avoiding hemorrhagic sequelae in at risk patients.

Table 1 (abstract P127). Baseline post-cardiotomy coagulation parameters following ECMO initiation by bleeding status
Fig. 1 (abstract P127).

Multivariable ROC curve for Kaolin Thromboelastography R-time, adjusting for weight and the presence of hypoplastic left heart syndrome

P128 Impact of technical features of intra-aortic balloon pump placement on device related morbidity and mortality in cardiac surgery patients

R Samalavicius1, L Puodziukaite2, I Norkiene2, I Misiuriene1, P Serpytis3

1II Department of Anaesthesiology, Center of Anaesthesiology, Intensive Care and Pain Management, Vilnius University Hospital Santaros Klinikos, Vilnius, Lithuania; 2Clinic of Anaesthesiology and Intensive Care, Vilnius University, Faculty of Medicine, Vilnius, Lithuania; 3Clinic of Cardiac and Vascular Diseases, Vilnius University, Faculty of Medicine, Vilnius, Lithuania

Introduction: Intra-aortic counterpulsation remains the most frequently used mechanical assist device worldwide. We aimed to evaluate intra-aortic balloon pump (IABP) device related morbidity and mortality in cardiac surgery patients.

Methods: Single centre retrospective review of 402 cardiac surgical patients treated with IABP counterpulsation between 2013 and 2018 in the University Hospital. Patient outcomes, complications and IABP insertion technical characteristics were analysed. According to tip position patients were divided into three groups: acceptable (at T2-T4 vertebrae), malposition (T5-T6) or severe malposition (at T7 or below).

Results: Median patient age was 67[IQR, 59-75] years, 289(71.9%) were male, median EuroSCORE II was 4.55[IQR, 2.5-10.9]%. Indications for IABP treatment were: treatment before surgery in 44(10.9%), prophylactic use in high risk 171(42.5%), hemodynamically unstable patient before CPB in 5(1.3%), during off-pump CABG in 10(2.5%), facilitate weaning from CPB in 42(10.5%) and low cardiac output following surgery in 130(32.3%) patients. 327(86.3%) patients survived to ICU discharge. In hospital mortality was 5.3% in elective IABP placement group vs 26.4% in urgent IABP implementation group (p<0.001). In 23(5.7%) patients due to severe aortoiliac disease IABP was placed intraoperatively through ascending aorta. In remaining 379 patients percutaneous IABP placement was performed, balloon position was acceptable in 134(35.4%), malpositioned in 172(45.4%), severely malpositioned in 64(16.9%) and unavailable for 9(2.4%) cases. In 340(89.7%) patients sheathless technique was used. Vascular complications rate was 15(3.7%) and showed tendency to be lower (2.65% vs 10.25%, p<0.05) when sheathless technique was used.

Conclusions: Elective balloon placement was associated with lower morbidity and mortality. Sheathed insertion showed a tendency to be associated with higher rate of vascular complications. Suboptimal IABP tip position was not associated with higher morbidity or mortality rates.

P129 A cut-off point of cardiac troponin elevation indicating poor prognosis for critically ill patients admitted in an intensive care unit: a retrospective study

P Vasileiou, A Boultadakis, E Tsigou, E Boutzouka, T Katsoulas, G Fildissis

Timiou Stavrou and Noufaron, ICU, Attiki, Greece

Introduction: Elevated cardiac troponin (cTn) level in patients (pts) admitted in the intensive care unit (ICU) is multifactorial and has been associated with a worse prognosis. The aim of the study was to review the frequency and the main cause of cTn elevation and to calculate a discriminating index.

Methods: We retrospectively assessed all pts admitted in our eight-bed general ICU during a 6-month period with at least one measurement of cTn during their ICU stay. We recorded clinical characteristics, the level of cTn on admission, the maximum cTn during ICU stay and the possible causes of elevation. Variables are expressed as mean ± SD or as median and Interquartile Ratio (IR), according to the normality of their distribution. Student´s Ô test or the Mann Whitney U tests were used to compare the group of elevated cTn with the group of normal cTn. The prognostic performance of elevated cTn was evaluated by the Receiver Operating Characteristics (ROC) curve. Statistical analysis was performed using SPSS version 21.0 (SPSS, Inc., Chicago, Illinois).

Results: In 84 out of 92 pts that cTn was measured at least once, abnormal levels (>15.6 pg/ml) were found in 58 (69%) of them, and the maximum cTn value was 633 (1718.25) pg/ml. The clinical characteristics of the pts are depicted in Table 1. Sepsis was the main cause of troponin elevation, which complicated by Acute Kidney Injury (AKI) in 20 pts (34%). Maximum cTn, AKI and the difference of maximum - admission cTn (ÄcTn) differed significantly between pts who survived and pts who died (p=0.029 and 0.001, respectively). The Area Under the Curve (AUC) was 0.753 and the optimal prognostic cut-off value of ÄcTn was 57 pg/ml with a sensitivity of 0.656 and a specificity of 0.833

Conclusions: Raised cardiac troponin values is a frequent finding in ICU pts and sepsis is the driving cause. AKI and the difference between maximum and admission cTn measurements differ significantly between pts who survive and pts who die. An elevation of cTn during ICU hospitalization >57 pg/ml seems to be a threshold indicating poor prognosis regarding both mortality and AKI.

Table 1 (abstract P129). Clinical characteristics of the pts

P130 The prognostic role of NT-pro-BNP in septic patients with elevated troponin T level

A Shilova1, D Shchekochikhin2, M Gilyarov2, A Svet2, M Petrushin3

1Moscow City University Hospital #1 n.a. Pirogov, ICCU, Moscow, Russia; 2Moscow City University Hospital #1 n.a. Pirogov, Cardiology, Moscow, Russia; 3Tver Regional Hospital, ICU, Tver, Russia

Introduction: Sepsis is frequently accompanied with release of cardiac troponin T (TnT) and NT-pro-BNP, but the clinical significance of this myocardial injury and cardiac dysfunction remains unclear [1]. TnT is known to be an independent predictor of mortality, whereas the prognostic role of NT-pro-BNP is uncertain.

Methods: We have prospectively enrolled 67 patients with sepsis in our intensive care unit from June the 1st 2017 to June the 1st 2018. Subjects with a clinically apparent cause of troponin release were excluded. High-sensitivity cardiac troponin I (hs-cTnI), NT-pro-BNP, CRP and PCT and lactate concentration in plasma were measured at admission

Results: Mean SOFA on admission was 6,1 points. 38.8% of patients were women. Mean age ranged from 52 ±7.8 years in men and 54.38±9.2 years in women. 37.3% of patients were admitted in shock. 70.15% patients underwent noncardiac surgery. 30 days mortality rate was 62.7% (42 pts). Average TnT 0.21±0.09 ng/dl. Average Nt-proBNP 238.8±72.4 ng/ml. TnT and NT-pro-BNP levels on admission were significantly lower in survived patients. 0.11 vs 0.28 ng/ml, p=0.007 for TnT and 194.4 pg/ml vs 272.3 pg/ml, p= 0.05 for NT-pro-BNP. There was significant correlation between TnT level and PCT, APACHE-II, SOFA, lactate and NT-pro-BNP values (p < 0.05). We’ve performed stepwise multiple regression model, that included age, PCT, CRP, APACHE-II, SOFA, lactate, TnT and NT-pro-BNP. Elevation of NT-pro-BNP was founded to be the only independent predictor of mortality in septic patients with elevated HR 1.6 [1.15; 2.24], p= 0.013.

Conclusions: Elevated NT-pro-BNP is an independent predictor of unfavorable prognosis in septic patients with elevated TnT level.


1. Fei Wang et al. Crit Care. 2012; 16(3): R74.

P131 The incidence and risk factors of acute kidney injury after left ventricular asisst device implantation

A Ozdemirkan, H Sahinturk, F Atar, O Ersoy, P Zeyneloglu, A Pirat

Baskent University Faculty of Medicine, Anesthesiology and Critical Care, Ankara, Turkey

Introduction: Cardiac surgery-associated acute kidney injury (AKI) is a serious complication of cardiac surgery with an incidence of 5-42%. The aim of the study was to evaluate AKI in the early postoperative period after left ventricular assist device (LVAD) surgery and to compare patients with and without AKI to determine incidence, risk factors and clinical outcomes.

Methods: In this retrospective cohort study, the medical records of all patients aged between 18-75 years undergoing LVAD implantation from January 2011 to December 2016 was reviewed. AKI was defined according to KDIGO criteria.

Results: Out of 57 patients, 10 (18%) were female and the mean age of the cohort was 44.6±16.1 years. Thirty six patients (63%) developed AKI following LVAD implantation. Duration of cardiopulmonary bypass was longer in patients with AKI (162.5±58.2 vs 128.7±48.6, p=0.039). Patients with AKI had lower mean blood pressures (54.9±13.1 vs 62.8±6.9, p=0.013), higher frequency of norepinephrine usage [16 (44%) vs 3(14%), p=0.020), and higher cumulative fluid balance on the postoperative first day (1.4±0.6 L vs 1.2±0.9L, p= 0.042). The frequency of prolonged mechanical ventilation among patients with AKI was higher when compared to those without AKI [21 (58%) vs 5 (24%), p= 0.012]. Logistic regression analysis revealed duration of cardiopulmonary bypass, mean blood pressure and cumulative fluid balance on the postoperative first day as independent risk factors for AKI (OR:1.013, CI 95% 1.000-1.025, p:0.045; OR: 0.929, CI 95% 0.873-0.989, p:0.021; OR: 1.001, CI 95% 1.000-1.001 respectively). Hospital mortality (58% vs 24%, p=0.012) and 30-day mortality (39 % vs 5%, p= 0.005) were significantly higher in patients who had AKI.

Conclusions: AKI develops in 63% of patients after LVAD surgery. Risk factors for occurance of AKI includes longer duration of cardiopulmonary bypass, lower mean blood pressures and higher cumulative fluid balance on the postoperative first day.

P132 Early inotropic therapy in patients after venoarterial extracorporeal membrane oxygenation implantation - a single center experience

V Zotzmann, C Lang, T Wengenmayer, D Dürschmied, C Bode, D Staudacher

University Heart Center, Cardiology, Freiburg, Germany

Introduction: Venoarterial extracorporeal membrane oxygenation (VA-ECMO) increases afterload. This can negatively impact left ventricular (LV) output in case of decreased LV-function resulting in increased filling pressures, pulmonary edema, LV-distension or even stasis in the pulmonary circulation. If positive inotropic agents should be used is discussed controversial in literature.

Methods: Here, we report data of VA-ECMO-patients, treated with dobutamine, levosimendan, suprarenin or no inotropic agens, in respect of 30-day survival. All data were collected retrospectively (10/2010 to 10/2018) at a single center, all patients with a survival below 24 hours were excluded. While treatment of VA-ECMO patients is strongly guided by standard operation procedures at our institution, no recommendation on positive inotropic therapy could be made.

Results: A total of 232 VA-ECMO patients were evaluated, of which 47 patients were treated with levosimendan within 48 hours after cannulation. 30day survival in the whole cohort was 41.9%. A total of 99 patients did not receive any positive inotropic therapy at 24 hours after implantation (survival 48.0%). Survival was best in the levosimendan plus dobutamine group 50%, followed by dobutamine mono-therapy 46.2% and levosimendan mono 41.2%. Survival with suprarenin mono was 33.3%, suprarenin plus levosimendan 25.0% and suprarenin plus dobutamine 18,8%. Pooling data, we found no evidence that levosimendan and/or dobutamine (survival 45.8%, n=83, p=0.882) improves survival over no inotropic therapy (Fig 1). Therapy with any combination including suprarenin however resulted in poor survival (27.7%, n=65, p=0.009). Adjustment for lactate levels or eCPR did not change the results.

Conclusions: This retrospective analysis of 232 VA-ECMO patients shows no evidence that early inotropic therapy improves outcomes in VA-ECMO patients. This conclusion is obviously biased by retrospective design. Until randomized data are available, suprarenin however should be avoided.

Fig. 1 (abstract P132).

30 day survival after va-ECMO implantation

P133 Survey of non-resuscitation fluids in septic shock

A Linden-sonderso1, P Bentzer1, M Spångfors2, J Undén3, T Kander4, D Griesdale5, J Boyd6, M Jungner4

1Helsingborg Hospital, Department of Anesthesiology and Intensive Care, Helsingborg, Sweden; 2Lund University, Kristianstad Hospital, Department of Clinical Sciences Lund, Anesthesiology and Intensive Care Medicine, Kristianstad, Sweden; 3Hallands Hospital, Department of Operation and Intensive Care, Halmstad, Sweden; 4Skåne University Hospital, Department of Intensive and Perioperative Care, Lund, Sweden; 5Vancouver General Hospital, British Columbia, Department of Anesthesiology, Pharmacology and Therapeutics, Vancouver, Canada; 6St. Paul´s Hospital, British Columbia, Department of Intensive Care, Vancouver, Canada

Introduction: Positive fluid balance is associated with poor outcome in septic shock. The objective of the present study was to characterize non-resuscitation fluids in early septic shock.

Methods: Consecutive patients >18 years of age were screened for inclusion criteria during a 4-month period in 8 ICUs in Sweden and in Canada. Inclusion criteria were septic shock per SEPSIS-3 definition within 24 hrs of ICU admission. A maximum of 30 patients per center were included. Type, indication and volume of non-resuscitation fluids were recorded during the first 5 days of admission. Fluids other than colloids, blood products and crystalloids given at rate > 5ml/kg/h were considered to be non-resuscitation fluids. The study was registered on (NCT03438097). Data are presented as median (interquartile range).

Results: A total of 201 patients were included between March 1st and June 30th 2018 (see Table 1 for demographics). Patients received 7886 (4051-12670) milliliters (ml) of non-resuscitation fluids and 3450 (2000-5625) ml of resuscitation fluids during the observation period. Non-resuscitation fluids consisted of vehicle for drugs 2418 (1289-4655) ml, glucose 1485 (0-3124) ml, crystalloids at a rate of < 5 ml/kg/h 600 (0-2000) ml, enteral nutrition 334 (0-1609) ml, enteral water 270 (0-995) ml and parenteral nutrition 0 (0-0) ml. Vehicles were mainly used for administration of vasoactive drugs and antibiotics, and total volumes were 455 (185-900) ml and 560 (125-1100) ml, respectively. Daily volume and type of fluid are presented in Figure 1. At day 2 and onwards non-reuscitation fluids represented the major part of total daily volume (Figure 1).

Conclusions: These recent data suggests that non-resuscitation fluids is the major source of fluids in septic shock already at day 2 after admission to the ICU. Strategies aimed at reducing this source of fluids may reduce positive fluid balance.

Table 1 (abstract P133). Patient characteristics
Fig. 1 (abstract P133).

Median daily volume and type of fluids. Data for type of fluid each day are presented as percent of total daily volume. Please note that sum of the median daily volume does not equal sums of median non-resucitation and resuscitation fluids over the whole observation period as described in results because of the skewed distribution of data. Number of patients are (n)

P134 Effect of targeted or standard protocols of intra-operative fluid therapy on early graft functions during live related renal transplantation- a randomized comparative trial

R Ramachandran1, A Singh1, V Rewari1, V Bansal2, B Ray1, A Trikha1

1All India Institute of Medical Sciences, Anaesthesiology, Pain Medicine and Critical Care, New Delhi, India; 2All India Institute of Medical Sciences, Surgical Disciplines, New Delhi, India

Introduction: Aim was to compare different fluid hydration regimens, standard fluid replacement method and targeted fluid replacement before declamping of anastomotic renal vessels, on early graft function during live related renal transplantation (LRRT). Primary outcome of study was serum creatinine (creat) on first postoperative day

Methods: 40 consenting patients undergoing LRRT were randomized. In Group S- patients received intravenous fluid according to standard guidelines for intraoperative fluid infusion using Holliday-Segar rule. Patients in Group T- received intravenous fluids at start of vascular anastomosis to achieve a CVP of > 15 mm Hg at de-clamping of vessels. Serum creat, blood urea, serum electrolytes and urine output were measured in postoperative period. Hemodynamic parameters including blood pressure, CVP, cardiac output and cardiac index were compared in both groups

Results: Difference in serum creat on first post-operative day, 2.04 (0.73) and 2.31 (1.19) [Mean (SD)] in Group T and S, respectively, were statistically insignificant. Total urine output on first post-operative day was significantly more in Group T as compared to Group S (P value-0.03). Graft diuresis time and graft turgidity after vascular de-clamping were significantly better in Group T. Total volume of crystalloids infused during intra-operative period was significantly less in Group T (P value-<0.001)

Conclusions: Serum creat on first post-operative day was similar in patients receiving fluid according either to standard practice or according to focused infusion prior to de-clamping of vascular anastomosis despite intervention group receiving significantly less intravenous fluid intraoperatively. Graft turgidity score and diuresis time were significantly better with focused infusion

P135 A retrospective analysis of intravenous (IV) fluid and medication administration in a United Kingdom district general Intensive Care Unit (ICU)

B Scrace, L Squire, D Ashton-Cleary, L Moore, M Spivey

Royal Cornwall Hospitals Trust, Critical Care, Truro, United Kingdom

Introduction: We aimed to ascertain the extent and make-up of fluid overload in critically ill patients and to identify whether delivery of more concentrated medications could reduce this. Positive fluid balance is associated with increased mortality [1]. A recent study has shown that the predominant component of fluid overload was from IV medications and maintenance fluid [2].

Methods: We reviewed 20 sequential patients admitted to our ICU with an APACHE II score of greater than 15 and a length of stay (LOS) greater than 72 hours. The patients’ electronic admission summary was interrogated to establish: length of stay (LOS) fluid balance at 72 hours, total volume administered as IV medications, total volume administered as maintenance fluid and total fluid administered as bolus fluid. Predicted reduction in volume resulting from administration of the most concentrated format of IV medications received in the first 72 hours LOS was calculated for each patient. The percentage reduction in total IV fluid administration was then calculated.

Results: 90% of patients had a positive fluid balance at 72 hours LOS. Median fluid balance at 72 hours was positive 4170.5 mls. IV medications (median 2423 mls) and maintenance fluid (median 3362.5 mls) represented a greater proportion of intravenous fluid administration than bolus fluid (median 1900 mls). Use of the most concentrated format of IV medications gave a predicted median volume reduction of 687.75 mls and a median percentage reduction in total IV administration of 11%.

Conclusions: The proportion of patients with a positive fluid balance and the breakdown of IV fluid administration reflect that seen in recent studies [2]. By administering medications in the highest concentration possible, reductions in fluid administration can be made. A shift away from using maintenance IV fluid in the acute phase of critical illness may lead to reductions in fluid overload.


1. Rhodes A et al: Intensive Care Med; 43:304–377. 2017

2. Silversides et al: Crit Care Med; 46:1600–1607. 2018

P136 Impact of 10% intraoperative fluid overload on patients undergoing major abdominal surgery & admitted to general SICU

S Kongsayreepong, A Piriyapatsom

Siriaj Hospital, Mahidol University, Anesthesiology, Bangkok, Thailand

Introduction: Fluid overload(>10% of admission BW) has been associated with increased complications & organ dysfunction [1]. The aim of this study was to study the incidence &effects of intraop fluid overloads [IFO] on the incidence of prolonged ICU stays (>72 hr); 90 day mortality&periop complications in patients undergoing major abdominal surgery

Methods: A prospective, observational study was done on1,680 patients admitted to the general SICU from Jan 2014 - Dec 2017 after major abdominal surgery. study data included patients’demographic data; admitted & preop serum alb, Hb & creatinine up to 72 hrs; ASA physical status; surgery type & urgency; anesthesia type & duration; type & amount of intake & output; perioperative complications; septic shock on ICU admission; ventilator days, ICU & hospital LOS, APACHE II score on postop day 1, 90 day mortality.

Results: Incidence of IFO was 15.2%. Patients with IFO had significantly longer anesthetic times, lower preop & admitted serum alb; higher preop Hb, fluid balances; longer ICU & hospital LOS. Significantly higher combined general & neuraxial block & septic shock on ICU admission were seen in the IFO pts. Significantly higher periop complications (e.g. CHF, serious cardiac arrhythmias, intraabdominal hypertension, wound infection, anastomosis leakage) were found in IFO patients esp in septic shock patients. AKI was the most organ dysfunction (28.8%) found in IFO & 17% needed RRT from fluid overload. IFO was a significant predictor of prolonged ICU stay (OR 9.78; 95% CI 3.28 -10.9, p<0.001) & who had intraop fluid balances>130 mL/kg were significantly associated with higher 90-day mortality.

Conclusions: IFO had a high impact in critically ill patients undergoing major noncardiac surgery in terms of prolonged ICU stay, increasing periop major complications, and mortality. Care should be taken to prevent IFO.


1. Glatz T, BMC Surg 2017;17:6

P137 Quantification of fluid needed for burns resuscitation in children

H Bangalore1, S Ray2, N Martin3

1Great Ormond Street Hospital, Burns Intensive Care, London, United Kingdom; 2Great Ormond Street Hospital, Paediatric Intensive Care, London, United Kingdom; 3St Andrew´s Burns Center, Chelmsford, United Kingdom

Introduction: In children less than 30 kilograms, maintenance fluids are routinely added to the resuscitation requirements calculated using Parkland’s or other formulae. The contribution of this component for fluid resuscitation in children can add a significant quantity to total estimated fluid requirements. For example, in a child who is 10 kilograms with a 25% burn, the maintenance fluid requirement is 1000 mls per 24 hours and the resuscitation component per Parkland’s will be 4 x 10 x 25%=1000 mls. Hence, the maintenance requirement can exceed the resuscitation requirement in this child if the burn surface area is less than a 25 % burn. The contribution of maintenance fluids to the total fluid requirements in small children with thermal injuries is under-recognised and not frequently studied.

Methods: To understand the contribution of maintenance fluids to the total fluid requirements in children less than 30 kilograms who need resuscitation for thermal injuries of different sizes, we numerically simulated 1. children who had similar weights but different burn sizes and 2. Children with similar burn size but different weights. The results are as shown in Figure 1 and 2.

Results: 1. The contribution of maintenance fluids is greatest in smaller children with a smaller burn. 2. In bigger children and larger burns, as the total burn fluid requirements are higher, the proportion contributed by maintenance fluids to the total fluid requirements is smaller. 3. Decreasing the quantity of maintenance fluids (like in other areas of intensive care) may help decrease fluid overload, particularly in smaller children and smaller burns

Conclusions: Understanding of the contribution of maintenance to the total fluid requirements in children is important for accurate fluid management.

Fig. 1 (abstract P137).

Total fluid requirements for different Burn sizes

Fig. 2 (abstract P137).

Total fluid requirements for different weights

P138 Quantification of fluid needed for burns resuscitation in children

H Bangalore1, S Ray2, N Martin3

1Great Ormond Street Hospital, Burns Intensive Care, London, United Kingdom; 2Great Ormond Street Hospital, Paediatric Intensive Care Uniti, London, United Kingdom; 3St Andrews Burns Center, Burns and reconstructive surgery, Chelmsford, United Kingdom

Introduction: Accurate quantification of fluid in resuscitation of thermal injuries is important for benchmarking, comparing and improving outcomes. In adults, it is usually expressed as mls/kg/%TBSA. In children, maintenance fluids are added to the resuscitation requirements. This is kept constant and the resuscitation component is titrated to meet pre-defined end points—usually urine output. Maintenance fluids are not uniformly stratified across the weight ranges. We propose that quantification of fluids in mls/kg/%TBSA in children does not accurately capture fluid needs for resuscitation due to the maintenance component of the fluid requirement.

Methods: We conducted this retrospective study in children admitted to a single-center Burns Intensive Care Unit (BICU) between January 2010 and December 2014. Children ≤30 kilograms with TBSA ≥15% admitted within 8 hours of their injury were included. OE (Observed to expected ratio) and fluid in mls/kg/% TBSA were calculated as shown in Figure 1.

Results: There were 33 children in the cohort with half requiring invasive mechanical ventilation in the BITU and nearly a quarter requiring inotropic support. The demographic details are as shown in Table 1. The OE ratio at the end of 24 hours in the cohort was 1.01 (0.68-1.36). The total fluid given was 6.9 (6.1, 8) mls/kg/ % TBSA. The Titrated resuscitation component was 4.2 (3.7, 5.6) mls/kg/TBSA. Total fluid (which included the maintenance fluid) had a poor correlation with OE ratio R2=0.34 (Fig 2). Exclusion of the maintenance fluid had a better correlation with the OE ratio R2=0.75

Conclusions: To capture differences in the titratable resuscitation component rather than differences in the maintenance requirements, fluid should be quantified in children by excluding the maintenance component when expressed as mls/kg/%TBSA.

Table 1 (abstract P138). Patient details
Fig. 1 (abstract P138).

Fluid Calculations

Fig. 2 (abstract P138).

Correlation of OE ratio and fluids-total & titrated resuscitation component

P139 Dynamic arterial elastance for predicting mean arterial pressure responsiveness after fluid challenges in acute respiratory distress syndrome patients

P Luetrakool1, S Morakul1, V Tangsujaritvijit2, C Pisitsak1

1Ramathibodi, Department of anesthesiology, Bangkok, Thailand; 2Ramathibodi, Department of medicine, Bangkok, Thailand

Introduction: Dynamic arterial elastance (Eadyn; pulse pressure variation/stroke volume variation; PPV/SVV) is a dynamic parameter of arterial load that can be continuously monitored. Previous study proposed that Eadyn was able to predict mean arterial pressure (MAP) responsiveness after fluid challenge [1-5]. The objective of this study was to assess whether the Eadyn was able to predict MAP responsiveness in acute respiratory distress syndrome (ARDS) patients ventilated with low tidal volume.

Methods: We performed a prospective study of diagnostic test accuracy in adult ARDS patients with acute circulatory failure and fluid responsiveness. All patients are continuously monitored blood pressure via arterial line connected with Flotrac® transducer and Vigileo® monitor. Once the attending physicians decided to load intravenous fluid, we recorded PPV/SVV and also other hemodynamic parameters before and after fluid bolus. MAP responsiveness was defined as an increase in MAP ≥ 10% from baseline after fluid challenge.

Results: Twenty-three events were included. Nine events (39.13%) were MAP-responsive. Cardiac output, heart rate and stroke volume were similar in both MAP-responder and MAP-nonresponder group. Baseline MAP, diastolic blood pressure (DBP) and pulse pressure (PP) were significantly different after fluid challenge in MAP-responder group. Eadyn of preinfusion phase was failed to predict MAP responsiveness after fluid challenge (area under the curve = 0.603, 95%CI = 0.38-0.798).

Conclusions: One of the arterial load parameters such as Eadyn derived from non-calibrated pulse contour analysis method was unable to predict MAP responsiveness in ARDS patients with low tidal volume ventilation.


1. Monge Garcia MI, et al. Crit Care. 2011;15(1):R15.

2. Garcia MI et al. Crit Care. 2014;18(6):626.

3. Cecconi M et al. Anesth Analg. 2015;120(1):76-84.

4. Lanchon R et al. Anaesth Crit Care Pain Med. 2017;36(6):377-82.

5. Wu CYet al. Eur J Anaesthesiol. 2016;33(9):645-52.

P140 The change of central venous pressure and mean systemic filling pressure during a fluid challenge in patients with septic shock

I Cisterna, R Barelli, T Taccheri, A Dell´Anna, M Antonelli

Catholic University of the Sacred Heart, Rome, Italy

Introduction: Fluids administration remains a cornerstone treatment for patient in shock state. Increase in Mean Systemic Filling Pressure (Pmsf) with minimal variation of Central Venous Pressure (CVP), should follow volume load in fluid responders (FR). Our aim is to test the hypothesis that in FR septic shock patients, fluid load will determine a significant increase in Pmsf but not in CVP.

Methods: We prospectively included all mechanically ventilated patients with diagnosis of septic shock with invasive hemodynamic monitoring (transpulmonary thermodilution VolumeView-EV1000 Edwards©). We collected hemodynamic and metabolic data and Pmsf with the inspiratory holds technique, before and after a fluid challenge (FC) of 500 ml of ringer lactate in 10 minutes). FR was defined as an increase in cardiac output (CO)>15%.

Results: 13 measures were obtained in 11 patients. In 8 case we observed FR. We found a significant increase in Pmsf after a FC (mean difference(md) 16.9±21.5 mmHg, p=.015). CVP increased significantly (md 2.5±2.4 mmHg, p=.002). Pmsf increased significantly in non-FR (md 27±10mmHg, p=.013) but not in FR while CVP was higher after FC only in FR (md 2.4±2.1mmHg, p=.008). Venous return gradient (Pmsf-CVP) globally increased after FC (md 14±22 mmHg, p=.03), but only in non-FR such increase was significant (md 24±12 mmHg, p=.03). No correlation was found between the variation CO and venous return gradient. We did not find any improvement in metabolic parameters after the fluid challenge.

Conclusions: Pmsf and combined CVP variations do not correlate with FR in our cohort of septic shock patients. Inspiratory holds may not be adequate to infer Pmsf in such context. Further studies are warranted to investigate the effect of FC on Pmsf in this field.

P141 Evaluation of pre-load dependence over time in patients with septic shock

I Douglas1, P Alapat2, K Corl3, M Exline4, L Forni5, A Holder6, D Kaufman7, A Khan8, M Levy3, G Martin9, J Sahatjian10, W Self9, E Seeley9, J Weingarten9, M Williams9, C Winterbottom11, D Hansell12

1Denver Health Medical Center, Denver, Colorado, United States; 2Ben Taub Hospital, Houston, United States; 3Rhode Island Hospital, Providence, United States; 4Ohio State University, Columbus, United States; 5Royal Surrey Hospital, Guilford, United States; 6Emory University, Atlanta, United States; 7NYU Langone Medical Center, New York, United States; 8Oregon Health and Sciences University, Portland, United States; 9Indiana University, Indianapolis, United States; 10Cheetah Medical, Newton Center, United States; 11Bridgeport Hospital, Bridgeport, United States; 12Massachusetts General Hospital, Boston, United States

Introduction: Cardiac function is known to be negatively impacted by sepsis. Stroke volume (SV) change in response to Passive Leg Raise (PLR) is an effective method to predict fluid responsiveness (FR) or cardiac response to preload expansion. We have previously shown that fluid responsiveness is a dynamic state, changing frequently over a 72 hour monitoring period.

Methods: FRESH is a currently enrolling prospective randomized controlled study, evaluating the incidence of FR and patient centered outcomes in critically ill patients with sepsis or septic shock (NCT02837731). Patients randomized to PLR guided resuscitation were evaluated every 6-12 hours over the first 72 hours of care and classified as FR if the SV increased > 10% when measured with non-invasive bioreactance (Starling SV, Cheetah Medical). The time of first FR was noted.

Results: A total of 608 PLR assessments were performed in 86 patients over a 72 hour monitoring period. 56 % were female, and the average age was 61 years. PLRs were evaluated over time, with time 0 representing initial fluid resuscitation (Figure 1). When individual subjects were evaluated over time, 100% of subjects who became FR only after 24 hours showed evidence of LV/RV dysfunction (Figure 2).

Conclusions: Fluid responsiveness or preload dependence frequently changes for septic shock patients over the first 72 hours of care. Evidence suggests it is beneficial to periodically perform an assessment of preload responsiveness to guide fluid administration, as preload dependence is a dynamic and changing state. Preload dependence provides additional information beyond fluid responsiveness. Those patients who remain primarily fluid non-responsive (preload independent) are more likely to demonstrate ECHO confirmed LV/RV dysfunction, as the delay in return to cardiac function may be related to underlying cardiac deficits. Further evaluation may be indicated in preload independent patients.

Fig. 1 (abstract P141).

Fluid Responsive PLR Time Series

Fig. 2 (abstract P141).

Time of Initial Fluid Responsive PLR by Subject

P142 Early 20% vs 5 % albumin in addition to crystalloid resuscitation in surgical septic shock admitting to the general SICU

S Kongsayreepong, S Nonthiaraj

Siriaj Hospital, Mahidol University, Anesthesiology, Bangkok, Thailand

Introduction: Fluid resuscitation is one the main management in patients with septic shock. Large amount of fluid may be needed in surgical septic shock patients. The aim of this study was to compare the effect of 20% vs 5% albumin in addition to crystalloid solution in the early phase of septic shock resuscitation in surgical septic shock patient undergoing major surgery for source control on perioperative outcome.

Methods: This prospective observation study was done in consecutive surgical septic shock patients (age > 18 yrs) undergoing surgery for source control who received only crystalloid solution, 5% albumin or 20 % albumin in addition in crystalloid solution in the early period of shock resuscitation (12 hours after shock detection) and admitting to the general surgical ICU during October 2016-September 2018. Study data included patient demographic data, co-morbidities, ASA physical status, diagnosis, type and duration of operation, Type & amount of fluid used, fluid intake/output in the first 72 hrs, time to reverse shock, incidence of early postoperative complications (AKI, ARDS, intraabdominal hypertension), shock reversal time, ICU & hospital LOS), 28 & 90 day mortality

Results: There were no statistical significant different in the demographic data & comorbidity in the 3 groups (180 patients). Significant more patients in the 2016 received only crystalloid resuscitation. No allergy was found in the albumin groups. Patients who received co-administration of albumin solution had significant less shock reversal time, fluid balance, incidence of AKI, ARDS, intraabdominal hypertension, ICU length of stay & 28 days mortality. Patient who receive 20% albumin had significant less shock reversal time, fluid balanced, abdominal hypertension & ICU length of stay.

Conclusions: Hyperoncotic 20% albumin could be used as a safe additional resuscitation fluid to crystalloid solution in surgical septic shock patient with less shock reversal time & less complications from positive fluid balance.

P143 Lack of association between low cumulative dose of hydroxyethyl starch and acute kidney injury in patients with acute ischemic stroke

SH Park, TJ Kim, HB Jeong, SB Ko

Seoul National University Hospital, Department of Neurology and Critical Care, Seoul, South Korea

Introduction: Hydroxyethyl Starch (HES), a synthetic colloid, has been used as a volume expander, and is associated with renal impairment in patients with sepsis. However, a small dose of HES (6%, 130/0.4) has sometimes been used in acute ischemic stroke. Therefore, we investigated whether a small dose of HES was linked with renal deterioration in patients with acute ischemic stroke.

Methods: A consecutive 524 patients with acute ischemic stroke within 7 days from onset were included between January 2012 and May 2016 (Fig 1). We collected admission serum creatinine (SCr), estimated glomerular filtration rate (eGFR), and renal function was assessed using KDIGO definition of acute kidney injury on hospital days 5 to 9 as to patient’s hospitalization period. Proportion of patients with good functional outcome [(mRS) 0-2] at 90 days were compared between HES group and controls.

Results: Among the included patients (mean age, 68.6; male, 56.5 %), 81 patients (15.5 %) were treated with HES (median cumulative dose, 1450 mL). Initial SCr was lower (0.87 ± 0.43 vs. 1.15 ± 1.15, P < 0.001) and initial eGFR was higher (86.91 ± 24.27 vs. 74.55± 29.58, P < 0.001) in HES group compared with controls (Table 1, 2). The rate of acute kidney injury (AKI) was not different between HES group and control (4.9 % vs. 8.1 %, P = 0.320) (Fig 2). Moreover, use of HES decreased followed-up SCr (0.81 ± 0.47, P= 0.001) and increased followed-up eGFR (97.02 ± 27.90, P < 0.001) similar to control group (SCr, 1.05 ± 0.97, P = 0.001; eGFR, 80.35 ± 32.91, P < 0.001). HES treatment did not lead to better functional outcome after 3 months of acute ischemic stroke (58.5% vs.64.3%, P = 0.279).

Conclusions: HES did not negatively affect renal outcome in patients with acute ischemic stroke if HES was used with a small dose.


1. Thanvi B et al. Postgrad Med J. 2008; 84:412-7.

2. van der Jagt M. Crit Care. 2016 May 31;20(1):126

Table 1 (abstract P143). Baseline characteristics of study population
Table 2 (abstract P143). Differentiation of initial and follow up laboratory finding
Fig. 1 (abstract P143).

Flow diagram of patients selection

Fig. 2 (abstract P143).

Baseline renal function and alteration of renal function in HES treatment (N= 81)

P144 Dynamic in vivo assessment of mean systemic filling pressure: clinical application

F Policastro1, M Pelli1, RA De Blasi2

1Sant´Andrea Hospital - Sapienza University of Rome, Scuola di Specializzazione in Anestesia e Rianimazione, Roma, Italy; 2Sant´Andrea Hospital - Sapienza University of Rome, Dipartimento di Scienze Medico-Chirurgiche e Medicina Traslazionale/Unità di Terapia Intensiva, Roma, Italy

Introduction: Venous return and systemic perfusion are determined by pressure and volumes within the Venular Network. Mean Systemic Filling Pressure (MSFP) is crucial for venous return and volaemic status, and as such it is a useful parameter in physiology and clinical settings alike. We tested whether: Near Infra-Red Spectroscopy (NIRS) could be effective at measuring MSFP both in healthy individuals and in conditions with a rise in interstitial pressures; after an occlusion pressure is relieved, the decrease in venular blood volume could allow calculation of τ (time constant) and thus Venous Resistances (Rv). In order to verify these hypotheses we used a forearm NIRS probe on healthy individuals at rest and during different degrees of Maximal Voluntary Contraction (MVC).

Methods: 10 healthy subjects volunteered in the study that took place at Sant’Andrea Hospital in Rome (Italy). All subjects had venular pressures and volumes assessed via a NIRS probe positioned on the forearm using a pressure-cuff in steps of 5 mmHg from 0 to 50 mmHg, at rest and at 10% and 20% MVC. For each patient MSFP, unstressed Volume (Vu) and stressed Volume (Vs) were measured. A temporary 30 mmHg occlusion was obtained and volume time course was calculated upon release, to derive τ .

Results: P-V relationship was found to have a 3-slopes shape reflecting venular network changes. We measured Vu, Vs, and obtained MSFP values of 6.74 ± 1.83 mmHg, p<0.001; during exercise no changes in Vu and Vs were noted but MSFP values rose; value was found to be 2.9 ± 0.1 sec at rest and 0.9 ± 0.03 sec after exercise, reflecting a reduction in Rv.

Conclusions: NIRS measurements on healthy subject may have implications in the clinical assessment of critical care patients where changes in interstitial pressure are possible.


1. De Blasi RA et al. Microcirculation 21:606-614, 2014

P145 Diagnosis of hepatic arterial flow disorders in patients after abdominal surgery with development of multiple organ dysfunction syndrome

SA Tachyla1, AL Lipnitski1, AV Marachkou1, VA Livinskaya2

1Mogilev Regional Hospital, Department of Anesthesiology and Intensive Care, Mogilev, Belarus; 2Belarusian-Russian University, Department “Economics and Management”, Mogilev, Belarus

Introduction: In the pathogenesis of multiple organ dysfunction syndrome (MODS) important role plays the development of hepatic dysfunction. A known method for assessing hepatic blood flow is reohepatography (RHG). However, it requires the analysis of a large number of parameters of the rheogram curve.

The aim of this study was to develop a method for assessing arterial hepatic blood flow based on the RHG in patients with MODS after abdominal surgery.

Methods: 55 patients in the department of anesthesiology and intensive care unit were included in a prospective study (36 men and 19 women, age 60.1 ± 16.1 years, weight 78.5 ± 13.9 kg.). All patients were divided into two groups: group 1 - patients after orthopedic and trauma surgery (n = 28), group 2 - patients after abdominal surgery with MODS (n = 27). Patients in the groups did not have statistical differences by sex, age, body weight, height. RHG was carried out using the “Reo-Spectr” (Russian Federation).

Results: We have compared the RHG indicators between the groups (Table 1). We have developed a method for assessing hepatic arterial blood flow, which consists in determining the area under the arterial part of RHG curve using the Simpson’s rule. Its normal values range from 43.2 mΩ *s to 49.0 mΩ *s. The method is non-invasive, can be applied at the patient´s bed. Its advantage is simplicity, it can be used for rapid diagnosis and monitoring the effectiveness of treatment. Area under the RHG curve in the group 1 were 46.1 ± 8.1 mΩ *s and 20.0 ± 16.8 mΩ *s in the group 2 (p <0.05).

Conclusions: Patients after abdominal surgery with MODS have impaired hepatic blood flow, which may be associated with liver pathology caused by main surgical disease (obstructive jaundice) and hemodynamic disorders caused by acute cardiovascular failure. The method we developed allows us to determine disorders of hepatic arterial blood flow in the early stages before signs of liver dysfunction appear.

Table 1 (abstract P145). Comparison of RHG indicators between groups, * – p<0.05

P146 Comparison of pulse oximetry hemoglobin with laboratory measurement of arterial and central-venous hemoglobin: the prospective CLIMATE-II study

W Huber, T Lahmer, U Mayr, G Batres-Baires, S Rasch, R Schmid, L Offman

II. Medizinische Klinik, Station 2/11, Munich, Germany

Introduction: Non-invasive pulse oximetry hb-monitoring (SpHb) has been validated predominatly in the peri-operative setting and with controversial results. Monitoring of SpHb in parallel with arterial and central-venous hb-monitoring is part of the CLIMATE-database. The CLIMATE-studies investigate clinical examination, arterial and central-venous blood gas analysis, circulating blood volume, perfusion index, SpHb (Radical-7; MASIMO; USA) and transpulmonary thermodilution. The CLIMATE-II study comapred SpHb with arterial (Hb_art) and central-venous Hb (HB_cv).

Methods: A total of 240 datasets were recorded within 24h in 30 patients of a general ICU (8 datasets per patient). Primary endpoint: Accuracy, precision and trending capacities of SpHb compared to Hb_art and Hb_cv (RapidPoint; Siemens; Germany). Statistics: IBM SPSS 25.

Results: Age 65+/-12 years; mechanical ventilation 240/240, vasopressors 156/240 (65%) of measurements.

SpHb was poorly correlated with Hb_art (r=0.306; p<0.001) and Hb_cv (r=0.312; p<0.001), while Hb_art and Hb_cv were strongly correlated (r=0.949; p<0.001). SpHb (9.60+/-1.11g/dL) overestimated Hb_art (8.83+/-1.64g/dL; p<0.001) and Hb_cv (9.07+/-1.69g/dL; p<0.001). This resulted in bias, lower (LLOA) and upper (ULOA) limits of agreement of 0.77, -2.17 and 3.71g/dl for SpHb vs. Hb_art and 0.53, -2.47 and 3.53g/dL for SpHb vs. Hb_cv. Percentage error values were 31.9 and 32.1%. Trending analysis (subtraction of two subsequent values) did not demonstrate a correlation of Delta-SpHb with Delta-Hb_art (r=0.31; p=0.654) and Delta-Hb_cv (r=0.094; p=0.171), whereas Delta_Hb_art and Delta-Hb_cv were significantly associated (r=0.543; p<0.001). 4-quadrant-plot concordance for Delta-SpHb was poor vs. Delta_Hb_art (59%) and vs. Delta-Hb_cv (59%).

Conclusions: In mechanically ventilated ICU patients SpHb provided a moderate bias, but inacceptable precision and trending compared to Hb_art and Hb_cv.

P147 Clinical examination, LiMON, MASIMO and thermodilution for haemodynamic monitoring: the prospective CLIMATE-I study

W Huber, T Lahmer, G Batres-Baires, U Mayr, S Rasch, R Schmid, L Offman

II. Medizinische Klinik, Station 2/11, Munich, Germany

Introduction: There is a lack of studies comparing the use clinical examination to apparative haemodynamic monitoring. Therefore, the CLIMATE-studies compared clinical examination including recap-time (RCT), circulating blood volume (CBV; LiMON; Pulsion; Germany), perfusion index (MASIMO; USA) and transpulmonary thermodilution (TPTD; PiCCO; Pulsion; Germany).

Methods: The CLIMATE-I study compared PI with clinical findings and parameters of more invasive technologies. Primary endpoint: prediction of a cardiac index (CI) ≤ 2.5mL/min/sqm by non-invasive parameters. A total of 240 datasets were recorded within 24h in 30 patients (8 datasets per patient). Each dataset included clinical examination (RCT), infrared measurements of the surface temperatures on forehead, forearm, finger and great toe (Thermofocus; Technimed; Italy), CBV (only 1st measurement), perfusion index (PI; MASIMO) and data from TPTD. Statistics: IBM SPSS 25.

Results: Age: 65+/-12 years; mechanical ventilation 240/240, vasopressors 156/240 (65%) of measurements. In univariate analysis (p<0.001 except as indicated) PI had a stronger association to flow (Spearman-correlations with PI: stroke volume index: r=0.456; cardiac index CI: r=0.294) than to preload (CVP: r=0.214; p=0.001; GEDVI: r=0.254; SVV: r=-0.349). The strongest associations of PI were found with clinical markers as finger tip temperature (FTT; r=0.610) and RCT (r=-0.402). In multivariate analysis (R=0.607), PI was independently associated with FTT (p<0.001; T=7.859) and SVI (p<0.001; T=6.591), but not with CVP, GEDVI, SVV, CI, CBV and RCT.

Among the non-invasive technologies only FTT (ROC-AUC=0.681; p=0.028) and RCT (AUC=0.688; p=0.023) predicted CI ≤ 2.5mL/min/sqm. PI (AUC=0.446; p=0.512) was not predictive.

Conclusions: FTT and RCT are strongly associated with PI and predict a CI ≤ 2.5L/min/sqm. PI did not predict a CI ≤2.5L/min/sqm.

P148 Subjective right ventricle assessment in the critically ill vs objective measures

S Orde1, M Slama2, S Yastrebov3, A Mclean1, S Huang1

1Nepean Hospital, Intensive Care Unit, Sydney, NSW, Australia; 2Amiens University Hospital, Medical ICU, Amiens, France; 3St George Hospital, ICU, Sydney, Australia

Introduction: Subjective right ventricle (RV) assessment is common and guidelines suggest its utility in adequately trained clinicians. This has never been assessed in a robust manner. We compared subjective assessment of RV size and function in ICU patients by Intensivists with advanced echo qualifications vs RV free wall strain (RVfwS) assessed by speckle tracking echo, and conventional methods.

Methods: 52 Intensivists reviewed 2D loops from 80 critically ill patients. Inclusion criteria: mechanically ventilated with PaO2:FiO2 <300. Exclusion criteria: congenital heart disease, cardiac surgery, unable to perform RVfwS (19 patients, feasibility 80%). Echo measures included RV size: end diastolic area (EDA) and RV dimensions; RV function: fractional area change (FAC), S’, TAPSE, RVfwS. Binary (normal vs abnormal) and ordinal analysis (normal, mild/moderate or severe) was performed for RV size and function.

Results: 80 patients: 54% male, median 68years (59-73); P:F ratio 174 (132-208); PEEP 10 (7-12); APACHE III 80.5 (26); median ventilation time 6 days (3-9). Fair agreement was seen in subjective assessment vs objective measures with binary assessment of RV size and function. Ordinal data analysis showed poor agreement with RVfwS (Figure 1) and RV dimensions. If one-step disagreement was allowed the agreement was good (Table 1, 2). Significant overestimation of severity of abnormalities was seen comparing subjective assessment with RV EDA and TAPSE, s’ and FAC. There was no difference in agreement values when accounting for clinician echo experience, perceived expertise (at level of cardiologist) or type of qualifications.

Conclusions: Relatively low levels of agreement were seen with subjective assessment vs objective measures of RV size and function assessed by echo. It seems prudent to avoid subjective RV assessment in isolation and a combination of objective and subjective measures should be used.

Table 1 (abstract P148). Agreement of RV size assessment
Table 2 (abstract P148). Agreement of RV function assessment
Fig. 1 (abstract P148).

Subjective RV function and size assessment vs objective measures

P149 Predicting hypotension episode with numerical vital sign signals in the intensive care unit

JH Yoon1, V Jeanselme2, A Dubrawski2, MR Pinsky1, G Clermont1

1University of Pittsburgh, Critical Care Medicine, Pittsburgh, United States; 2Carnegie Mellon University, Auton Lab, Pittsburgh, United States

Introduction: Even short periods of hypotension are associated with increased morbidity and mortality. Using high-density numerical physiologic data, we developed a machine learning (ML) model to predict hypotension episodes, and further characterized risk trajectories leading to hypotension.

Methods: A subset of subjects with 1/60Hz physiological data was extracted from MIMIC2, a richly annotated multigranular database. Hypotension was defined as >5 measurements of systolic blood pressure ≤ 90 mmHg and mean arterial pressure ≤ 60 mmHg, within a 10-minute window. Derived features using raw measurements of heart rate, respiratory rate, oxygen saturation, and blood pressure were computed. Random Forest (RF), K-Nearest Neighbors (KNN), and Logistic Regression models were trained with 10-fold cross validation to predict instantaneous risk of hypotension using features extracted from the data leading to the first episode of hypotension (cases) or ICU discharge in subjects never experiencing hypotension (controls). For a given subject, risk trajectory was computed from the collation of instantaneous risks.

Results: From a source population of 2808 subjects, 442 subjects met our definition of hypotension, and 724 subjects without hypotension comprised the control group. 204 features were generated from the four vital signs. The area under the curve (AUC) for Random Forest classifier was 0.829, out-performing Logistic Regression (AUC 0.826) or K-Nearest Neighbors (AUC 0.783) (Fig 1). Risk trajectories analysis showed average controls risk scores <0.3 (<30% risk of future hypotension), while the hypotension group had a rising risk score (0.45 to 0.7) in the 8 hours leading to the first hypotension episode, and significantly higher scores leading into subsequent episodes (Fig 2).

Conclusions: Hypotension episodes can be predicted from vital sign time series using supervised ML. Subjects developed hypotension have an increased risk compared to controls at least 8 hours prior to the episode.

Fig. 1 (abstract P149).

Comparison of performance of three different supervised machine learning algorithm with receiver operating characteristic (ROC) curve

Fig. 2 (abstract P149).

Risk evolution over the last 8 hours before the episode in the ICU. On the x-axis, time 0 is the time for hypotension episode (case: red) or time of discharge or end of monitoring (control: blue)

P150 Cardiac output estimation using uncalibrated pulse wave analysis (PWA) in postoperative cardiac surgery patients: a method comparison study using pulmonary artery thermodilution as the reference method

G Greiwe1, K Luehsen2, B Saugel1

1University Medical Centre Hamburg-Eppendorf, Department of Anesthesiology, Hamburg, Germany; 2University Medical Centre Hamburg-Eppendorf, Hamburg, Germany

Introduction: In critically ill patients or in patients undergoing major surgery, monitoring of CO is recommended[1-3]. Less-invasive advanced hemodynamic monitoring with PWA is increasingly used in perioperative and critical care medicine. In this study, we evaluate the measurement performance of an uncalibrated pulse wave analysis (PWA) device (MostcareUp, Vygon, Ecouen, France) compared with cardiac output (CO) assessment by pulmonary artery thermodilution (PATD) in patients after cardiac surgery.

Methods: In patients after cardiac surgery, we performed seven sets of PATD measurements to assess PATD-CO. Simultaneously, we recorded the PWA-CO and compared it to the corresponding PATD-CO. To describe the agreement between PWA-CO and PATD-CO we used Bland-Altman analysis showing the mean of the differences and 95%-limits of agreement and calculated the percentage error.

Results: We included 17 patients in the analysis. The bias between PWA-CO and PATD-CO was 0.01 L*min-1. Upper and lower 95% limits of agreement were +1.40 L*min-1 and -1.38 L*min-1. The percentage error was 28.1%.

Conclusions: PWA-CO estimated with using the MostcareUp device shows good agreement with pulmonary artery thermodilution-derived CO in patients after cardiac surgery.


1. Cecconi et al. (2014) Consensus on circulatory shock and hemodynamic monitoring. Task force of the European Society of Intensive Care Medicine.

2. Saugel B, Vincent JL (2018) Cardiac output monitoring: how to choose the optimal method for the individual patient.

3. Vincent JL et al. (2015) Perioperative cardiovascular monitoring of high-risk patients: a consensus of 12.

P151 Non-invasive monitoring using photoplethysmography technology

K Tomita1, T Nakada1, T Oshima1, T Oami1, T Aizimu2, T Aizimu2, S Oda1

1Chiba University Graduate School of Medicine, Emergency and Critical Care Medicine, Chiba, Japan; 2Chiba University, Center for Frontier Medical Engineering, Chiba, Japan

Introduction: Non-invasive continuous blood pressure monitoring devices have been investigated, however, these devices did not have sufficient accuracy and precision. We developed a continuous monitor using the photoplethysmographic technique and tested the accuracy and precision of this system to ensure it was comparable to conventional continuous monitoring methods used for critically ill patients.

Methods: The study device was developed to measure blood pressure, pulse rate, respiratory rate, and oxygen saturation, continuously with a single sensor using the photoplethysmographic technique. Patients who were monitored with arterial pressure lines in the ICU were enrolled. The physiological parameters were measured continuously for 30 minutes at 5-minute intervals using the study device and the conventional methods. The primary outcome variable was blood pressure.

Results: Pearson fs correlation coefficient between the conventional method and photoplethysmography device were 0.993 for systolic blood pressure, 0.985 for diastolic blood pressure, 0.998 for mean blood pressure, 0.996 for pulse rate, 0.995 for respiratory rate, and 0.963 for oxygen saturation. Percent errors for systolic, diastolic and mean blood pressures were 2.4% and 6.7% and 6.5%, respectively. Percent errors for pulse rate, respiratory rate and oxygen saturation were 3.4%, 5.6% and 1.4%, respectively.

Conclusions: The non-invasive, continuous, multi-parameter monitoring device presented high level of agreement with the invasive arterial blood pressure monitoring, along with sufficient accuracy and precision in the measurements of pulse rate, respiratory rate, and oxygen saturation.

P152 Trending, accuracy and precision of stroke volume measurement from noninvasive bioreactance and pulse wave transit time compared with oesophageal Doppler

C Pisitsak

Ramathibodi Hospital, Department of Anesthesiology, Bangkok, Thailand

Introduction: Evidences have shown that perioperative goal directed hemodynamic management using oesophageal Doppler monitoring (ODM) can improve patient outcomes. The ODM is considered minimal invasive device. The authors compared the trending, accuracy and precision of non-invasive stroke volume measurement based on bioreactance technique and pulse wave transit time (PWTT) with ODM.

Methods: Ten patients who underwent abdominal surgery under general anesthesia were included for repetitive measurement of stroke volume on five minutes interval after anesthesia. The ODM, bioreactance and PWTT based stroke volume measurement were monitored in each patient. The correlation of stroke volume between bioreactance and PWTT with ODM were analyzed using Pearson correlation coefficient. The Bland and Altman analysis was performed to evaluate the accuracy and precision of PWTT and bioreactance against ODM.

Results: The total data sets of 227 were included for analysis. The results demonstrated the correlation coefficient of 0.75 (p<0.001, 95% CI 0.62-0.78), a bias of 0.28 ml (limits of agreement of -30.92 - 31.38 ml) and the percentage error of 46.72% between ODM and bioreactance (Fig 1) and the correlation coefficient of 0.48 (p<0.001, 95%CI 0.44-0.72), a bias of -0.18 ml (limits of agreement of -40.28 - 39.92 ml) and the percentage error of 60.25% between ODM and PWTT (Fig 2).

Conclusions: Stroke volume measurement using bioreactance technique had strong correlation with ODM while PWTT had moderate correlation. Both devices had small bias with wide limits of agreement and percentage error compared with ODM. Therefore, these devices are not interchangeable with ODM. However, using trends in stroke volume to guide treatment might still be acceptable.


1. Abbas SM, Hill AG. Anaesthesia 2008;63(1):44–51.

Fig. 1 (abstract P152).

Comparison between oesophageal Doppler and bioreactance technique for stroke volume measurement, Bland and Altman Plot (left panel), regression line and data (right panel). SV = stroke volume, OED = oesophageal Doppler

Fig. 2 (abstract P152).

Comparison between oesophageal Doppler and pulse wave transit time for stroke volume measurement, Bland and Altman Plot (left panel), regression line and data (right panel). SV = stroke volume, OED = oesophageal Doppler, PWTT = pulse wave transit time

P153 Hemorrhage diagnosis using sequential deep learning models of waveform vital sign data

F Falck1, MR Pinsky2, G Clermont2, A Dubrawski1

1Carnegie Mellon University, Auton Lab, School of Computer Science, Pittsburgh, United States; 2University of Pittsburgh, Department of Critical Care Medicine, School of Medicine, Pittsburgh, United States

Introduction: Hemorrhage is the most common cause of trauma deaths and the most frequent complication of major surgery. It is difficult to identify until profound blood loss has already occurred. We aim at detecting hemorrhage early and reliably using waveform vital sign data routinely collected before, during, and after surgery.

Methods: We use waveform vital sign data collected at 250 Hz during a controlled transition from a stable (non-bleeding) to a fixed bleeding state of 93 pigs. These vital signs include airway, arterial, central venous and pulmonary arterial pressures, venous oxygen saturation (SvO2), pulse oximetry pleth and ECG heartrate, continuous CO, and stroke volume variation (LiDCO). We used Gated Recurrent Units (GRU), Long Short-Term Memory (LSTM) and dilated, causal, one-dimensional convolutional neural networks presenting time windows drawn from the raw vital signs as inputs during training.

Results: Our GRU model reaches Area Under ROC (AUC) at a clinically relevant False Positive Rate (FPR) <1% of 0.7469 and True Positive Rate of 0.5639, thus achieving an operationally relevant performance for a real-world clinical application (Table 1). However, outside of the very low FPR range (cf. ROCs in Fig. 1 and 2), our models appear inferior to a referenced Random Forest (RF) classifier.

Conclusions: Our work demonstrates the applicability of deep learning models to diagnose hemorrhage based on raw, waveform vital signs. Future work will address why the RF classifier can address the greater homogeneity of subjects when they bleed compared to an apparently wide dispersion of their statuses when being stable.


This work is partially supported by NIH GM117622.

Table 1 (abstract P153). Experimental results (superslow bleeding subjects)
Fig. 1 (abstract P153).


Fig. 2 (abstract P153).


P154 Can myocardial perfusion imaging with echo contrast help recognise type 1 acute myocardial infarction in the critically ill?

S Orde1, M Slama2, F Pathan3, S Huang4, A Mclean4

1Nepean Hospital, Intensive Care Unit, Sydney, NSW, Australia; 2Amiens University Hospital, Medical ICU, Amiens, France; 3Nepean Hospital, Cardiology Department, Sydney, NSW, Australia; 4Nepean Hospital, Sydney, NSW, Australia

Introduction: We sought to assess feasibility of myocardial contrast perfusion echo (MCPE) in the critically ill and if MCPE could aid correlation of clinical acumen in detecting type 1 myocardial infarction [acute MI]. Diagnosis of acute MI in ICU is difficult and inappropriate intervention can be harmful.

Methods: Single centre, prospective, observational study. Critically ill adult patients with Troponin I >50ng/L and acute MI being considered were included. Exclusion criteria: poor echo windows (2pts), known ischaemic heart disease, contrast contraindications.

MCPE analyses flow in the myocardium. After transient microbubble destruction (‘flash’), combined replenishment rate and plateau intensity estimate myocardial blood flow in ‘region of interest’ (ROI) (Figure 1).

Ischaemia assessed by angiography (22pts), nuclear imaging (1pt), MRI (3pts), CTA (3pts) or normal repeat echo in stress induced cardiomyopathy (14pts). 2 cardiologists and 6 intensivists, blinded to outcome, analysed history, ECG, Troponin and 2D echo images to estimate likelihood of MI.

Results: 40 patients (28 female), age 65 years (IQR 44-73) with Troponin 1987 ng/L (400-4384). 6 had acute ischaemia (15%). MCPE analysis was feasible in 415/640 segments (68.8% [IQR 50-81]). No adverse events seen. Regional wall analysis and global longitudinal strain (GLS) were both abnormal, but no significant difference seen in acute MI group vs none. Significant difference was seen in MCPE assessment (3.3 vs 2.4dB/s, p=0.050). A myocardial blood flow 2.8dB/s had 67% sensitivity and 88% specificity vs acute MI. Clinical acumen had mild agreement with outcomes (r2 0.1, p=0.054; AUC 0.74); however, combining clinical acumen with MCPE analysis correlation improved (r2 0.3, p=0.03; AUC 0.87) (Figure 2).

Conclusions: MCPE is feasible in the critically ill and showed better correlation with presence of acute MI vs clinical acumen alone.

Fig. 1 (abstract P154).

Myocardial contrast perfusion echocardiography (MCPE)

Fig. 2 (abstract P154).

Receiver operating curves for detection of type 1 myocardial infarction

P155 Non-invasive hemorrhage detection approach using photoplethysmography

JH Yoon, Y Chen, MR Pinsky, G Clermont

University of Pittsburgh, Critical Care Medicine, Pittsburgh, United States

Introduction: Many instances of significant bleeding may not occur in highly monitored environment, contribution in the delay in recognition and intervention. We therefore proposed a non-invasive monitoring for early bleeding detection using photoplethysmography (PPG).

Methods: Fifty-two Yorkshire pigs were anesthetized, stabilized and bled to hemorrhagic shock, and their invasive arterial blood pressure (ABP), and PPG data were collected [1]. Time series of vital signs were divided into data frames of 1 minute updated every 30 seconds and beat to beat features were computed. The final feature matrix contained 18 ABP features and 85 PPG features. A supervised machine-learning framework using Least Absolute Shrinkage and Selection Operator regularized logistic regression model was constructed to score the probabilities for hemorrhage of each data frame. Data in stabilization was set as negative and data in bleeding was set as positive. Model performance was evaluated by receiver operating characteristic (ROC) area under the curve (AUC) with leave-one-out cross validation, and its precision was assessed with activity monitoring operative characteristic (AMOC).

Results: Two different models were proposed using ABP and PPG features separately. Figure 1 showed the PPG model could classify the hemorrhage with AUC = 0.89, where the AUC of ABP model was 0.91. Figure 2 showed the PPG model could detect the hemorrhage on average 15.5 minutes (equals to 320 ml blood loss) if the false alarm rate of 1/100 was tolerated, whereas the average detection time of ABP model were 12.5 minutes at same threshold of false alarm rate.

Conclusions: We proposed a novel non-invasive bleeding detection approach using PPG signals only. This method potentially can improve the identification of hemorrhage with in patients and environments where invasive monitoring is unavailable.


1. Gomez H, et al. J Surg Res 2012.

Fig. 1 (abstract P155).

Comparison of model performance with two different vital sign feature sets, on Receiver Operating Characteristic (ROC) curve

Fig. 2 (abstract P155).

Comparison on precision, on Activity Monitoring Operating Characteristic (AMOC) curve

P156 Feasibility of ultrasound guided central venous catheterization in critically ill patients without Trendelenburg position

A Dumoulin1, W Stockman1, P Lormans1, D Benoit2

1AZ Delta, Intensive Care medicine & Anaesthesiology, Rumbeke, Belgium; 2University Hospital Gent, Intensive Care UZ Gent, Gent, Belgium

Introduction: In our experience, most central venous catheters in ICU patients can be placed without using Trendelenburg position (TP) if ultrasound is used before and during the procedure, thereby avoiding unwanted side effects of TP.

Methods: We conducted a prospective observational single centre study from 28/02/2018 till 08/06/2018 in 30 consecutive mixed ICU patients who needed central venous catheterization using real-time ultrasound guidance. Ultrasound was used pre-procedural to determine the appropriate location for catheterization and the size of the vessel in various non TP positions. If deemed necessary the physician could adjust the bed position to optimize ultrasound vessel visualization and patient comfort. We registered the angle of the upper body position using a digital protractor.

Results: Baseline characteristics are shown in Table 1, catheter and procedure characteristics are shown in Table 2. The median angle of bed position was 26°. No patients were positioned in neutral or TP. All procedures were successful with a mean of 1.3 punctures per patient, and a maximum of 3. The median procedure time was 12.5 minutes. No major complications occurred in any of our patients.

Conclusions: Central venous catheterisation in moderate upright position is feasible and can be done safely when using real-time ultrasound by well-trained physicians. We recommend performing clinical assessment and pre-procedural ultrasound to choose the optimal puncture site and position in order to attain an optimal ultrasound visualisation of the vessel and patient comfort.

Table 1 (abstract P156). Patient demographics
Table 2 (abstract P156). Catheter+ procedure characteristics

P157 A survey of practice for central venous catheter placement confirmation methods

K Chen, C Day

Royal Devon & Exeter Hospital, Critical Care, Exeter, United Kingdom

Introduction: Once a central venous catheter (CVC) has been inserted, its position should be confirmed and complications excluded. Ultrasound is commonly used to aid insertion but the extent to which it is used in confirmation of site and exclusion of complications is not known for the UK.

Methods: We conducted a telephone survey contacting all critical care units in England. The nurse in charge or an Advanced Critical Care Practitioner was asked to provide information about what constituted usual practice for most CVC lines inserted in their ICU. If they were busy a call back was arranged. If this failed then a further call was made to that ICU.

Results: 147 of the 190 ICU contacted provided information (77%). 94.6% used ultrasound to confirm the position of the guide wire during insertion. This could include exclusion of a wire in the subclavian vein or visualisation as far caudally as possible. 95.2% of units used a chest X-ray. 65.3% used waveform analysis from the line. 42.9% would use ultrasound to look for the catheter tip and/or to exclude pneumothorax. 35.4% used a blood gas taken from the line.

Conclusions: There is still a great dependancy on the chest x-ray to confirm position and exclude complications following CVC insertion. USS is used, mostly as an adjunct in confirming position and excluding complications. The use of USS is very dependant on an operator being available. There were several comments that practice during the day was different to those at night. During the day there was more likely to be a skilled operator, able to use USS for confirmation of line site and placement.


American Society of Anaesthesiologists Task Force on Central Venous Access. Anaesthesiology 2012; 116: 539-75;

P158 Ultrasound-guided oblique right brachiocephalic vein central venous cannulation (CVC) in adult: a novel approach

R Ismail, A Osman, A Ahmad, A Azil

Hospital Raja Permaisuri Bainun, Emergency & Trauma, Ipoh Perak, Malaysia

Introduction: The aim of this study was to evaluate the feasibility right brachiocephalic vein central vein cannulation (CVC) using high frequency linear ultrasound probe via an oblique needle trajectory [-3].

Methods: A retrospective analysis of 10 patients presenting to tertiary-care emergency department who required CVC for vasopressor administration was carried out. All central venous cannulation into the right brachiocephalic vein was performed with ultrasound guidance using the high frequency linear probe. Right brachiocephalic vein was visualised in its long axis. The needle was positioned just beside the centre of ultrasound probe 15 degrees below the coronal plane and 10 degrees angle to the ultrasound probe and advanced just behind the clavicle.

Results: The mean puncture time taken to perform this procedure, calculated from the needle piercing the skin until to the aspiration of blood from the brachiocephalic vein through the needle, was 23 ± 6.8s. No procedure-related complications were detected.

Conclusions: The oblique needle trajectory of right brachiocephalic vein CVC in adult is feasible and able to visualised well the anatomical structure, hence avoid complications.


1. Breschan, et al. Br JAnaesth 106 (5): 732–7 (2011)

2. Byon HJ, et al. Br J Anaesth 111 (5): 788–92 (2013)

3. Klug W, et al. Saudi J Anaesth. 2016;10(2):143-148.24.

P159 Identifying risk factors of central venous catheter placement related complications: an observational multicenter study

JM Smit1, TS Steenvoorden1, ME Haaksma1, EH Lim1, MJ Blans2, FH Bosch2, M Petjak3, B Vermin3, HR Touw1, AR Girbes1, L Heunks1, PR Tuinman1

1Amsterdam UMC, Location de Boelelaan, Intensive Care Medicine, Amsterdam, Netherlands; 2Rijnstate Hospital, Intensive Care Medicine, Arnhem, Netherlands; 3Groene Hart Ziekenhuis, Intensive Care Medicine, Gouda, Netherlands

Introduction: Central venous cannulation, a routine procedure on intensive care units, is associated with a low complication rate. As a consequence, the routine use of chest X-ray (CXR) or ultrasound (US) to assess these complications is under discussion. Our aim was to identify risk factors for central venous catheter (CVC) placement associated complications that can help decide whether or not follow-up using CXR and/or US is indicated.

Methods: Multicenter prospective, observational study. Consecutive critically ill adult patients who underwent CVC placement. Either the internal jugular vein or subclavian vein was cannulated. Complication rates were determined. Predicting factors were obtained through a questionnaire filled in by physicians after placing a CVC. If the questionnaire was incomplete or data was missing, analyses were performed using the available data. Patient characteristics were duplicated if a patient recieved more than one CVC. Outcomes were iatrogenic pneumothorax and malposition. Pneumothorax was detected using US, whereas CXR was used to determine CVC malposition.

Results: We included 756 CVC insertions in 727 patients. Median age was 68 (IQR: ± 15.0) years and men outnumbered women more than 2:1 (519 vs. 239). Mechanical complications were identified in 4.8 (n=36) of all CVC placements:11 pneumothoraces and 26 malpositions. Clinically relevant complications occurred in 1.2% (n=9) of the cases. Risk factors for CVC-placement related complications are shown in Table 1. US-guidance, insertion site, and setting were predictive for complications.

Conclusions: The overall CVC placement associated complication rate is low and multiple risk factors associated with the occurrence complications were identified. A complication rate this low, strongly suggests that routine post-procedural diagnostics is superfluous. Therefore, we suggest, provided that uneventful execution of the procedure is assured, post-procedural diagnostics are only necessary in selected cases with (multiple) risk factors.

Table 1 (abstract P159). Peri-procedural factors and the outcomes of CVC placement they correspond with

P160 Supraclavicular approach to ultrasound-guided subclavian vein cannulation. A preliminary study

Y Aissaoui

Avicenna Military Hospital - Caddi Ayyad University, Department of Anesthesiology and Intensive Care, Marrakesh, Morocco

Introduction: The use of ultrasound for subclavian vein cannulation (SCV) has developed poorly due to the difficulty of visualizing this vein via the classical infraclavicular approach. We explored the feasibility of ultrasound-guided subclavian vein catheterization via a supraclavicular approach

Methods: Prospective study conducted over six-month period in intensive care unit. After approval of the ethics committee, we included patients over 18 years of age and requiring central venous access. Exclusion criteria were: hemostasis disorders, puncture area infections and cervico-thoracic vascular malformations The procedure consisted of catheterization of the VSC with a supraclavicular approach under ultrasound guidance using an ultrasound in plane approach (Fig 1 and 2). Data collection included clinical and ultrasound data: SCV depth, diameter and length, catheterization time, number of needle redirection, cannulation success and complications.

Results: Thirty four patients were included. age: 57 ± 15 (Mean ± SD), 60% of whom were male. The success rate of SCV catheterization was 97% (one failure). The depth of the SCV was 11 ± 4.5 mm and its diameter was 11 ± 3.5 mm. The puncturable length of the SCV was 33 ± 7mm and the puncture angle was 36 ± 15 °. The time required to obtain an adequate ultrasound image was 25 ± 9 seconds. The interval between the beginning of the puncture and the insertion of the guidewire into the vein was 42 ± 29 sec. The total catheterization time was 69 ± 33 seconds. The number of needle redirection 0.8 +/- 1.2 redirects. The quality of the ultrasound image was excellent or good in 87.5% of cases. An arterial puncture was observed in two patients

Conclusions: This preliminary study demonstrated the feasibility of the subclavian vein cannulation via the supraclavicular approach. More study are required to confirm its safety and to compare this approach to the infraclavicular acces using ultrasound.

Fig. 1 (abstract P160).

Probe positioning in the supraclavicular fossa during ultrasound guided subclavian vein cannulation via supraclavicular approach

Fig. 2 (abstract P160).

Ultrasound image of the subclavian vein after cannulation and insertion of the guidewire

P161 Contribution of the ultrasound guidance in the set up of central venous catheters in emergency situations

W Sellami, I Ben mrad, Z Hajjej, M Chniti, I Labbene, M Ferjani

Department of critical care medicine and anesthesiology Military Hospital, Tunis, Tunisia

Introduction: In intensive care unit, we are confronted to place venous catheters in urgent situations. Although the femoral site is preferred, it remains difficult to handle in case of abdominal pelvic surgery or ventral position. In addition it does not allow to develop a diagnostic approach. The purpose of our study was to see if ultrasound guided cannulation of the internal jugular vein could be an alternative to the femoral one.

Methods: It was a prospective, monocentric, observational, and comparative study conducted in the anesthesia resuscitation department of the Military Hospital Main Instruction of Tunis over a period of 12 months. There were 118 patients, 58 in the group “guided femoral vein catheterization (FVC)” and 60 in the group “guided internal jugular vein catheterization (IJVC)”. The rate of failure and complications (mechanical, infectious and thrombotic) was compared; the number of punctures and access times too. The threshold of statistical significance was chosen at 0,05.

Results: The failure rate was 10.3% for FVC, compared to 8.3% for IJVC (p = 0.47, (p >0.05)). The risk of hematoma was 3.4% for FVC, 1.7% for IJVC (p = 0.5).No case of pneumothorax was noted, 2 malpositions of the IJVC were reported.Catheter-related infection was 6.9% for FVC, 3.3% for IJVC (p = 0.3). Venous thrombosis was 20.7% for FVC, 10% for IJVC, (p = 0.08). The number of attempts and the access time were lower for IJVC (respectively 1.3 ± 0.4 and 204s ± 46.3 vs 1.9 ± 0.7 and 256.5s ± 90.4 for the FVC, p <0.0001).

Conclusions: The failure rate and complications were comparable between the 2 groups, but the ultrasound-guided internal jugular catheter appears to be faster to insert and requires fewer punctures, so it could be an alternative to the femoral one in emergency situations.

P162 Correlation of lung ultrasound B-lines and New York Heart Association(NYHA) functional classification in congestive heart failure

CY Cheong, A Osman, AH Ahmad, MH Mohamed Sakan

Hospital Raja Permaisuri Bainun, Emergency and Trauma Department, Ipoh, Malaysia

Introduction: Lung ultrasound B-lines, a comet-like reverberation artefacts arising from water-thickened interlobular septa, indicate extravascular lung water which is a key variable in heart failure management and prognosis. Aim of this study is to measure the correlation between lung ultrasound B-lines and NYHA functional classification.

Methods: This is a 6 months prospective study on congestive heart failure patients conducted in 2 urban emergency departments in Malaysia. Following enrolment, patients had their functional capacity categorised based on NYHA classification, followed by Point of Care Ultrasound (POCUS) lung scan using a 12MHz linear probe. The scanning was performed by trained emergency physicians. The longitudinal scan done at the recommended 6 zones of both left and right lungs and the total number of B-lines identified were summed up as the comet score. Comet Score of 0, 1, 2 and 3 were categorised based on amount of B-lines of less than 5, 5-15, 15-30 and more than 30 B-lines respectively.

Results: Hundred and twenty-two patients were analysed (69 males(56.6%) and 53 females(43.4%)) ranging from 24 to 88 years old. Comet Score of 1,2 and 3 were found to be statistically significant with presence of paroxysmal nocturnal dyspnoea, elevated jugular venous pressure, lung crackles, bilateral pitting oedema and chest radiographic findings. A moderate correlation between NYHA classes with Comet Score 1,2 and 3 (rs= 0.77 (P<0.01)) was documented.

Conclusions: Our study demonstrated a moderate correlation between NYHA classes and lung ultrasound B-lines. Lung ultrasound may be a potential tool to objectively determine the functional capacity in patients with congestive heart failure and monitor its changes in response to treatment and disease progression.

P163 The use of online communications as a tool for ultrasound peer review: a survey of users

O Olusanya1, J Wilkinson2, P Parulekar3

1St Mary´s Hospital, Intensive Care Unit, London, United Kingdom; 2Northampton General Hospital, Intensive Care Unit, Northampton, United Kingdom; 3East Kent Hospitals University NHS Trust, Intensive Care Unit, Kent, United Kingdom

Introduction: Point of Care Ultrasound (POCUS) is a tool of increasing utility in the management of the critically ill patient. Guidelines exist for training and accreditation in POCUS [1, 2] however the widespread use of POCUS has been hampered by a lack of mentors. Online communication with end-to-end security, such as WhatsApp ™ are increasingly used in medicine as a communication aid [3]. Some individuals are using such communications to share POCUS images for review- the overall sentiment around these tools is unknown.

Methods: An online survey of POCUS users was conducted via Twitter ™. The question was “in situations where an expert opinion on an ultrasound is not immediately available, is it acceptable to get an expert review via an online medium such as WhatsApp, and would you be happy to be that expert?”

Results: 304 votes were received. Voters were a mix of POCUS users from the USA, Europe, and Australia. 58% said the medium was acceptable, and that they would be happy to provide expertise. 34% voted “no”, with 8% voting “other” (Fig 1).

Conclusions: In this international survey of POCUS users, 58% were happy to provide and receive mentorship using remote software such as WhatsApp. Distance mentorship for POCUS training should be explored.


1. Volpicelli G, et al. ICM 2012; 38(4):577-591.

2. Via G et al. J Am Soc Echocardiogr. 2014 Jul;27(7):683.e1-683.e33

3. NHS England. (last checked 27th Nov)

Fig. 1 (abstract P163).

Distribution of votes in Poll

P164 A modified Delphi process to develop an integrated scheduled inter-professional assessment of critically ill patients with ultrasound. INSIGHT

E Corcoran1, N Hare2, D Hadfield1, S Helyar1, P Hopkins3

1Kings College Hospital London - London, Research, London, United Kingdom; 2GSTT, Research, London, United Kingdom; 3Kings College Hospital London - London, Intensive Care Medicine, London, United Kingdom

Introduction: Point of care, focused ultrasound offers safe, non-invasive imaging for immediate evaluation, resuscitation and guidance of therapeutic procedures in the critically ill patient [1]. Our previous research showed all critical care professions within our institution support a clinical effectiveness trial of nurse-led scheduled ultrasound [2]. A description of the development and refinement of INSIGHT - a feasibility and clinical effectiveness randomized controlled trial.

Methods: A modified Delphi exercise was used to select the most beneficial ultrasound windows and imaging questions to ask for each window in scheduled inter-professional ultrasound. 260 nurses, 46 doctors and 6 physiotherapists from critical care were given the same information regarding potential utility of each window. The windows and associated questions were individually ranked; each window and question tested against three further criteria; and filtered by ease of training to level 1 standard; clinical usefulness; time of practical delivery and applicability across an inter-professional group.

Results: The Modified Delphi exercises and prioritization exercise ranked ease of adoption by training; feasibility within the time frame and clinical usefulness to develop a core INSIGHT scan of 5 domains, each with set binary questions (Tables 1 and 2)

Conclusions: We have developed a research intervention that will allow us to test the effectiveness of inter-professional scheduled whole body assessment of critically ill patients by ultrasound. We now plan to conduct a clinical effectiveness trial with an internal pilot to confirm feasibility.


(1) Beaulieu Y&Marik P. 2005. Chest 128: 1766-81.

(2) Hare N, et al. 2014. Intensive Care Medicine 40 (conference abstract).

Table 1 (abstract P164). Results from the first stage of the Delphi exercises: 10 windows across the whole body were found to be useful. These are ranked in order of importance (1-10). A subsequent prioritization exercise by ease of adoption by training showed number of scans required for each window to reach level one standard (ranked 1-10). Time taken to scan each window (ranked 1-10) and clinical usefulness of each window (ranked 1-10). Total added up for each window and sums helped identify 5 key windows for use in the trial
Table 2 (abstract P164). Final choice of windows (subcostal cardiac/IVC, right hemi-diaphragm/right lower lobe/subphrenic assessment, abdominal ultraosund using bladder as acoustic window, left hemidiaphragm, central veins - internal jugular/subclavian/femoral. The questions that are required to be answered when scanning each window (outcome of Delphi exercises) are identified

P165 Central venous pressure cannot be reliably estimated by transthoracic echocardiography in critically ill patients

M Jozwiak, A Boilève, JL Teboul, C Richard, X Monnet

Hôpital Bicêtre, service de médecine intensive réanimation, Le Kremlin-Bicêtre, France

Introduction: The central venous pressure (CVP) remains a helpful variable at the bedside for hemodynamic management of critically-ill patients. Moreover, transthoracic echocardiography (TTE) is currently the first-line tool for evaluating the hemodynamic conditions in such patients. We sought to assess the ability of TTE to estimate and to track changes in CVP in critically ill patients.

Methods: In 31 patients (25 mechanically ventilated and four with atrial fibrillation), concomitant TTE examination and CVP measurement (internal jugular catheter) were performed before and after a passive leg raising test (n=10) or the infusion of 500-mL of saline (n=21).

Results: After pooling all values, CVP was only correlated to the end-expiratory inferior vena cava (IVC) diameter and the respiratory variations of the IVC diameter (r=0.40 and r=-0.26, respectively, p<0.05). At baseline, an end-expiratory IVC diameter ≤14 mm predicted a CVP ≤10 mmHg with a specificity of 100% (95%CI: 81-100%) but a sensitivity of 29% (95%CI: 8-58%). An end-expiratory IVC diameter >26 mm predicted a CVP >10 mmHg with a specificity of 100% (95%CI: 77-100%) but a sensitivity of 0% (95%CI: 0-20%). The respiratory variations of the IVC diameter could predict neither a CVP ≤10 mmHg nor a CVP >10 mmHg (AUC=0.683, p=0.08). Relative changes in CVP were weakly correlated to relative changes in S/D ratio and in velocity-time integral (VTI)-systolic filling fraction of the supra-hepatic vein flow (r=0.44 and r=0.43, respectively, p<0.05). The concordance rate between changes in CVP and S/D ratio or changes in VTI-systolic filling fraction was 48% and 55%, respectively.

Conclusions: In critically-ill patients, TTE cannot reliably estimate CVP or track changes in CVP induced by a passive leg raising test or fluid administration. However, an end-expiratory IVC diameter ≤14 mm or >26 mm can predict a CVP ≤10 mmHg or >10 mmHg, respectively, with a specificity of 100%.

P166 Lung water assessment by lung ultrasonography

A Ramos1, A Dogliotti2, F Acharta3, D Latasa1, F Gerber1, J Robles1, C Lovesio1

1Grupo Oroño, Critical Care, Rosario, Argentina; 2Grupo Oroño, Statistics and Epidemiology, Rosario, Argentina; 3Sanatorio Parque, Critical Care, Carlos Pellegrini, Argentina

Introduction: The objective was to investigate the accuracy of lung ultrasonography in the quantification of lung water in critically ill patients by using transpulmonary thermodilution technique as the gold standard for the determination of extravascular lung water level (ELWI).

Methods: Prospective study of 20 consecutive, fully mechanically ventilated patients who required hemodynamic monitoriztion with EV1000, were included. ELWI was determined using the transpulmonary thermodilution technique, a value above 10 ml/kg was considered as the presence of pulmonary water . Semiquantitative ultrasound assessment of lung water was performed by determining the ultrasound B-line, defined as the total number of B-lines detectable in an anterolateral lung ultrasonography examination, ≤ 3 B-lines was considered as absence of pulmonary water and >4 B-lines was considered as the presence of lung water.

Results: Good correlations was found between the presence of more than 3 B-line and ELWI; ≤ 3 B-lines, Kappa < 0 (no agreement) (p= 0.06); >3, Kappa 1 (good agreement) (p <0.0001). Sensitivity 58.3% (95% CI: 27 to 84), specificity 100%, positive predictive value 100%, negative predictive value 61.5% (95% CI: 31 to 86), (p= 0.006).

Conclusions: More than 3 B-lines by ultrasonography agreed with transpulmonary thermodilution technique. For the entire population: >3 B-lines by ultrasonography has a good positive predicted value. Lung ultrasonography examination may provide a reliable, simple and bedside lung densitometry in the intensive care setting.

P167 Search for optimal strength and time for capillary refilling time

R Kawaguchi1, T Nakada1, T Oshima1, M Shinozaki2, T Nakaguchi2, H Haneishi2, S Oda1

1Chiba University Graduate School of Medicine, Department of Emergency and Critical Care Medicine, Chiba, Japan; 2Chiba University Center for Frontier Medicine Enginieering, Chiba, Japan

Introduction: Capillary refilling time (CRT) has been widely used in clinical settings. However, CRT measurement conditions, including pressing strength and time, are not standardized. Search for optimal strength and time for CRT measurement seems to be needed.

Methods: We developed a novel device, which can adjust pressing strength and time, and can precisely measure CRT using an electric actuator and strength and color sensor. CRT was in 31 healthy adults using the device under conditions of pressing strength 1, 3, 5 and 7 N, and pressing time 1, 2, 3, 4, 5 and 6 s.

Results: There was a significant difference in CRT for pressing strength but not for pressing time (two-way ANOVA: power P<0.001, time P=0.97). There was a significant difference in CRT between pressing strength of 1 and 3 N but not for 3, 5 and 7 N (P=0.16). Thus 3 N seems to be needed for the pressing strength. To search for optimal pressing time, the plots from the color sensor during nail bed compression were analyzed. We found two phases in the color sensor plots. In the initial part of compression, the plots changes rapidly (rapid phase) and then the slope of plots reduces (slow phase). The pressure release during the rapid phase could destabilize the measurement. The longest period of the rapid phase was 1.9999 s among all the study subjects. Thus, a pressing time of 2 s seems to be needed to obtain stable CRT measurements.

Conclusions: On our study for the investigation of standard pressing time and strength for CRT measurements, pressing the nail bed with 3-7 N and 2 s appears to be optimal.

P168 Detection of pancreas ischemia with microdialysis and CO2-sensors in a porcine model

K Rydenfelt1, R Strand-Amundsen2, R Horneland3, S Hødnebø1, TI Tønnessen1, H Haugaa1

1Oslo university hospital, Department of Anesthesiology, Oslo, Norway; 2Oslo university hospital, Department of Clinical and Biomedical Engineering, Oslo, Norway; 3Oslo university hospital, Department of Transplantation Medicine, Section of Transplantation Surgery, Oslo, Norway

Introduction: Pancreas transplantation is associated with a high rate of early graft thrombosis. Current postoperative monitoring lack tools for early detection of ischemia, which could precipitate a graft-saving intervention. We are currently exploring the possibility of ischemia detection with microdialysis and CO2-sensors in the organ tissue or on the surface in a porcine model.

Methods: In anesthetized pigs, CO2-sensors and microdialysis catheters are inserted into the parenchyma or attached to the surface of the pancreas. PCO2 is measured continuously and lactate is sampled with microdialysis every 15 min. Ischemia is induced by sequential arterial and venous occlusions for 45 minutes, with 120 minutes of reperfusion in between.

Results: PCO2 increased and decreased in response to ischemia and reperfusion within minutes. Lactate increased and decreased with the same pattern, but with a considerable delay as compared to PCO2. An example is depicted in Figure 1. The values are presented in Table 1 (n=3).

Conclusions: Preliminary data suggest that PCO2 increase and decrease within minutes in response to ischemia and reperfusion in the pancreas and seem to correlate well with lactate measured with microdialysis both intraparenchymally and on the surface of the organ.

Table 1 (abstract P168). Tissue and surface lactate and PCO2 in response to ischemia and reperfusion, results from 3 pigs
Fig. 1 (abstract P168).

Ischemia detected with CO2-sensors (PCO2) and microdialysis (lactate), in the parenchyma and on the surface of the pancreas. Example from one pig

P169 Validation of automated software for analysis of the sublingual microcirculation in clinical and experimental settings

MP Hilty1, P Giaccaglia1, S Akin1, O Erdem1, B Ergin1, DM Milstein2, F Toraman3, Z Uz1, G Veenstra4, C Ince1

1Erasmus MC, Department of Intensive Care, Rotterdam, Netherlands; 2University Medical Center, Amsterdam, Netherlands; 3Acıbadem Mehmet Ali Aydınlar University School of Medicine, Istanbul, Turkey; 4Medical Center Leeuwarden, Leeuwarden, Netherlands

Introduction: Reliable automated handheld vital microscopy (HVM) image sequence analysis is a prerequisite for use of sublingual microcirculation measurements at the point-of care according to the current consensus statement. We aim to validate a recently developed advanced computer vision algorithm [1] versus manual analysis in a wide spectrum of populations and contexts.

Methods: Our collaborators were invited to contribute raw data of published or ongoing institutional review board approved work. Inclusion criteria were use of the Cytocam HVM device, manual analysis with the AVA software, and image quality as independently assessed by Massey score of <10 in >50% of recordings in a random subset of each study. 233 subjects from 11 studies were included, covering clinical and experimental populations, major shock forms and interventions to recruit the microcirculation (Table 1).

Results: 2,599,710 red blood cells were tracked by the algorithm across 150,163 frames in 1462 measurements in real time. A good to excellent correlation was found between algorithm-determined and manual capillary density (p<0.0001, r 0.6–0.9, Figure 1). Capillary perfusion was classified using space-time diagram derived red blood cell velocity (RBCv), yielding good correlation with manual analysis for functional capillary density und proportion of perfused vessels. Microcirculatory alterations during disease and interventions were equally detected by the algorithm and manual analysis. Change in flow short of severe abnormality was reflected in absolute RBCv but not microcirculatory flow index.

Conclusions: We demonstrate the validity of automated software for HVM image sequence analysis across broad populations, disease conditions and interventions. Thus, microcirculatory assessment at the bedside may finally complement point-of-care evaluation of disease severity and treatment response in critically ill patients and during surgery.


1. Hilty et al, ICMx 2018;6(suppl 2):40;172-173.

Table 1 (abstract P169). Our study population covers a wide range of major patient populations, clinical conditions associated with specific microcirculatory abnormalities, as well as interventions to recruit the microcirculation
Fig. 1 (abstract P169).

A good correlation was found between algorithm-determined and manually measured capillary total vessel density (TVD). Substudy indices (A-K) correspond to Table 1. Dashed lines represent identity lines. TVD, capillary total vessel density; ICU, intensive care unit; LVAD, left ventricular assist device

P170 Validating the point of care microcirculation (POEM) score to characterize sublingual microcirculatory flow

H Fargaly1, A Deitchman1, M Gilani1, J Assadi1, S Hutchings2, M McCurdy1

1University of Maryland, School of Medicine, Pulmonary and critical care, Baltimore, United States,2King´s College Hospital, Critical Care, London, United Kingdom

Introduction: In 2016, Naumann et al introduced the POEM score as a real-time, point-of-care score to assess sublingual microcirculation [1]. Our study aimed to determine the reproducibility of the POEM score.

Methods: Two expert operators used a sidestream darkfield (SDF) videomicroscope (Cytocam, Braedius, Netherlands) to separately acquire four high-quality video clips and assign a POEM score to each image in 20 adult mechanically ventilated patients. Each operator was blinded to the other’s images and analysis. Video clip scores and acquisition times were recorded.

Results: Of the 20 patients enrolled in this study, 45% (n=9) required vasopressors. We categorized POEM scores 4-5 as “normal” and POEM scores 1-3 as “impaired.” (Fig 1). With only one instance of interrater disagreement (i.e., a single image scored as 3 versus 4), Cohen’s kappa (0.86) confirmed a strong correlation between interpreters. The mean time to complete a study session was 9 minutes.

Conclusions: The present inability to quickly characterize the quality of sublingual microcirculation as either normal or impaired at the point of care limits real-world clinical application of this resuscitative endpoint. The rapidly obtained POEM score appears to be reproducible between bedside interpreters. Future studies should assess the effect of POEM score-guided resuscitation.


1. Naumann DN et al. Crit Care. 2016; 20:310.

Fig. 1 (abstract P170).

Microcirculatory images for two different subjects

P171 Microcirculatory response to passive leg raising in recreational marathon runners

I Kiudulaite1, Z Pranskuniene2, J Arstikyte3, M Brazaitis4, A Pranskunas5

1Lithuanian university of health sciences, Department of intensive care medicine, Kaunas, Lithuania; 2Lithuanian university of health sciences, Department of Drug Technology and Social Pharmacy, Kaunas, Lithuania; 3Lithuanian university of health sciences, Institute for Digestive Research, Kaunas, Lithuania; 4Lithuanian Sports University, Sports Science and Innovation Institute, Kaunas, Lithuania; 5Lithuanian university of health sciences, Department of Intensive Care Medicine, Kaunas, Lithuania

Introduction: We hypothesize that marathon runners have different microcirculatory response to passive leg raising (PLR) evaluated before and after marathon.

Methods: Eleven healthy marathon runners with a training volume of at least 5 h per week for a minimum of 2 years volunteered for the study. PLR were performed two times: 24 hours prior to their participation in the Kaunas Marathon (distance: 41.2 km) and directly after finishing the marathon. The systemic hemodynamic and microcirculatory measurements were obtained before and after PLR. Systemic hemodynamic parameters were measured using impedance cardiography (Niccomo, Medis, Medizinische Messtechnik GmbH, Germany). Sublingual microcirculatory images were obtained using a Cytocam-IDF device (Braedius Medical, Huizen, The Netherlands) and analyzed using standardized published recommendations.

Results: The median age of participants was 32 years. We found no significant difference in proportions of hemodynamic responders before and after marathon (64% vs 55%, p=0.819). Also we did not find differences between PLR induced changes of total vessel density (TVD) and proportion of perfused vessels (PPV) of small vessels before and after marathon. Correlations between changes of sroke volume and changes of TVD or PPV of small vessels during PLR were not significant.

Conclusions: Marathon running did not change microcirculatory responsiveness.

P172 Correlation between perfusion index and lactate level in critically ill patients

K Kumwilaisak, J Sereeyotin

King Chulalongkorn Memorial Hospital, Anesthesiology Department, Bangkok, Thailand

Introduction: Macrocirculation and microcirculation are the key components to determine organ perfusion. Early detection of tissue hypoperfusion would guide us to actively manage and lead to the better outcome. Presently, we commonly use serum lactate as a representative of global tissue perfusion but it needs blood sampling and obtains intermittent data. Perfusion index(PI) based on analysis of pulse oximetry signal is noninvasive continuous monitoring for peripheral tissue perfusion. Previous studies have shown correlations between PI and microcirculation parameters in critically ill patients, however, there is still few data. We aim to investigate the correlation between PI and serum lactate level in critically ill patients

Methods: A prospective observational study was performed. All patients who had signs of tissue hypoperfusion and serum lactate level ≥ 2 mmol/L were enrolled and received hemodynamic resuscitation based on standard clinical practice of KCMH. PI was measured by Masimo Radical-7®Pulse CO-Oximeter® and recorded values simultaneously with serum lactate at 0, 2, 6 and 24 hours during resuscitation

Results: Of the 42 critically ill patients (Table 1), we found significant correlation between PI and lactate level at 0 and 2 hours (r=-0.397, p=.009 and r=-0.311, p=.045 respectively). The change in PI also significantly correlate with lactate clearance at first 6 hours of resuscitation (r=0.444, p=.003 at 0-2 hour and r=0.370, p=.017 at 2-6 hours) (Fig 1). 24 patients(57%) had lactate clearance ≥ 10% within 2 hours, whereas 18 patients(42.8%) had not (Table 2, Fig 2). The cut-off value of increasing in PI<0.86 predicted patients who were not lactate clearance at 2 hours with sensitivity of 88.9% and specificity of 54.2%.(AUC 0.699, 95%CI 0.54-0.86, p=.029)

Conclusions: PI may has value as an adjunct continuous monitoring for peripheral tissue perfusion in early resuscitation period by using concurrently with serum lactate. Increasing of PI< 0.86 within 2 hours after resuscitation prompt us to do further management with our patients.

Table 1 (abstract P172). Patients characteristics
Table 2 (abstract P172). Comparison of parameters between lactate clearance group and lactate non-clearance group
Fig. 1 (abstract P172).

Relationship between change in PI and lactate clearance (%)

Fig. 2 (abstract P172).

Comparison of the change in PI between lactate clearance group and lactate non-clearance group


P174 Acute changes in lactate levels after bolus fluid treatment in critically ill patients

C Pierrakos 1, T Nguyen1, D Velissaris2, R Attou1, L Kugener1, J Devriendt1, PM Honore1, D De Bels1

1CHU-Brugmann, Brussels, Intensive Care, Brussels, Belgium; 2University Hospital of Patras, Internal Medicine Department, Patras, Greece

Introduction: Increased serum lactate is a frequent triggering factor for initiation of bolus fluid treatment (BFT). The aim of this study was to investigate the clinical utility of monitoring acute changes in serum lactate levels for assessing the response to BFT.

Methods: This is a prospective observational study in critically ill adult patients, including patients who received either 1 L crystalloids or 500 ml colloids within 30 to 40 min after lactate measurement. Cardiac index (CI) and serum lactate levels were measured before and after BFT. Patients with base-line levels of lactate <2mmol/L (Group 1), and patients with lactate ≥2mmol/L (Group 2) were analyzed separately. Responders to BFT were considered those who had an increase in CI ≥15%.

Results: We assessed 69 critically ill patients who received BFT. The patients’ mean age was 69±16 years, and the APACHE II score 22.7±8. Thirty-four patients had a baseline lactate value of <2 mmol/L, and 35 had ≥2mmol/L. Group 1: Fifteen patients (44%) responded to BFT (%dCI:37% ± 14). A statistically significant increase in lactate levels was observed in these patients (from 1.39 mmol/L ±0.3 to 1.65 mmol/L ±0.4, p<0.01) after the end of BFT. Lactate levels remained unchanged in non-responders (1.41 mmol/L ±0.4 versus 1.56 mmol/L±0.5, p=0.11). Group 2: Twelve patients (34%) responded positively to BFT (%dCI: 29% ±8) but had no significant changes in lactate levels (3.6 mmol/L ±0.4 versus 3.8 mmol/L ±0.5, p=0.32). Changes in lactate were also not observed in non-responders (3.9 mmol/L ±0.4 versus 4.1 mmol/L ±0.5, p=0.66).

Conclusions: Acute changes in lactate levels cannot be used for assessing responses to BFT. An increase in lactate levels can be observed in BFT responders likely because of a washout phenomenon. We did not observe any significant decrease in lactate concentration secondary to hemodilution effect in non-responders.

P175 Mitochondrial oxygen availability and regional saturation during CPB

E Mik, R Kortlever, M Ter Horst, F Harms

Erasmus MC - University Medical Center Rotterdam, Anesthesiology, Rotterdam, Netherlands

Introduction: Clinical measurement of mitochondrial oxygen tension (mitoPO2) has become available with the COMET system [1]. A question with any novel technique is whether it is feasible to use in clinical practice and provides additional information. In elective cardiac surgery patients we measured cutaneous mitoPO2 and tissue oxygenation (StO2).

Methods: Institutional Research Board approved observational study in patients undergoing cardiopulmonary bypass (CPB). mitoPO2 measurements were performed on the left upper arm (COMET, Photonics Healthcare B.V.) by oxygen-dependent delayed fluorescence of 5-aminolevulinic acid (ALA)-induced protoporphyrin IX [2]. Priming of the skin was done with ALA (Alacare, Photonamic GmbH) applied the evening before surgery. StO2 measurements (INVOS, Medtronic) were done in close proximity to the COMET sensor.

Results: At the time of writing 28 of 40 patients were enrolled and mitoPO2 measurements were feasible in this clinical setting. mitoPO2 appeared sensitive with a high dynamic range. For example, high-dose vasopressor therapy decreased mitoPO2 and blood transfusion increased a low mitoPO2 but not a high mitoPO2. In the example in Figure 1, mitoPO2 is clearly dependent on CPB flow and the restored cardiac circulation is able to maintain good cutaneous oxygenation after CPB even before returning of cellsaver blood. StO2 had the tendency to provide relatively stable values within a small bandwidth and little response to even major hemodynamic changes.

Conclusions: mitoPO2 shows the effect of interventions on mitochondrial oxygenation and provides additional information compared to standard monitoring and StO2.


1. Ubbink R et al. J Clin Mon Comput 31:1143-1150, 2017

2. Mik EG. Anesth Analg 117:834-846, 2013

Fig. 1 (abstract P175).

Example of mitoPO2 and StO2 during CPB

P176 Frailty assesment in cardiopulmonary arrest, is it necessary?

N Arriero Fernández, JA Silva Obregón, A Estrella Alonso, Z Eguileor Marin, MA Tirado Fernández, R Viejo Moreno, JE Romo Gonzales, C Marian Crespo

Hospital universitario de Guadalajara, Intensive care unit, Guadalajara, Spain

Introduction: The aim of this study is to describe the impact of frailty in cardiopulmonary arrest (CPA)

Methods: A descriptive, observational and retrospective study was carried out from January 2015 to June 2018. We analyzed interconsultations to ICU for CPA in this period. Age, sex, Clinical Frailty Scale(CFS), causes, limitation of CPR and outcomes were determined.

For statistical analysis of the data, quantitative variables are expressed as means +/-standard deviation; qualitative variables as percentages. The statistic analysis was made by using Chi scuare test. Multivariate analysis was performed using a logistic regression model.

Results: We analyze 106 patients, 18 of them were excluded. Frailty divide patients in 2 categories: prefrail and frail (PF-F)(CFS>3) and non-frail (NF)(CFS<4). The univariate analyses revealed no statistically significant difference between groups in age (mean PF-F age=75.98 years +/- 11.28; mean NF age=72.37 years+/-9.58; p=0.15) (Table 1). Most of patients were men (68.2%). Men were significant more likely to be frail (59% PF-F men vs 88,9% NF men OR 5.56 (CI95% 1.51-20.48) p=0.01). Main CPA causes were respiratory 46,6%; coronary heart disease (CHD) 17%; other causes 18,2%; unknow causes 18,2%. Patients suffering from CHD were significant less likely to be frail OR 0.28 (CI95% 0.08-0.97) p=0.04. CPR were stopped in 19.3%, patients who were limited were significant more likely to be frail (27.9% vs 0%; OR 2 (CI95% 1.01-2.63) p=0.04). 94.3% died. Patient that did not survive to discharge were significant more likely to be frail (98.4% vs 85.2% OR 10.43 (CI95% 1.11-98.35) p=0.04). Multivariate analysis included several variables. Only frailty was statistically significant (OR 11.35 (CI95% 1.03-125.52) p=0.04

Conclusions: Assessing frailty could predict patient mortality in CPA. It may help clinicians in decision-making

Table 1 (abstract P176). Univariate analysis

P177 Prehospital ventilation strategies during cardiopulmonary resuscitation

G Jansen, N Kappelhoff, K Götz

Evangelisches Klinikum Bethel, Anaesthesiology, Intensive Care and Emergency Medicine, Bielefeld, Germany

Introduction: While there is increasing evidence for the importance of chest compressions for the outcome of cardiopulmonary resuscitation (CPR), recommendations for ventilation strategies during CPR are rather poor. Although continuous ventilation at 10 breaths per minute and the avoidance of hyperventilation is suggested after intubation, the guidelines provide no information regarding the implementation of differentiated ventilation strategies (DVS) [1]. The present survey was designed to determine the DVS used in the prehospital setting by German emergency physicians (GEP) during CPR

Methods: An anonymous web-based questionnaire encompassing 10 questions was sent to GEP during September and October 2018. The EP were asked to specify the ventilator model, their preferred VS during CPR and the initial set of ventilator parameters, i.e.: volume- vs. pressure-controlled ventilation, respiratory rate, tidal volume, maximum and minimum pressure limits.

Results: 61 % of the questionnaires were completed (151/247). 95 % of respondents had preclinical access to respirators with the ability to perform DVS. 75 % of the participants chose volume-controlled, 19 % pressure-controlled ventilation modes during CPR. 6 % of participants indicated that they were performing manual ventilation. 53% chose a non-guidance-compliant ventilation with respiratory rates > 12/min. The corresponding group characteristics are shown in Table 1.

Conclusions: Despite the possibility of preclinical implementation of differentiated VS, GEP predominantly use volume-controlled ventilation modes. In contrast to the current guidelines, 53 % of GEP hyperventilate the patients under the CPR. Future studies are necessary to evaluate the influence of prehospital DVS on outcome of preclinical CPR.


[1] Perkins GD et al. Resuscitation 123:43-50, 2018

Table 1 (abstract P177). Prehospital ventilation strategies during cardiopulmonary resuscitation

P178 Traumatic asphyxia, focused on presence of cardiac arrest and the therapeutic reactivity

S Kikuta, S Ishihara, S Kai, H Nakayama, S Matsuyama, T Kawase, S Nakayama

Hyogo Emergency Medical Center, Department of Emergency and Critical Care Medicine, Chuo, Kobe, Hyogo, Japan

Introduction: Traumatic asphyxia is a rare condition in which breathing and venous return is impaired due to a strong compression to the upper abdomen or chest region, and induces swelling, purplish red appearance, and petechiae around the face and neck. To our knowledge, there are no reports describing details of traumatic asphyxia including the clinical course and the therapeutic reactivity from cardiac arrest. We focused on cardiac arrest among all traumatic asphyxia patients treated at our hospital, and investigated their clinical features and therapeutic reactivity.

Methods: Sixteen cases of traumatic asphyxia involved with our hospital between April 2007 and March 2018 were reviewed by using the pre-hospital activity record, medical record, and Hyogo prefectural inspection record. These patients were divided into three groups. The first group had already cardiac arrest at the time of rescue from the trapped place (Group A; 6 cases). The second group became cardiac arrest after the rescue (Group B; 5 cases). The third group did not experience cardiac arrest (Group C; 5 cases).

Results: All cases had abnormal findings in skin or conjunctiva (Table 1). Total mortality rate reached 56%, but among 11 cases of Group A and B who resulted in cardiac arrest, there were 10 cases with Injury Severity Score 16 or more and Abbreviated Injury Scale in the chest 3 or more. They had pneumothorax, flail chest, pericardial hematoma. Seven of them restored spontaneous circulation, and two cases achieved neurologically full recovery.

Conclusions: There are some cases of traumatic asphyxia whose therapeutic reactivity is very good even after cardiac arrest, so it is important not to spare efforts for life support in such cases.

Table 1 (abstract P178). Patients´ characteristics of each group

P179 Physiological derangements are cumulatively associated with poor outcome after cardiac arrest

Z Haxhija1, A Bilkanovic1, L Lucas2, J Dziodzio1, R R. Riker1, T May1, N Nielsen3, H Friberg4, D B. Seder1

1Maine Medical Center, Department of Critical Care Services and Neuroscience Institute, Portland, United States; 2Maine Medical Center, Center for Outcomes Research and Evaluation, Portland, United States; 3Helsingborg Hospital, Department of Anesthesia and Intensive Care, Helsingborg, Sweden; 4Skåne University Hospital, Department of Anaesthesia and Intensive Care, Lund, Sweden

Introduction: Hyperglycemia, hypotension, seizures, dysoxia and dyscarbia may contribute to secondary brain injury and have been identified as independent risk factors for poor outcome after cardiac arrest. It is unknown, however, if these physiological derangements have a cumulative association with outcome, and may represent an underappreciated treatment target.

Methods: We performed retrospective analyses of prospectively collected data from the International Cardiac Arrest registry. Demographics, clinical characteristics and physiological parameters from the first 24 hours were abstracted and assessed in bivariate and multivariable analyses, with standard confounders forced in. The primary outcome variable was delayed functional outcome, defined by the Cerebral Performance Status (CPC) dichotomized to good (CPC 1–2) or poor (CPC 3–5) outcome, measured about 6 months after resuscitation. Missing data were assessed by multiple imputation.

Results: Among the 2024 patients eligible for analysis, with a median ischemic time of 25 (15–36) minutes, 44% shockable initial heart rhythm and 75% witnessed arrest, five hundred ten (25%) patients had a good functional outcome at 6-months. Physiological derangements were each negatively associated with outcome in bivariate analysis at the p <0.001 level. A summary score of physiological derangements was included with potential confounders in the final regression model, and was independently associated with outcome with the chance of a good outcome decreasing by 46% for each increase of one physiologic derangement (95% CI 0.46–0.63).

Conclusions: Uncorrected physiological derangements are independently and cumulatively associated with worse outcome after cardiac arrest. Although causality cannot be established, it is reasonable to consider that the correction of physiological parameters may be an important step in the chain of survival after resuscitation.

P180 Characteristics and clinical outcomes of cardiac arrest patients admitted at the Intensive Care Unit from 2010-2016 in the Netherlands

L Mandigers1, F Termorshuizenb2, N De Keizerc3, D Gommers1, D Dos Reis Mirand1, W Rietdijk1, C Den Uila1

1Erasmus MC-University Medical Center, Department of Intensive Care Medicine, Rotterdam, Netherlands; 2Rivierduinen Psychiatric Institute Leiden, Department of Psychosis research, Leiden, Netherlands; 3University of Amsterdam, NICE foundation, Academic Medical Center, Amsterdam, Netherlands

Introduction: Worldwide, cardiac arrest remains a major cause of death. Many of the post cardiac arrest patients will be admitted at the Intensive Care Unit, however little is known about these patients. This patient group is relevant, because of the amount of patients and their vulnerability. For this reason, this study was set up to describe the characteristics and clinical outcomes of cardiac arrest patients admitted at the Intensive Care Unit.

Methods: We used data from the Dutch National Intensive Care Evaluation registry [1] from 2010-2016 in the Netherlands. All adult patients admitted at the Intensive Care Unit with cardiac arrest as admission diagnosis were included.

Results: In this study, characteristics and clinical outcomes of 26,056 cardiac arrest patients are described. Over time, the survival rate of Out-of-hospital cardiac arrest patients significantly decreases, while the survival rate of In-hospital cardiac arrest patients showed no significant relation.

Conclusions: This study shows a nationwide overview of the characteristics and outcomes of cardiac arrest patients admitted at the Intensive Care Unit. The decrease in mortality rate of out-of-hospital cardiac arrest patients may point to improvements made in post cardiac arrest care, however future research has to be done to specify the cause of this decrease.


1. NICE. Data in beeld. 2016;

P181 Outcomes following cardiac arrest at a UK cardiac arrest centre: a retrospective observational study

R Pugh1, L Pugh2, Z Habib3

1Glan Clwyd Hospital, Anaesthetics/ Intensive Care, Denbighshire, United Kingdom; 2Cardiff University, Cardiff University, Cardiff, United Kingdom; 3Glan Clwyd Hospital, Critical Care, Denbighshire, United Kingdom

Introduction: Glan Clwyd Hospital (GCH) was recently designated one of three Cardiac Arrest Centres for Wales. It has offered a 24/7 Percutaneous coronary angiography (PCI) service to a geographically dispersed North Wales population of approximately 690,000 since June 2017. Prior to this, urgent coronary angiography was available on a more limited basis to patients requiring PCI. The aim of this study was to investigate factors associated with hospital mortality after critical care admission following cardiac arrest.

Methods: Retrospective review of the Ward Watcher critical care database at GCH to identify patients who had undergone CPR in the 24 hours prior to critical care admission in 2013-18. Patients likely to have sustained OOHA of cardiac aetiology (OOHA-C) were identified from primary and secondary diagnoses and free text entry. Data were subsequently analysed using Excel and SPSS. The project was registered as a service evaluation with GCH Audit Department.

Results: There were 190 cardiac arrest admissions over this period, increasing from 25 in 2013-14 to 69 in 2017-18. Of these 122 were OOHA, of which 103 were considered OOHA-C. Although OOHA-C hospital mortality appeared to decrease over the time period (89%% to 56%), this was not statistically significant (p=0.149). Factors associated with survival to hospital discharge are presented in the Tables below. On logistic regression, only PCI and low pH within the first 24 hours of critical care remained statistically significant (p=0.027 and p<0.001 respectively).

Conclusions: Although we have been unable to make a distinction between patients presenting following STEMI and NSTEMI, and appreciating a potential influence of selection bias, the significant association between PCI and survival to hospital discharge supports the introduction of clinical pathways enabling PCI access following OOHA-C [1].


1. Nolan J, et al. Resuscitation.2015; 95: 202-222

Table 1 (abstract P181). Factors associated with survival to hospital discharge (categorical)
Table 2 (abstract P181). Factors associated with survival to hospital discharge (continuous, median)

P182 Determination of theoretical personalised optimum chest compression point with anteroposterior chest radiography

S Chon1, S Kim1, W Oh2, S Cho1

1CHA Bundang Medical Center, CHA University School of Medicine, Emergency medicine, Seongnam, South Korea; 2Kangwon National University School of Medicine, Internal Medicine, Chuncheon, South Korea

Introduction: Following the traditional assumption that the point (P_max.LV) beneath which the left ventricle (LV) is at its maximum diameter should be compressed to maximise stroke volume, we reported how to locate personalised P_max.LV using posteroanterior chest radiography. [1] Here, we aimed to derive and validate rules to estimate P_max.LV using anteroposterior chest radiography (chest_AP), which is performed for critically-ill patients urgently needing determination of personalised P_max.LV.

Methods: A retrospective, cross-sectional study was performed with non-cardiac arrest adults who underwent chest_AP and computed tomography (CT) within 1 h (derivation:validation=3:2). On chest_AP, we defined CD (cardiac diameter), RB (distance from right cardiac border to midline) and CH (cardiac height, from carina to uppermost point of left hemi-diaphragm) (Fig 1, 2). [1] Setting P_zero (0, 0) at the midpoint of xiphisternal joint and designating leftward and upward directions as positive on x and y axes, we located P_max.LV (x_max.LV, y_max.LV). The coefficients of the following mathematically-inferred rules were sought: x_max.LV=a0*CD-RB; y_max.LV=ß0*CH+γ . (a0: mean of (x_max.LV+RB)/CD; ß0, γ : representative coefficient and constant of linear regression model, respectively). [1]

Results: Among 360 cases (52.0±18.3 y, 102 females), we derived: x_max.LV=0.643*CD-RB and y_max.LV=55-0.390*CH (Table 1, Fig 2). This estimated P_max.LV (19±11) was as close as the averaged P_max.LV (19±11, P=0.13) and closer than three equidistant points representing the current guidelines (67±13, 56±10, and 77±17; all P<0.001) to the reference identified on CT, and thus was validated (all units: millimetre) (Table 2).

Conclusions: Personalised P_max.LV can be estimated with chest_AP. Further studies with actual cardiac arrest victims are needed to verify the safety and effectiveness of the rule.


1. Cho S et al. Resuscitation 2018;128:97-105.

Table 1 (abstract P182). The constants derived from the derivation set and those among total cases including both the derivation and validation set
Table 2 (abstract P182). The differences among estimated P_max.LV (x_max.LV, y_max.LV) using posteroanterior chest radiography, those of the reference P_max.LV measured on CT, and the three points along the lower sternal half with (top) at its top, (middle) at its middle and (bottom) representing the sites of chest compression recommended by the current CPR guidelines
Fig. 1 (abstract P182).

Identification of the theoretical optimum chest compression point (modified from a figure of reference 1)

Fig. 2 (abstract P182).

Estimation of the theoretical optimum chest compression point from anteroposterior chest radiography (modified from a figure of reference 1)


P184 Correlation between end-tidal CO2 and degree of compression of heart cavities measured by transthoracic echocardiography during cardiopulmonary resuscitation for out-of-hospital cardiac arrest

R Skulec, P Vojtisek, V Cerny

Masaryk hospital Usti nad Labem, Dept. of Anesthesiology, Perioperative Medicine and Intensive Care, Usti nad Labem, Czech Republic

Introduction: We realized clinical study to evaluate whether changes of the left ventricular (LV), right ventricular (RV) and inferior caval vein (IVC) diameter induced by chest compressions during cardiopulmonary resuscitation (CPR) for out-of-hospital cardiac arrest (OHCA) evaluated by transthoracic echocardiography correlate with end-tidal CO2 (EtCO2) levels.

Methods: 20 consecutive patients resuscitated for OHCA were included to the study. Transthoracic echocardiography was performed from subcostal view during ongoing chest compressions in all of them. This was repeated three times during CPR in each patient and EtCO2 levels were registered. From each investigation, video loop was recorded. Later on, maximal and minimal diameter of LV, RV and IVC were obtained from the recorded loops and compression index of LV (LVCI), RV (RVCI) and IVC (IVCCI) was calculated as (maximal-minimal/maximal diameter)x100. CImax defined as the value of LVCI or RVCI, whichever was greater was also assessed. Correlations between EtCO2 and LVCI, RVCI, IVCI and CImax were expressed as Spearman´s correlation coefficient (r).

Results: Evaluable echocardiographic records were found in 13 patients and a total of 39 measurements of all parameters were obtained. Chest compressions induced significant compressions of all observed cardiac cavities (LVCI=21.58±13.25%, RVCI=38.23±22.46%, IVCCI=62.26±30.84%, CImax=40.55±20.53%). We identified positive correlation between EtCO2 and LVCI (r=0.579, p<0.001), RVCI (r=0.752, p<0.001) and IVCCI (r=0.628, p<0.001). The most significant positive correlation has been found between EtCO2 and CImax (r=0.805, p<0.001) (Fig 1).

Conclusions: Evaluable echocardiographic records were reached in most of the patients. EtCO2 positively correlated with all parameters under consideration, while the strongest correlation was found between CImax and EtCO2. Therefore, CImax is a candidate parameter for real-time monitoring of haemodynamic efficacy of chest compressions during CPR.

Fig. 1 (abstract P184).

Correåation between CImax and EtCO2

P185 Table cut-off values of time interval between witness and first CPR for neurologically favourable 1-m survival in CC-only and CC+MMV first CPR groups

H Inaba, H Kurosaki, K Takada, A Yamashita, Y Tanaka

Kanazawa University Graduate School of Medicine, Department of Circulatory Emergency and Emergency Medical Science, Kanazawa, Japan

Introduction: Time interval between witness and first CPR (Bystander- or EMS-performed CPR, whichever performed earlier) is known to be one of critical time factors associated with survival from out-of-hospital cardiac arrest (OHCA). This study aimed to compare cut-off values of this time interval for neurologically favourable one-month survival between bystander-witnessed OHCA groups receiving chest compression-only (CC-only) and chest compression + ventilation (CC+V )first CPR.

Methods: From nationwide OHCA database during 2014-2016, we extracted 82,888 bystander-witnessed OHCA cases. The cut-off values for neurologically favourable one-month survival were determined using Yoden index in the two groups receiving CC-only and CC+V first CPR before and after propensity-score matching including characteristics of OHCA associated with bystander CPR.

Results: The matching procedure did not alter the cut-off values (Table 1). The values were 3 and 7 min in CC-only and CC+V first CPR groups, respectively. When analysed for subgroups, the difference in cut-off value was largest in subgroup of presumed non-cardiac aetiology: 3 vs. 8 min. A hazard model disclosed that effect of first CPR on survival was more largely diminished by delay in start of time first CPR in CC-only group than a in CC+V group: hazard ratio (95% CI), 1.89 (1.84-1.94). The overall survival rates (8.9% before matching and 8.6% after matching) in CC-only group were higher than those in CC+V group (5.6% and 6.2%). However odds ratio (OR) was small when adjusted by other factors associated with survival: adjusted OR (95% CI), 1.23 (1.15-1.32) before matching, 1.13 (1.09-1.22) after matching.

Conclusions: The effect of CC-only CPR on survival after OHCA was more largely diminished by delay in initiation than that of CC+ V CPR.

Table 1 (abstract P185). Cut-off values of time interval between witness and first CPR for neurologically favourable 1-M survival in CC-only and CC+MMV first CPR groups

P186 Post cardiac arrest management

S Chaudhry, M Ranganathan, M Bishop

South Warwickshire NHS Foundation Trust, Anaesthetics and Intensive Care Unit, Warwick, United Kingdom

Introduction: The UK Resuscitation Council has set out guidelines for management of patients post cardiac arrest [1]. This is in line with European Resuscitation Council guideline. We set out to find if we are following the guideline.

Methods: We did a retrospective audit over the course of 2 years looking at the data of 24 patients who had in hospital and/or out of hospital cardiac arrest and after the return of spontaneous circulation were admitted to the Intensive Care Unit (ICU). We focused on whether the care they received was as per the standards set by the UK Resuscitation Council.

Results: We had 11 in the hospital and 13 out of hospital cardiac arrests; 11 patients had less than 10 minutes of CPR, 9 had more than 10 minutes CPR and 4 patients the data was not recorded; 16 patients needed more than 61 minutes to reach from the site of arrest to the ICU. The partial pressure of carbon dioxide was >5.5 Kpa in 11 patients at two or more occasions. Target MAP was not documented in 9 patients; blood sugar target was not documented in 15 patients and was not maintained within limits in 5 patients. Target temperature was not documented in 10 patients. The withdrawal of treatment was not delayed for 72 hours in 1 patient out of 24. In 4 patients neurological tests were not documented. Multimodal assessment tools were not used in 1 patient. Electroencephalography and serum Neuron Specific Enolase were not used to diagnose brain deaths as they were not available at our trust. 11 patients were discharged, 12 died in the ICU and 1 died in hospital after discharge from ICU.

Conclusions: The audit reflected our local practice and showed that our mortality was in line with the acceptable limits; poor documentation of plan of care which posed problems in analyzing the care that these patients received; some of the parameters were not being maintained as set by UK Resuscitation guideline.


1. Nolan JP et al. Advance life support resuscitation council UK seventh edition 2016. Post-resuscitation care; 13: p 153 - 162.

P187 Chest compression quality depending on provider position during intra-hospital transportation

G Jansen1, K Kipker2, E Latka2, D Marx3, R Borgstedt1, S Rehberg1

1Evangelisches Klinikum Bethel, Anaesthesiology, Intensive Care and Emergency Medicine, Bielefeld, Germany; 2Studieninstitut Ostwestfalen-Lippe, Fachbereich Medizin und Rettungswesen, Bielefeld, Germany; 3Medi-Learn GbR, Kiel, Germany

Introduction: High-quality chest compressions (CC) with minimized interruptions are one of the most essential prerequisites for an optimal outcome of resuscitation. Therapy of reversible causes of cardiac arrest often requires intra-hospital transportation (IHT) during ongoing CPR. The present study investigated CC quality during transportation depending on the position of the provider.

Methods: 20 paramedics were enrolled into a manikin study with four groups: a reference group with the provider kneeling beside manikin on the floor (group 1), and 3 groups performing CC during a simulated IHT of 100 meters: walking next to the bed (group 2), kneeling beside the patient in bed (group 3, Fig. 1) or squatting above the patient in bed (group 4, Figure 2). Indicators of CC quality were measured as defined in the ERC Guidelines 2015 (pressure point and depth, compression frequency, complete relief, sufficient pressure depth) [1]. All 20 paramedics performed CC during each scenario (group 1-4).

Results: There were no statistical differences in quality of CC between groups 1, 3 and 4. Notably, group 2 performed significantly worse in respect to the proportion of CC with correct pressure point (p = 0.044 vs group 1), correct CC depth (p=0.004 vs. group 1, p=0.035 vs. group 3, p=0.006 vs. group 4). The results are shown in Table 1.

Conclusions: Carrying out guideline-compliant CC [1] during IHT is feasible with multiple provider positions. Based on the present results, kneeling or squatting position next to the patient (Figure 1 and 2) is recommended, whereas “walking next to the bed” while performing CC should be avoided.


[1] Perkins GD et al. Resuscitation 95:81-99

Table 1 (abstract P187). Indicators of chest compression quality
Fig. 1 (abstract P187).

Chest compressions in kneeling position while intra-hospital transportation

Fig. 2 (abstract P187).

Chest compressions in squatting position while intra-hospital transportation

P188 Ineffective ventricular and aortic compression during chest compression in CPR: resuscitative transesophageal echocardiography (TEE) findings

A Osman1, M Mohamed Sakan2, A Ahmad1, C Yen1, A Abdullah3

1Hospital Raja Permaisuri Bainun, Emergency Department, Perak, Malaysia; 2Hospital Raja Permaisuri Bainun, Emergency Medicine, Perak, Malaysia; 3Hospital Serdang, Cardiology Department, Selangor, Malaysia

Introduction: In cardiac arrest, resuscitative transesophageal echocardiography (TEE) provides continuous unobstructed visualisation of the heart during each compression in real-time fashion enabling instant feedbacks to improve the quality of subsequent chest compressions. The 2015 American Heart Association only regarded resuscitative TEE as an adjunct to find treatable causes and guide treatment decisions [1].

Methods: We presented a case series of cardiac arrest situations incorporating the use of resuscitative TEE to review the effectiveness of high quality chest compression following 2015 AHA recommendations.

Results: Four cardiac arrest patients underwent resuscitative TEE during CPR. Hands were placed at lower half of sternum and compressed to the depth of between 5 to 6cm. Mid esophageal four chamber view (to visualise left ventricular compression) and mid esophageal long axis view (to visualise aortic compression) were reviewed. In 3 out of 4 cases, chest compressions led to aortic outflow tract (LVOT) obstruction necessitating corrective hand relocations.

Conclusions: Chest compression is a blind intervention and external chest landmarks do not always correlate with the location of left ventricle [2]. Maximal compression at the recommended lower half of sternum might result in compression of the LVOT. Resuscitative TEE is a robust tool that should be equipped in centres that strive for better outcomes among cardiac arrest patients.


1. Brooks SC et al, Circulation. 2015;132:S436-S443

2. Qvigstad E et al, Resuscitation 2013.

P189 Abnormal movements but not agonal breathing is a concomitant sign associated with shockable initial rhythm and better outcomes in “sudden” cardiac arrest

H Inaba, H Kurosaki, K Takada, A Yamashita, Y Tanaka

Kanazawa University Grauate School of Medicine, Department of Circulatory Emergency and Emergency Medical Science, Kanazawa, Japan

Introduction: Emergency medical service (EMS) personnel may detect sudden onset of concomitant signs in EMS-witnessed out-of-hospital cardiac arrest (OHCA). This study aimed to investigate the incidences of concomitant signs and their associations with characteristics and outcomes of EMS-witnessed “sudden” OHCAs.

Methods: In addition to standard recommendation, EMS prospectively recorded concomitant signs which lead to start of advance life support in 385 cases with EMS-witnessed out-of-hospital cardiac arrest (OHCA) during the period of April 2012 to March 2018. Detailed database was completed after additional interview to EMS personnel. After excluding 268 cases with impending cardiac arrest, 117 cases were analysed.

Results: In addition to common and standard signs for cardiac arrest, the following signs were recorded as a concomitant signs which lead to start¡¡of advanced life support: abnormal movements of extremities and jaw which mimic tonic convulsion (17.9%), abnormal eye movement or position (3.4%), shout or groan (5.1%) and vomiting (5.1%). Agonal breathing was recorded in 29.1% (34/117) of all EMS-witnessed “sudden” OHCA cases and after cease of abnormal movements in 17.4% (4/22) of cases with abnormal movements. While abnormal movements were significantly associated with shockable initial rhythm (unadjusted OR; 95% CI, 4.39;1.56-12.3) and better neurologically favourable one-year survival (5.11;1.87¨C14.0), agonal breathing was not associated with shockable initial rhythm (0.59;0.22-1.63) or neurologically favourable survival (0.45;0.14-1.44). Multivariable analysis including all concomitant signs and agonal breathing confirmed the results of univariate analyses: adjusted OR (95% CI) of abnormal movement, 5.45 (1.82-16.7) for shockable rhythm, 4.16 (1.40-12.3).

Conclusions: Abnormal movements mimicking tonic convulsion are likely to be a concomitant sign associated with shockable initial rhythm and better outcomes in “sudden” cardiac arrest.

P190 Introduction of a treatment bundle improves post-cardiac arrest care by reducing variation in practice

C Platt, P Hampshire

Royal Liverpool University Hospital, Critical Care Department, Liverpool, United Kingdom

Introduction: This retrospective audit aimed to assess adherence to the post-cardiac arrest care bundle utilised by our intensive care unit (ICU). Introduced in 2017, following publication of International Liaison Committee on Resuscitation (ILCOR) guidance[1], the care bundle was designed to standardise care for this patient cohort which previously often varied according to the responsible clinician.

Methods: A retrospective review of clinical notes was undertaken for patients admitted to ICU following return of spontaneous circulation but whom remained comatose. This audit encompassed three-month periods before and after introduction of the care bundle in October 2017. Audit standards were assigned from target parameters documented in the bundle and reflected guidance from the Cheshire and Merseyside Critical Care Network.

Results: 39 patients were included in our audit; 15 admitted prior to and 24 admitted following implementation of the care bundle. In patients whom targeted temperature management was indicated, improved adherence to thermoregulation between 34-36°C was observed (50 vs 71%). Significant improvements were since in the observance to target values for oxygen saturation (0 vs 70.8%, p=0.0023) and mean arterial pressure (20 vs 95.8%, p<0.0001) following the introduction of the care bundle. Improved observance of ventilation targets was also seen; maintenance of PaCO2>4.5 kPa (40 vs 75%, p=0.0441) and tidal volumes <8 ml/kg ideal body weight (0 to 37.5%, p=0.0069).

Conclusions: The introduction of a post-cardiac arrest care bundle in our ICU has improved care by providing discrete physiological targets to guide nursing staff and standardising management between clinicians. Variations in care are associated with poorer patient outcomes [2] and introduction of this bundle has reduced disparities in practice.


1. Nolan JP et al. Resuscitation 95:202-222, 2015.

2. Peterson ED et al. JAMA 295:1912-1920, 2006.

P191 Cardiac causes and outcome of out-of-hospital cardiac arrest – single centre experience

P Pekic1, M Bura2, N Maric3

1University Hospital “Sveti Duh”, Department for cardiovascular disease - Cardiac intensive care and arrhythmology unit, Zagreb, Croatia; 2Neuropsychiatric hospital “dr. Ivan Barbot”, Internal Medicine, Popovača, Croatia; 3University Hospital “Sveti Duh”, Intensive Care Unit, Zagreb, Croatia

Introduction: Out-of-hospital cardiac arrest (OHCA) is a major health problem worldwide with an annual incidence of approximately 9,000 in Croatia. Cardiac causes of OHCA can be attributable to a wide array of cardiac diseases and reported survival rate is low in spite of advances in resuscitation and EMS services.

Methods: Single-centre retrospective study analyzed outcomes of 42 OHCA patients admitted to cardiac ICU between 2010.-2015. We studied demographic data, initial rhythm, type of CPR, comorbidities and various post admission diagnostic findings in order to identify their impact on survival.

Results: OHCA comprised 0,5% of all admissions. Mean LOS was 8.24 days (0-64). Mean age was 65,4y (29-90), M: F ratio 29:13 and bystander CPR was performed in only 19% OHCA patients. The most common initial rhythm was VF (45.2%), followed by VT (16.6%), PEA was found in 2,9% and asystole in 1.9% of pt More than half of pt received adrenalin (55%) and defibrillation (57%) and only 17% required a temporary pacemaker. 38% of pt had an ECG consistent with MI after ROSC, 19% underwent coronary angiography resulting in PCI in 75% of cases. In 5 pt (12%) therapeutic hypothermia protocol was performed. Most OHCA pt had hypertension (60%) and hyperlipidaemia (69%) as the most common risk factors followed by cardiomyopathy (33%), diabetes (29%) and CAD (21%). Only 9% had a preexisting significant valvular disease and the rest were extracardial comorbidities: chronic renal disease (17%), COPD (9%) and cerebrovascular disease (9%). 14 patients survived (33%) and GCS on admission was the only significant impact factor on survival along with comorbidities (mean GSC was 6 in survivors vs. 3 in deceased). Interestingly, age, initial rhythm, troponin I level, pH and therapeutic hypothermia had no impact on survival.

Conclusions: Our data demonstrate the importance of early on-site resuscitation as the most important factor of neuroprotection and outcome and puts an emphasis on the importance of CPR education for layman population.

P192 Prediction of acute coronary ischaemia and angiographic findings in patients with out-of-hospital cardiac arrest

J Higny1, A Guédès2, C Hanet2, V Dangoisse2, L Gabriel2, J Jamart3, C De Meester De Ravenstein4, E Schroeder2

1CHU UCL NAMUR, Pathologie Cardiovasculaire, Soins Intensifs, Yvoir, Belgium; 2CHU UCL NAMUR, Pathologie Cardiovasculaire, Yvoir, Belgium; 3CHU UCL NAMUR, Unité de Support Scientifique, Yvoir, Belgium; 4Pôle de Recherche Cardiovasculaire (CARD), Institut de Recherche Expérimentale et Clinique (IREC), Cliniques Universitaires Saint-Luc, Bruxelles, Belgium

Introduction: Coronary artery disease (CAD) is the leading cause of out-of-hospital cardiac arrest (OHCA). However, diagnosis of acute coronary ischaemia (ACI) remains challenging, particularly in patients without ST-segment elevation on the post-resuscitation ECG. In this regard, a consensus statement recommends the implementation of a work-up strategy in the emergency room (ER) to exclude non-coronary causes of collapse within 2 hours.

Methods: Retrospective single-centre study performed on 64 consecutive patients with resuscitated OHCA who underwent a diagnostic coronary angiography (CA). We present data on coronary angiograms for patients who underwent cardiac catheterization after resuscitation. Afterwards, we sought to identify parameters associated with ACI.

Results: ST-segment elevation was noted in 29 patients (45%). ST-segment depression or T-wave abnormalities were noted in 35 patients (55%). Invasive coronary strategy allowed to identify an acute culprit lesion in 46 cases (72%). 29 patients with ST-segment elevation underwent an immediate angioplasty for an acute coronary occlusion. 17 patients without ST-segment elevation underwent an ad hoc percutaneous coronary intervention for a critical lesion. Stable CAD was found in 9 cases (14%) and a normal angiogram was found in only 9 cases (14%) (Figure 1). The independent predictors of ACI were convertible rhythm (OR 16.02; 95% CI 4.48-57.29), personal history of CAD (OR 15.12; 95% CI 4.19-54.53) and presence of at least 2 cardiovascular risk factors (CVRF) (OR 10.68; 95% CI 2.55-44.74) (Table 1).

Conclusions: ACI was the leading precipitant of collapse. ST-segment elevation was highly predictive of coronary occlusion. In addition, a culprit coronary lesion was identified in nearly 50% of patients undergoing CA despite the lack of ST-segment elevation. Finally, our findings suggest that the identification of risk criteria may help to improve the recognition of ACI after OHCA.

Table 1 (abstract P192). Univariate and multivariate analysis for the prediction of ACI in patients with OHCA
Fig. 1 (abstract P192).

Angiographic findings in patients in patients undergoing cardiac catheterization after OHCA (n = 64)

P193 The prediction of outcome for in-hospital cardiac arrest (PIHCA) score

E Piscator1, K Göransson1, S Forsberg1, M Bottai2, M Ebell3, J Herlitz4, T Djärv1

1Karolinska Institutet, Center for Resuscitation Science, Department of Medicine Solna, Stockholm, Sweden; 2Karolinska Institutet, Unit of Biostatistics, Department of Environmental Medicine (IMM), Stockholm, Sweden; 3College of Public Health, University of Georgia, Department of Epidemiology and Biostatistics, Athens, United States; 4University of Borås and Sahlgrenska Academy, University of Gothenburg, Center of Prehospital Research, Faculty of Caring Science, Work-life and Welfare, University of Borås and Department of Molecular and Clinical Medicine, Sahlgrenska Academy, University of Gothenburg, Borås and Gothenburg, Sweden

Introduction: A do-not-attempt-resuscitation (DNAR) order may be issued when it is against the wishes of the patient to receive cardiopulmonary resuscitation (CPR), or when CPR is considered medically futile; that is when the chances of good outcome are minimal. A prearrest prediction tool can aid clinicians in consolidating objective findings with clinical judgement in the decision process for DNAR orders. Our aim was prediction model update of the Good Outcome Following Attempted Resuscitation score for favorable neurologic survival after in-hospital cardiac arrest (IHCA).

Methods: The prediction model update was based on a retrospective cohort of 717 adult IHCAs in Stockholm County 2013-2014 identified through the Swedish Cardiopulmonary Resuscitation Registry. It included redefinition and reduction of predictor variables, and addition of the predictor variable chronic co-morbidity, assessed as Charlson Comorbidity Index, to create a full model consisting of 9 predictors. The outcome was favorable neurologic survival defined as Cerebral Category Score ≤2 at discharge. The full model was recalibrated based on 1000-bootstrap resampling calibration. Model performance was evaluated with the area under the receiver operating curve (AUROC), calibration and classification accuracy with a cutoff of 3% likelihood of favorable neurologic survival.

Results: We call the updated prediction model the Prediction of outcome for In-Hospital Cardiac Arrest (PIHCA) score. It had an AUROC of 0.81 (95% CI:0.81 to 0.81) and showed evidence of good calibration, see calibration plot in Figure. Predictive value for classification into <3% likelihood of favorable neurologic survival was 97.4%. False classification into < 3% likelihood of favorable neurologic survival was 0.57%.

Conclusions: The PHICA score has potential to be used as an aid for objective prearrest assessment of the chance of favorable neurologic survival after IHCA, as part of decision making for a DNAR order.

Fig. 1 (abstract P193).

Calibration plot for the Prediction of outcome for In-Hospital Cardiac Arrest (PIHCA) score

P194 Extracorporeal membrane oxygenation and cytokine adsorption in post-resuscitation disease

P Morimont, T Desaive, T Amand, A Rego, J Koch, M Lagny, S Habran, B Lambermont

University Hospital of Liege, Medical Intensive Care, Liege, Belgium

Introduction: Prognosis of survival in patients with cardiac arrest remains poor. During and after cardiopulmonary resuscitation, pathophysiological disturbances in relation with a cytokine storm, are described as “post-resuscitation” disease like a combination of cardiogenic and vasodilatory shocks. Veno-arterial extracorporeal membrane oxygenation (VA ECMO) allows to restore adequate perfusion but little is known about its effect on left ventricular (LV) function and about the role of cytokines.

Methods: This study was performed in an experimental model of cardiac arrest performed in 3 groups of 3 anesthetized and mechanically ventilated pigs. Cardiac arrest was obtained by application of electrical current to epicardium inducing ventricular fibrillation. After a no-flow period of 5 minutes, medical resuscitation with catecholamines and vasopressors was performed in “CONTROL” group while VA ECMO was started in “ECMO” group and VA ECMO in combination with CYTOSORB (extracorporeal blood purification therapy designed to reduce excessive levels of inflammatory mediators such as cytokines) was started in “ECMO-CYTO” group. LV function was assessed with transthoracic echocardiography and arterial pressure with aortic pressure catheter.

Results: Hemodynamic stability was obtained after 22 ± 6 and 24 ± 7 minutes in ECMO and ECMO-CYTO groups, respectively. No return of spontaneous circulation was observed in CONTROL group. At 15 minutes following cardiac arrest, LV area fractional change on short axis was normalized in ECMO and ECMO-CYTO groups (31± 3 and 34 ± 4 %, respectively). Vasopressor requirements were significantly lower in ECMO-CYTO group than in ECMO group.

Conclusions: After cardiac arrest (no-flow) of 5 minutes duration, VA ECMO allowed complete LV recovery and hemodynamic stability within 30 minutes of “post-resuscitation” disease. CYTOSORB added to VA ECMO could contribute to reduce post-resuscitation vasodilatation.

P195 Impact of rapid response car system on ECMO in out-of-hospital cardiac arrest: a retrospective cohort study

M Nasu1, R Sato2, K Takahashi1, Y Kitai1, Y Kitahara1, T Hozumi1, H Fukui1, T Yonemori1

1Urasoe General Hospital, Emergency and Critical Care, Urasoe, Japan; 2Department of Internal Medicine, John A. Burns School of Medicine, University of Hawaii at Manoa, Department of Internal Medicine, Hawaii, United States

Introduction: Extracorporeal life support (ECLS) has been reported to be more effective than conventional cardio-pulmonary resuscitation (CPR). In ECLS, a shorter time from arrival to implantation of extracorporeal membrane oxygenation (ECMO; door-to-ECMO) time has been reported to be associated with better survival rates. This study aimed to examine the impact of the physician-based emergency medical services (P-EMS) using a rapid response car (RRC) on door-to-ECMO time in patients with out-of-hospital cardiac arrest (OHCA).

Methods: In this retrospective cohort study, adult patients with OHCA who were admitted to a Japanese tertiary care hospital from April 2012 to November 2018 and underwent veno-arterial ECMO were included. Patients were either transferred by emergency medical service (EMS only group) or with physician using RRC (RRC group). Primary outcome was door-to-ECMO time. Secondary outcome was onset-to-ECMO time, 24-hour survival, 30-day survival and neurologically favorable outcome. Wilcoxon rank-sum test was used to compare the outcome between the two groups.

Results: A total of 46 patients were included in this study, and outcome data were available for all patients. The door-to-ECMO time was significantly shorter in the RRC group than in the EMS only group (median, 20 min vs. 33 min; P =0.004). Additionally, the onset-to-ECMO time was significantly shorter (median, 50 min vs. 60 min; P =0.005). There was no significant difference in the 24-hour survival, 30-day survival rates and neurologically favorable outcome.

Conclusions: The physician-based RRC system was associated with a shorter door-to-ECMO time. Combination of the RRC system with ECLS may lead to better outcomes in patients with OHCA.

P196 The evaluation of ICU outcome of extracorporeal cardiopulmonary resuscitation (ECPR) for adult patients with out-of-hospital cardiac arrest (OHCA)

R Hokama, S Miyakawa, M Koie, M Nakashio, J Maruyama, Y Irie, K Hoshino, H Ishikura

Fukuoka university hospital, Department of Emergency and Critical Care Medicine, Fukuoka city, Japan

Introduction: Although there have been some reports to examine the indication and prognostic factors of ECPR, the indication for ECPR has not been established in patients with OHCA. The purpose of this study was to establish prognostic factors for ECPR in adult patients with OHCA.

Methods: This retrospective, observational study was conducted on OHCA patients aged over 18 who were delivered to the ER from January 2001 to December 2017. And we enrolled the patients who were performed ECPR immediately after arrival and survived 24 hours or more. These patients were divided into ICU survival groups (S) and non-survival groups (NS) and evaluated of ICU outcome of ECPR. We reviewed medical records collected at the time of admission and compared demographic data and experimental data between the two groups.

Results: In this study, 70 patients were included, 22 patients (31.4%) were S (median age 55 years [IQR, 42-64 years]) and 48 patients were NS (median age 59 years [IQR, 53-65 years]). There was a significant difference between S and NS with the ratio of witnesses [S 21(95.5%) vs non-survivor 37(77.1%), p=0.013], the ratio of abnormal pupil diameter (5mm or more, 2mm or less)[S 7(31.8%) vs NS 36(75.0%), p<0.001] and a presence of pupillary light reflex[S 9(40.9%) vs NS 3(6.3%), p<0.001]. Multivariable logistic regression analysis performed for 5 explanatory variables (pupillary light reflex, dilated pupils (5mm or more), normal pupil diameter (3-4mm), the witness, and ventricular fibrillation), indicated pupillary light reflex (OR, 6.1; 95% CI, 1.14-32.50, p=0.03) and dilated pupils (OR, 0.06; 95%CI, 0.004-0.889, p=0.04) to be an independent prognostic factor.

Conclusions: In this study, we suggested that pupil findings on hospital arrival might become predictive factors of ECPR for adult patients with OHCA.

P197 Spontaneous CPR simulations in the ICU

N Tabak, Y Polishuk, E Kishinevsky, G Bregman

Kaplan Medical Center, ICU, Rehovot, Israel

Introduction: Quick identification and professional treatment of conditions requiring CPR must be practiced by all staff, with refresher simulations providing updated knowledge. Spontaneous CPR simulations lead to faster staff response times, improves team work, and increases the effectiveness of treatment during real CPR situations. Providing refresher resuscitation courses are not enough, and lack the advantages of spontaneous simulations, specifically practicing performing CPR.

Methods: The simulations all based on American Heart Association protocols and based on clinical situations of typical ICU patients. The simulations take place in a room with specialized equipment with feedback given immediately to participants. The participants of the simulations are Nursing staff and Doctors that are currently on shift, plus trainees and medical students. The simulations are run by experienced staff members. The simulations are held once a week covering all shifts. Judgement of staff performance uses a tool based on AHA protocols and checks effectiveness of CPR management, specifically response time, correct usage of equipment and devices, performance of chest compressions, intubation, drug management and team work.

Results: During the period between 03/17 - 10/18 47 spontaneous CPR simulations were performed in the ICU, the majority on the morning shift. During this period at least 57% of Nursing staff and 66% of Medical staff practiced CPR simulations at least twice, with an overall improvement in performance of 25%, improved identification of conditions requiring CPR, quicker response times, use of appropriate equipment and increased effective team work.

Conclusions: Improving resuscitation skills needs practice and team work, a continuous process which needs to include all Nursing and Medical staff and involves repeat CPR simulations. There is a need to expand simulations to cover all shifts. Staff responded positively to this process and reported increased confidence when managing real life CPR situations.

P198 Optimized infrastructure and regular training reduces time-to-extracorporeal life support for in hospital cardiac arrest

M Suverein1, A Bruekers1, T Delnoij1, R Lorusso2, M Bol1, J Sels1, P Weerwind2, P Roekaerts1, J Maessen2, M Van de Poll1

1Maastricht UMC, Intensive Care, Maastricht, Netherlands; 2Maastricht UMC, Cardiothoracic Surgery, Maastricht, Netherlands

Introduction: ECPR requires rapid activation of a multidisciplinary team and a well-organized ECPR infrastructure [1]. In May 2017, we started optimising our ECPR infrastructure for out-of-hospital cardiac arrest by introducing a dedicated ECPR cart, regular training sessions and simulation cases, and by evaluating all procedures. This infrastructure is also used for in-hospital cardiac arrest (IHCA). The objective of this retrospective analysis was to determine whether these changes had an effect on the time to extracorporeal life support (ECLS) in IHCA as well.

Methods: A single-centre, retrospective review was conducted a year after the start of training. All adult patients (≥18 years) with a witnessed, refractory IHCA who received ECPR from 2013 to 2018 were included. The primary outcome was time-to-ECLS, defined as time from start of arrest to adequate flow on ECLS. Data are presented as median [IQR]. Statistical analysis was performed using a Mann-Whitney U test.

Results: In the first period (2013 – April 2017), 13 patients received ECPR with a median age of 62 [60-67], 69.2% were male and with a Charlson Comorbidity Index (CCI) of 2.5 [2.0-3.0]. In the second period (May 2017 – May 2018), 4 patients received ECPR, with a median age of 53 [33-60], 50% were male and with a CCI of 1.5 [0.5-1.5]. The time-to-ECLS was significantly reduced from 40 minutes [25-45] to 27 minutes [22-33] (p = 0.019) (Fig 1). Successful weaning of mechanical circulatory support remained similar for both time periods (1st 46.2 % vs. 2nd 50%).

Conclusions: In conclusion, optimisation of the ECPR infrastructure significantly reduced time-to-ECLS, which may ultimately improve clinical outcome after ECPR for IHCA.


[1] Swol, J., et al. (2016). Perfusion 31(3): 182-188.

Fig. 1 (abstract P198).

Boxplot of time-to-ECLS (min) in the first and second period (p = 0,019)

P199 Interest of the “high fidelity” medical simulation as a pedagogical tool in training of medical residents

W Sellami, I Ben mrad, Z Hajjej, M Chniti, I Labbene, M Ferjani

Department of critical care medicine and anesthesiology Military Hospital, Tunis, Tunisia

Introduction: Medical simulation has become an essential teaching method for all health professionals. It is effective in terms of learning and improves skills. However, most studies have focused only on immediate benefits and short-term retention of skills. Main objective: To study the interest and the educational contribution in the short and medium term of medical simulation compared to a classical training.

Methods: Cohort, prospective, observational, single-center, randomized study with control group including 30 residents (20 in anesthesia resuscitation and 10 in emergency medicine). All benefited from a theoretical training with a reminder of the latest recommendations on the management of cardiac arrest and anaphylactic shock. They were randomized into 2 groups and received practical training on a high-fidelity simulator for the management of either cardiac arrest (ACC group) or anaphylactic shock (CA group). Each group was evaluated at 6 weeks (T0) and at 6 months on two scenarios: refractory ventricular fibrillation (FV) scored on 20 points and grade 3 anaphylactic reaction (RA3) scored on 30 points. Each group served as the control group for the pathology in which they did not receive specific simulator training. The results are expressed on average with their standard deviations with “p” <0.05.

Results: 30 residents were included and randomized into 2 groups of 15. The ACC and CA groups scored significantly higher in their respective 2-time scenarios: T0 (FV: ACC = 18.90 [17.8-20] vs CA = 9.35 [6.66-12.04]; RA3: CA = 26.5 [23.00-30] vs ACC = 26.28 [9.56-23]), T1 (FV: 18.0 [16.01-20] vs 15.03 [12,33-17,73] RA3: 28.50 [27-30] vs 21.3 [15.5-27.06]).

Conclusions: Simulator-trained participants have significantly better results than untrained ones, at both 6 weeks and 6 months.

P200 Benefits of simulation as a teaching tool in postgraduate education in emergency medicine “cardiac arrest scenario”

M Khrouf, H Sandid, K Hraiech, F Boukadida, Y Bel haj, Z Mezgar, M Methamem

Hopitall Farhat Hached 4000 Rue Ibn AL Jazzar, Sousse, Tunisia

Introduction: Simulation is a tool for improving the quality and safety of care, and its recognized as an essential method of evidence-based education. Emergency Medicine is a discipline in which there is a constant concern for the safety of patients. The emergency physician is often called upon to take charge of critical situations that use knowledge, know-how and knowledge as skills that must be mastered and whose theoretical learning alone is insufficient.

Methods: It´s a prospective study including residents in emergency medicine performing their specialty courses in emergency services and emergency medical assistance in the region of Sousse from January to June 2018. They were randomized into two groups: the one benefiting from a traditional education and the other from an education based on simulation sessions. The chosen scenario was the management of a cardiac arrest. A pre-test and a post-test were performed in both groups.

Results: We included 30 emergency residents who did not receive specialized training in the management of cardiac arrest, there was a female predominance with an average age of 27, there was no significant difference regarding the pretest between the two groups with 10.08 There was no significant difference with respect to the pre-test score between the two groups 10.08 ± 2.7 / 20 for the control group versus 10.34 ± 3.3 / 20 for the simulation group. There was a significant progression after the course with an average post-test score of 13.87 ± 1.8 in the simulation group while this score was 11.94 ± 2.3 in the control group with a statistically significant difference (p <0.001).

Conclusions: Simulation learning has led to a better acquisition of cognitive knowledge by learners. The simulation is not intended to replace bed-based teaching, nor theoretical or faculty teaching, but it is an essential complement . In Tunisia, the simulation must continue its current integration in the initial and continuous training of doctors.

P201 mTOR mediates neuronal death following transient global cerebral ischemia in the striatum of chronic high-fat diet-induced obese gerbils

JK Seo1, HJ Lim1, JB Moon1, TG Ohk1, MC Shin1, KE Kim1, MH Won2, JH Cho3

1Kangwon National University, Emergency Medicine, Chuncheonsi, South Korea; 2Kangwon National University, Neurobiology, Chuncheonsi, South Korea; 3Kangwon National University, Department of Emergency Medicine, Chuncheonsi, South Korea

Introduction: Recent studies have shown that obesity and its related metabolic dysfunction exacerbates outcomes of ischemic brain injuries in some brain areas, such as the hippocampus and cerebral cortex when subjected to transient global cerebral ischemia (tGCI). However, the impact of obesity in the striatum after tGCI has not yet been addressed. The objective of this study was to investigate the effects of obesity on tGCI-induced neuronal damage and inflammation in the striatum and to examine the role of mTOR which is involved in the pathogenesis of metabolic and neurological diseases.

Methods: Gerbils were fed with a normal diet (ND) or high-fat diet (HFD) for 12 weeks and then subjected to 5 min of tGCI. HFD-fed gerbils showed the significant increase in body weight, blood glucose level, serum triglycerides, total cholesterol, and low-density lipoprotein cholesterol without affecting food intake.

Results: In HFD-fed gerbils, neuronal loss occurred in the dorsolateral striatum 2 days after tGCI and increased neuronal loss were observed cholesterol days after tGCI; however, no neuronal loss was the in ND-fed gerbils after tGCI, as assessed by neuronal nuclear antigen immunohistochemistry and Fluoro-Jade B histofluorescence staining. The HFD-fed gerbils also showed severe activated microglia and further increased immunoreactivities and protein levels of tumor necrosis factor-alpha, interukin-1beta, mammalian target of rapamycin (mTOR) and phosphorylated-mTOR in the striatum during pre- and post-ischemic conditions compared with the ND-fed gerbils. In addition, we found that treatment with rapamycin, a mTOR inhibitor, in the HFD-fed gerbils significantly attenuated HFD-induced striatal neuronal death without changing physiological parameters.

Conclusions: These findings reveal that chronic HFD-induced obesity results in severe neuroinflammation and significant increase of mTOR activation, which could contribute to neuronal death in the stratum following tGCI. Abnormal mTOR activation might play a key role.

P202 Associations between partial pressure of oxygen and neurological outcome in out-of-hospital cardiac arrest patients

F Ebner1, S Ullén2, A Åneman3, T Cronberg4, N Mattsson4, H Friberg1, C Hassager5, J Kjærgaard5, M Kuiper6, P Pelosi7, J Undén1, M Wise8, J Wetterslev9, N Nielsen1

1Lund University, Helsingborg Hospital, Department of Clinical Sciences Lund, Anesthesia and Intensive Care, Helsingborg, Sweden; 2Skane University Hospital, Clinical Studies Sweden, Lund, Sweden; 3Liverpool Hospital, Department of Intensive Care, Sydney, Australia; 4Lund University; Skane University Hospital, Department of Clinical Sciences Lund, Neurology, Lund, Sweden; 5University of Copenhagen, Department of Cardiology and Department of Clinical Medicine, Rigshospitalet, Copenhagen, Denmark; 6Leeuwarden Medical Centrum, Intensive Care Unit, Leeuwarden, Netherlands; 7University of Genoa, Department of Anesthesia and Intensive Care, IRCCS San Martino Policlinico Hospital, Genoa, Italy; 8University Hospital of Wales, Adult Critical Care, Cardiff, United Kingdom; 9University of Copenhagen, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen, Denmark

Introduction: Exposure to hyperoxemia and hypoxemia is common in out-of-hospital cardiac arrest (OHCA) patients following return of spontaneous circulation (ROSC) but its effects on neurological outcome are uncertain and study results are inconsistent.

Methods: Exploratory post-hoc substudy of the Target Temperature Management (TTM) trial [1], including 939 patients after OHCA with ROSC. The association between serial arterial partial pressures of oxygen (PaO2) during 37 hours following ROSC and neurological outcome at 6 months, evaluated by Cerebral Performance Category (CPC), dichotomized to good (CPC 1-2) and poor (CPC 3-5), was investigated. In our analyses, we tested the association of hyperoxemia PaO2 > 40 kPa and hypoxemia PaO2 < 8 kPa, time weighted mean PaO2, (TWM-PaO2) (Fig 1), maximum PaO2 difference (Δ PaO2) and gradually increasing PaO2 levels (13.3 - 53.3 kPa) with poor neurological outcome. A subsequent analysis investigated the association between PaO2 and a biomarker of brain injury, peak serum Tau levels.

Results: 869 patients were eligible for analysis. 300 patients (35%) were exposed to hyperoxemia or hypoxemia after ROSC (Table 1). Our analyses did not reveal a significant association between hyperoxemia, hypoxemia, TWM-PaO2 exposure or Δ PaO2 and poor neurological outcome at 6-month follow-up after correction for co-variates (all analyses p= 0.146 - 0.847) (Fig 2). We were not able to define a PaO2 level associated with the onset of poor neurological outcome. Peak serum Tau levels at either 48 or 72 hours after ROSC were not associated with PaO2.

Conclusions: Hyperoxemia or hypoxemia exposure occurred in one third of the patients during the first 37 hours of hospitalization and was not significantly associated with poor neurological outcome after 6 months or with the peak s-Tau levels at either 48 or 72 hours after ROSC.


1. Nielsen et al. N Engl J Med. 2013;369:2197-206.

Table 1 (abstract P202). Outcome according to exposure groups in primary and secondary analyses
Fig. 1 (abstract P202).

Explanatory illustration depicting the primary and secondary analyses. The grey area under the chart represents the area from which PaO2-TWM was calculated. PaO2 = partial pressure of oxygen. kPa = kilopascal. T -1 = first blood gas analysis after admission but before inclusion into the TTM-trial. T0 to T36 = time points (hours) for protocolized blood gas sampling

Fig. 2 (abstract P202).

Adjusted OR´s (bullet points) with 95% CI´s (horizontal lines) for poor neurological outcome according to Cerebral Performance Category (CPC) across stepwise increasing PaO2 threshold values. OR below 1.0 indicates better outcome above the PaO2 threshold

P203 Clinical predictors of early and late mortality after cardiac arrest: a retrospective multicenter cohort study

P Kurtz1, C Storm2, M Soares3, F Bozza3, U Melo4, D Siefeld Araya5, B Mazza6, M S. Santino7, JL Salluh3

1Paulo Niemeyer State Brain Institute, Neuro ICU, Rio de Janeiro, Brazil; 2Charité – Universitätsmedizin, Berlin, Germany; 3D´Or Institute for Research and Education, Rio de Janeiro, Brazil; 4Hosp Estadual Alberto Torres, São Gonçalo, Brazil; 5Hospital Santa Paula, São Paulo, Brazil; 6Hospital Samaritano, São Paulo, Brazil; 7Hospital Barra d’Or, Rio de Janeiro, Brazil

Introduction: Data on outcome of survivors of cardiac arrest patients in low and middle-income countries is scarce. The purpose of this study was to investigate clinical predictors of hospital mortality in patients resuscitated after cardiac arrest in a large sample of Brazilian ICUs.

Methods: We performed a retrospective cohort study of survivors from cardiac arrest in 57 hospitals in Brazil, during 2014 and 2015. We retrieved patients’ clinical and outcome data from an electronic ICU quality registry. We used multivariable logistic regression analysis to identify factors associated with hospital mortality.

Results: A total of 2295 patients were included. Both patients with primary admission diagnosis of cardiac arrest (n = 800, 35%) and those that arrested on admission to the ICU (65%) were included. Median age was 67 (IQR 54 – 79) and 53% were male. Median SAPS 3 was 70 (57 – 83), SOFA was 10 (7 – 13) and hospital mortality was 83%. Among nonsurvivors, 47% died in the first 48h of ICU admission (early mortality). Only 1% received therapeutic hypothermia and 6% underwent withhold/withdrawal of life support. After adjusting for SAPS 3 and SOFA, early mortality was associated with hemodynamic compromise (systolic blood pressure <100 mmHg, lactate >4 mmol/L and use of vasopressors) and hypo or hyperthermia. Late mortality was independently associated with delayed admission to the ICU (>48h) and palliative care.

Conclusions: This large cohort of post cardiac arrest survivors demonstrates extremely high mortality rates and negligible implementation of hypothermia in Brazil, with half of nonsurvivors dying within 48 hours of admission with severe hemodynamic compromise and organ dysfunction. Late mortality was related to delayed admission to ICU and withdrawal/withholding measures.

P204 Cerebral oxygenation after out-of-hospital cardiac arrest: association with neurological outcome

P Jakkula1, J Hästbacka1, M Reinikainen2, V Pettilä1, P Loisa3, M Tiainen4, M Bäcklund1, S Bendel5, T Birkelund6, R Laru-Sompa7, M Skrifvars8

1University of Helsinki and Helsinki University Hospital, Perioperative, Intensive Care and Pain Medicine, Helsinki, Finland; 2North Karelia Central Hospital, Joensuu, Finland; 3Päijät-Häme Central Hospital, Lahti, Finland; 4University of Helsinki and Helsinki University Hospital, Department of Neurology, Helsinki, Finland; 5Kuopio University Hospital, Kuopio, Finland; 6Aarhus University Hospital, Aarhus, Denmark; 7Central Finland Central Hospital, Jyväskylä, Finland; 8University of Helsinki and Helsinki University Hospital, Department of Emergency Medicine and Services, Helsinki, Finland

Introduction: Cerebral hypoperfusion may aggravate the developing neurological damage after cardiac arrest. Near-infrared spectroscopy (NIRS) provides information on cerebral oxygenation but its clinical relevance during post-resuscitation care is undefined. We wanted to assess the possible association between cerebral oxygenation and clinical outcome after out-of-hospital cardiac arrest (OHCA).

Methods: We performed a post hoc analysis of a randomised clinical trial (COMACARE) where both moderate hyperoxia and high-normal arterial carbon dioxide tension (PaCO2) increased regional cerebral oxygen saturation (rSO2) as compared with normoxia and low-normal PaCO2, respectively. RSO2 was measured from 118 OHCA patients with NIRS during the first 36 h of intensive care and neurological outcome was assessed using the Cerebral Performance Category (CPC) scale at 6 months after cardiac arrest. We calculated the median rSO2 for patients with good (CPC 1-2) and poor (CPC 3-5) outcome and compared the results using the Mann-Whitney U test. We compared the rSO2 over time with outcome using a generalised mixed model. Finally, we added median rSO2 to a binary logistic regression model to control for the effects of possible confounding factors.

Results: The median (interquartile range [IQR]) rSO2 during the first 36 h of intensive care was 70.0% (63.5-77.0%) in patients with good outcome compared to 71.8% (63.3-74.0%) in patients with poor outcome, p = 0.943. We did not find significant association between rSO2 over time and neurological outcome (Figure 1). In the binary logistic regression model rSO2 was not a statistically significant predictor of good outcome (OR 0.99, 95% CI 0.94-1.04, p = 0.635).

Conclusions: We did not find any association between cerebral oxygenation during the first 36 h of post-resuscitation intensive care and neurological outcome at 6 months after cardiac arrest.

Fig. 1 (abstract P204).

Median (inter-quartile range) regional cerebral oxygen saturation (rSO2) in patients with good (Cerebral Performance Category [CPC] 1-2) and poor (CPC 3-5) neurological outcome during the first 36 h after intensive care unit (ICU) admission

P205 Cerebral oxygenation after out-of-hospital cardiac arrest: association with neuron-specific enolase at 48 h

P Jakkula1, M Reinikainen2, J Hästbacka1, V Pettilä1, M Lähde3, M Tiainen4, E Wilkman1, S Bendel5, T Birkelund6, A Pulkkinen7, M Skrifvars8

1University of Helsinki and Helsinki University Hospital, Perioperative, Intensive Care and Pain Medicine, Helsinki, Finland; 2North Karelia Central Hospital, Joensuu, Finland; 3Päijät-Häme Central Hospital, Lahti, Finland; 4University of Helsinki and Helsinki University Hospital, Department of Neurology, Helsinki, Finland; 5Kuopio University Hospital, Kuopio, Finland; 6Aarhus University Hospital, Aarhus, Denmark; 7Central Finland Central Hospital, Jyväskylä, Finland; 8University of Helsinki and Helsinki University Hospital, Department of Emergency Medicine and Services, Helsinki, Finland

Introduction: Near-infrared spectroscopy (NIRS) provides a non-invasive means to assess cerebral oxygenation during post-resuscitation care but its clinical value is unclear. We determined the possible association between cerebral oxygenation and the magnitude of brain injury assessed with neuron-specific enolase (NSE) serum concentration at 48 h after out-of-hospital cardiac arrest (OHCA).

Methods: We performed a post hoc analysis of a randomised clinical trial (COMACARE) comparing two different levels of carbon dioxide, oxygen and arterial pressure after OHCA and successful resuscitation. We measured rSO2 continuously with NIRS from 118 patients during the first 36 h of intensive care. We determined the NSE concentrations at 48 h after cardiac arrest from serum samples using an electrochemiluminescent immunoassay kit. The samples were tested for haemolysis and all samples with a haemolysis index > 500 mg of free haemoglobin per litre (n = 2) were excluded from the analyses. We calculated the median rSO2 for all patients and used a scatterplot and Spearman’s rank-order correlation to assess the possible relationship between median rSO2 and NSE at 48 h. In addition, we compared the NSE concentrations at 48 h after cardiac arrest in patients with good (Cerebral Performance Category Scale [CPC] 1-2) and poor (CPC 3-5) neurological outcome at 6 months using the Mann-Whitney U test.

Results: We did not find significant correlation between median rSO2 and serum NSE concentration at 48 h after cardiac arrest, rs = -0.08, p = 0.392 (Figure 1). The median (IQR) NSE concentration at 48 h was 17.5 (13.4-25.0) μg/l and 35.2 (22.6-95.8) μg/l in patients with good and poor outcome, respectively, p < 0.001.

Conclusions: We did not find any association between cerebral oxygenation during the first 36 h of post-resuscitation intensive care and NSE serum concentrations at 48 h after cardiac arrest.

Fig. 1 (abstract P205).

Scatter plots of serum neuron-specific enolase (NSE) concentration at 48 h after cardiac arrest vs. median regional cerebral oxygen saturation (rSO2) during the first 36 h in intensive care unit in patients with good (Cerebral Performance Category [CPC] 1-2) and poor (CPC 3-5) neurological outcome

P206 The association between lactate, cerebral oxygenation and brain damage in post-cardiac arrest patients

P Nuyens1, S Janssens1, J Dens2, C De Deyne2, B Ferdinande2, M Dupont2, PJ Palmers2, K Ameloot2

1University Hospitals Leuven, Cardiology, Leuven, Belgium; 2Ziekenhuis Oost-Limburg, Cardiology, Genk, Belgium

Introduction: patients admitted to the intensive care unit (ICU) after being successfully resuscitated from a cardiac arrest (CA) have a large cerebral penumbra at risk for secondary ischemic damage in case of suboptimal brain oxygenation. Therefore, resuscitation during ICU stay should be guided by parameters that adequately predict cerebral hypoxia. The value of lactate as resuscitation parameter may be questioned in post-CA patients since the brain critically depends on aerobic metabolism. We aimed to investigate the relationship between arterial lactate, cerebral cortex tissue oxygenation (SctO2) by near infrared spectroscopy (Foresight) and unfavorable neurological outcome at 180days (CPC score 3-5)

Methods: Subanalysis from the Neuroprotect post-CA trial. Lactate values and SctO2 were recorded hourly in 102 post-CA patients during 24hours TTM33 and subsequent rewarming.

Results: In total 3290 paired lactate/ SctO2 measurements were analysed. We found no correlation between paired lactate and SctO² measurements (R²= 0.01) (Fig.1). Moreover, temporary trends in lactate did not correlate with corresponding trends in SctO2 during the same one-hour time interval (R²=0.003) (Fig 2). If lactate values above 2.0 mmol/l are considered to be abnormal, lactate could not adequately detect clinical important brain ischemia (SctO2 < 60%): sensitivity 62% and specificity 53% (Table 1, 2). Nevertheless, time weighted lactate at 6h (OR 1.38; p 0.01), 12h (OR 1.38, p 0.01), 24h (OR 1.53; p 0.003) and 36h (OR 1.60; p 0.005) were inversely correlated with unfavorable neurological outcome at 180days (Fig 1, 2).

Conclusions: Although lactate was a marker of prognosis in post-CA patients, it should not be used to guide resuscitation since lactate values were not correlated with SctO2 and changes in lactate do not correspond with changes in SctO2 during the same time interval.

Table 1 (abstract P206). Logistic regression analysis of average lactate as predictor of poor neurological outcome (CPC 3–5) at 6 months
Table 2 (abstract P206). Sensitivity and specificity of lactate values > 2 mmol/L to detect SctO² < 60%
Fig. 1 (abstract P206).

Linear correlation between paired lactate and SctO² measurements

Fig. 2 (abstract P206).

Linear correlation between paired, hourly lactate and SctO² measurements

P207 Simplified monitoring and interpretation of continuous EEG after cardiac arrest - a useful bedside tool for the ICU physician

A Lybeck1, T Cronberg2, O Borgquist1, J During3, G Mattiasson1, D Piros3, S Backman4, H Friberg3, E Westhall4

1Skane Univeristy Hospital, Dept of Anaesthetics and Intensive Care, Lund, Sweden; 2Skane Univeristy Hospital, Dept of Neurology, Lund, Sweden; 3Skane University Hospital, Dept of Anaesthetics and Intensive Care, Malmo, Sweden; 4Skane Univeristy Hospital, Dept of Clinical Neurophysiology, Lund, Sweden

Introduction: The aim of the study was to investigate whether simplified continuous EEG monitoring (cEEG) [1] post-cardiac arrest can be reliably interpreted by ICU physicians after a short structured training, and whether acceptable interrater agreement compared to an EEG-expert can be achieved.

Methods: Five ICU physicians received training in interpretation of simplified cEEG (Fig 1) consisting of lectures, hands-on cEEG-interpretation, and a video tutorial - total training duration 1 day. The ICU physicians then interpreted 71 simplified cEEG recordings. Basic EEG background patterns and presence of epileptiform discharges or seizure activity were assessed on 5-grade rank-ordered scales based on a standardized EEG terminology [2]. An experienced EEG-expert was used as reference.

Results: There was substantial agreement (κ 0.69) for EEG background patterns and moderate agreement (κ 0.43) for epileptiform discharges between ICU physicians and the EEG-expert. Sensitivity for detecting seizure activity by the ICU physicians was limited (50%), but with high specificity (87%). Among ICU physicians interrater agreement was substantial (κ 0.63) for EEG background pattern and moderate (κ 0.54) for epileptiform discharges.

Conclusions: After a one-day educational effort clinically relevant agreement was achieved for basic EEG background patterns after cardiac arrest. Assessment of epileptiform patterns was less reliable, but bedside screening by the ICU physician may still be clinically useful for early detection of seizures. Interpretation of simplified cEEG requires awareness of its limitations and support from an EEG-expert when clinically indicated.


1 Friberg H, et al. Crit Care 17:233-242, 2013

2 Hirsch LJ, et al. J Clin Neurophysiol 30:1-27, 2013

Fig. 1 (abstract P207).

1 cEEG montage and example of simplified continuous EEG recording: F3, P3, F4, P4, reference Cz, ground Fz. The upper curves display the left and the right time-compressed aEEG, the y-axis displays the semilogarithmic uV scale. The shaded aEEG marks the time span of the aEEG corresponding to the original EEG below. The lower part displays the original EEG recording from the left and the right hemispheres; each division represents 1 sec on the x-axis

P208 Neurocardiac risk stratification 6 hours after resuscitation from cardiac arrest

A Bilkanovic1, Z Haxhija1, L Lucas2, J Dziodzio1, R R. Riker1, T May1, H Friberg3, D B. Seder1

1Maine Medical Center, Department of Critical Care Services and Neuroscience Institute, Portland, United States; 2Maine Medical Center, Center for Outcomes Research and Evaluation, Portland, United States; 3Skåne University Hospital, Department of Anaesthesia and Intensive Care, Lund, Sweden

Introduction: A tool to distinguish between the competing risks of neurologic- and circulatory-etiology death very early after return of spontaneuos circulation (ROSC) would facilitate triage decisions to extracorporeal membrane oxygenation, cardiac revascularization, and experimental neuroprotective therapies. We combined the electroencephalographic suppression ratio (SR) and CREST score (Coronary artery disease, initial non-shockable Rhythm, low Ejection fraction, admission Shock and Time to ROSC>25 minutes), a validated model to predict risk of circulatory-etiology death (CED), to create new neurocardiac risk classes.

Methods: Prospectively collected registry data included demographics, clinical data, the SR recorded at 6 hours after ROSC (SR6), CREST score, and outcomes. Utility of SR6 to predict neurological-etiology death (NED) was evaluated with a ROC curve. Clinically appropriate cutoffs to describe risk levels for each tool were selected, and patients stratified into 6 groups to demonstrate competing risks of CED and NED.

Results: Of 543 patients, 227 (42%) survived to hospital discharge, 239 (44%) had NED, 77 (14%) had CED. SR6 values predicted NED with AUC 0.89, and clinical thresholds were set at SR6 <37% (low risk of NED; 3-11%), SR6 37-82% (intermediate risk; 23-60%) and SR >83% (high risk of NED; 71-89%). CED occurred in 11-13% in low CREST (0-2) groups and 18-28% in the high CREST (3-5) groups. No patient with both high neurological and circulatory risk survived, and the lowest risk group had 87% survival. Relative risk of NED:CED in patients with lowest SR6 was low (1:4.3); in patients with highest SR6, NED:CED was high (5.2:1).

Conclusions: A neurocardiac risk model using early SR and CREST score allows for risk stratification at 6 hours after ROSC, and could be used to optimize the effectiveness of post-resuscitation care pathways and interventions.

P209 Interobserver variability in the interpretation of early head computed tomography following out of hospital cardiac arrest

A Caraganis1, RR Kempainen2, M Mulder3, R Brown4, M Oswood2, ME Prekker2

1University of Minnesota, Division of Pulmonary, Allergy and Critical Care, Minneapolis, United States; 2Hennepin Healthcare, Minneapolis, United States; 3Abbott Northwestern Hospital, Minneapolis, United States; 4University of Minnesota, Minneapolis, United States

Introduction: Hypoxic-ischemic injury on head computed tomography (CT), which manifests with varying degrees of cerebral edema and loss of gray-white matter differentiation, is a poor prognostic sign after resuscitated out-of-hospital cardiac arrest that may influence early clinical decision-making. Agreement among physicians on the presence of hypoxic-ischemic injury on early head CT is unknown.

Methods: We recruited 10 faculty physician participants (2 emergency medicine, 3 critical care, 3 neurocritical care, and 2 general radiology; average 8.2 years of practice) across 3 academic medical centers each with >100 admissions for resuscitated out-of-hospital cardiac arrest each year. Participants, blinded to clinical context, reviewed 20 unique head CTs obtained within 2 hours of cardiac arrest that were randomly selected from a local registry. A blinded neuroradiologist also reviewed all scans (gold standard). Participants determined if hypoxic-ischemic injury was present on each CT, and agreement was determined using multi- and dual-rater kappa statistics with 95% confidence intervals.

Results: Overall agreement among physicians regarding the presence of hypoxic-ischemic injury on head CT was fair (kappa 0.34; 95% CI, 0.19-0.49) with agreement consistent across most specialties (Table 1). When compared to the neuroradiologist, individual physician agreement ranged widely, from poor (kappa 0.12) to substantial (kappa 0.68), with 6 of 10 physicians having fair or worse agreement compared to the gold standard interpretation.

Conclusions: The finding of hypoxic-ischemic injury on early head CT after cardiac arrest had high interobserver variability as interpreted by acute care physicians and general radiologists. Pending the development of objective diagnostic criteria, clinicians should bear in mind the subjectivity and subtlety of cerebral edema or loss of gray-white matter differentiation soon after return of spontaneous circulation in these patients.

Table 1 (abstract P209). Interobserver agreement on the presence of hypoxic-ischemic injury on early head computed tomography after cardiac arrest

P210 Testing before stopping: a 6-year audit of neuro-prognostication tests after cardiac arrest

P Eiben, S Blakey, J Van Griethuysen, D Noakes, F Caetano, E Poimenidi, A Gupta, A Pineau Mitchell, G Patel, R Maharaj, S Vlachos

King´s College Hospital NHS Foundation Trust, Critical Care, London, United Kingdom

Introduction: We assessed institutional adherence of neuro-prognostication test use before Withdrawal of Life Sustaining Treatment (WLST) to international guidance. We also assessed the change in practice over the six-year time frame and compared the time periods before and after publication of the 2015 ERC-ESICM guidelines [1].

Methods: Retrospective analysis of adult comatose Out of Hospital Cardiac Arrest (OHCA) survivors, admitted to a tertiary referral Intensive Care Unit (ICU) between January 2012 and December 2017. Blinded data collection from electronic and paper records followed the Utstein template. Patients who were not intubated, had a non-cardiac arrest or stayed in ICU ²72h were excluded since their neuro-prognostication differed. We used Fisher´s exact tests for categorical variables, Wilcoxon tests for continuous variables and non-parametric tests for trend.

Results: We included 287 patients (Figure 1). Baseline characteristics and differences between the WLST and no-WLST groups are shown in Table 1. Utilization of neuro-prognostication tests is shown in Table 2. While CT and EEG were commonly employed, SSEP and MRI were used less frequently. Basic multimodal neuro-prognostication (arbitrarily defined as at least one CT or MRI, plus EEG, plus SSEP) was performed only in 34.1% of all patients undergoing WLST but the rate increased significantly over six years (p<0.001) and was higher in the time period after 2015, compared to the one prior to 2015 (Figure 2). This association remained significant after adjustment for confounders such as age, arrest rhythm, downtime, targeted temperature management, APACHE II score and organ failure in a logistic regression model (p=0.004).

Conclusions: In an institution with access to a wide range of imaging and neurophysiology tests, MRI and SSEP remained under-utilized but the rate of basic multimodal neuro-prognostication increased significantly over the study period, especially in the period after 2015.


1. Nolan JP, et al. Resuscitation. 2015;95:202-22.

Table 1 (abstract P210). Baseline characteristics of patients by category of withdrawal of life support (WLST)
Table 2 (abstract P210). Neuro-prognostication tests performed in patients after OHCA
Fig. 1 (abstract P210).

Patient flow during the study period

Fig. 2 (abstract P210).

Proportion of patients receiving basic multimodal neuro-prognostication over the study period

P211 Melatonin protects autophagy-like cell death cerebellar Purkinje cells following asphyxial cardiac arrest through attenuation of oxidative stress via MT2 receptor

JK Seo1, HJ Lim1, JB Moon1, TG Ohk1, MC Shin1, KE Kim1, MH Won2, JH Cho3

1Kangwon National University, Emergency Medicine, Chuncheonsi, South Korea; 2Kangwon National University, Neurobiology, Chuncheonsi, South Korea; 3Kangwon National University, Department of Emergency Medicine, Chuncheonsi, South Korea

Introduction: Although multiple reports using animal models have confirmed that melatonin appears to promote neuroprotective effects following ischemia/reperfusion-induced brain injury, the relationship between its protective effects and the activation of autophagy in cerebellar Purkinje cells following the asphyxial cardiac arrest and cardiopulmonary resuscitation (CA/CPR) remains unclear.

Methods: Rats used in this study were randomly assigned to 6 groups as follows; vehicle-treated sham-operated group, vehicle-treated asphyxial CA/CPR-operated group, melatonin-treated sham-operated group, melatonin-treated asphyxial CA/CPR-operated group, melatonin plus (+) 4P-PDOT (the MT2 melatonin receptor antagonist)-treated sham-operated group and melatonin+4P-PDOT-treated asphyxial CA/CPR-operated group.

Results: Our results demonstrate that melatonin (20 mg/kg, IP, 1 time before CA and 4 times after CA) significantly improved the survival rates and neurological deficits compared with the vehicle-treated asphyxial CA/CPR rats (survival rates ≥ 40% vs 10%). We also demonstrate that melatonin exhibited the protective effect against asphyxial CA/CPR-induced Purkinje cell death. The protective effect of melatonin in the Purkinje cell death following asphyxial CA/CPR paralleled a dramatic reduction in superoxide anion radical (O2·-), intense enhancements of CuZn superoxide dismutase (SOD1) and MnSOD (SOD2) expressions, as well as a remarkable attenuation of autophagic activation (LC3 and Beclin-1), which is MT2 melatonin receptor-associated. Furthermore, the protective effect of melatonin was notably reversed by treatment with 4P-PDOT.

Conclusions: This study shows that melatonin conferred neuroprotection against asphyxial CA/CPR-induced cerebellar Purkinje cell death by inhibiting autophagic activation by reducing expressions of ROS, while increasing of antioxidative enzymes, and suggests that MT2 is involved in the neuroprotective effect of melatonin in cerebellar Purkinje cell death induced by asphyxial CA/CPR.

P212 Pretreated fucoidan confers neuroprotection against transient global cerebral ischemic injury in the gerbil hippocampal CA1 area via reducing of glial cell activation and oxidative stress

HJ Lim1, JK Seo1, JB Moon1, TG Ohk1, MC Shin1, KE Kim1, MH Won2, JH Cho3

1Kangwon National University, Emergency Medicine, Chuncheonsi, South Korea; 2Kangwon National University, Neurobiology, Chuncheonsi, South Korea; 3Kangwon National University, Department of Emergency Medicine, Chuncheonsi, South Korea

Introduction: Fucoidan is a sulfated polysaccharide derived from brown algae and possesses various beneficial activities, such as anti-inflammatory and antioxidant properties. Previous studies have shown that fucoidan displays protective effect against ischemia-reperfusion injury in some organs. However, few studies have been reported regarding the protective effect of fucoidan against cerebral ischemic injury and its related mechanisms.

Methods: Therefore, in this study, we examined the neuroprotective effect of fucoidan against cerebral ischemic injury, as well as underlying mechanisms using a gerbil model of transient global cerebral ischemia (tGCI) which shows loss of pyramidal neurons in the hippocampal cornu ammonis 1 (CA1) area. Fucoidan (25 and 50 mg/kg) was intraperitoneally administered once daily for 3 days before tGCI.

Results: Pretreatment with 50 mg/kg of fucoidan, not 25 mg/kg fucoidan, attenuated tGCI-induced hyperactivity and protected CA1 pyramidal neurons from ischemic injury following tGCI. In addition, pretreatment with 50 mg/kg of fucoidan inhibited activations of resident astrocytes and microglia in the ischemic CA1 area. Furthermore, pretreatment with 50 mg/kg of fucoidan significantly reduced the increased 4-hydroxy-2-noneal and superoxide anion radical production in the ischemic CA1 area after tGCI and significantly increased expressions of superoxide dismutase 1 (SOD1) and SOD2 in the CA1 pyramidal neurons compared with the vehicle-treated-group. We found that treatment with diethyldithiocarbamate (an inhibitor of SODs) to the fucoidan-treated-group notably abolished the fucoidan-mediated neuroprotection in the ischemic CA1 area following tGCI.

Conclusions: These results indicate that fucoidan can effectively protect neurons from tGCI-induced ischemic injury through attenuation of activated resident glial cells and reduction of oxidative stress following increasing SODs. Thus, we strongly suggest that fucoidan can be used as a useful preventive agent in cerebral ischemia.

P213 The effects of cold fluids for induction of therapeutic hypothermia on reaching target temperature and complications– a sub-study of the TTH48 study

A Holm1, M Skrifvars2, FS Taccone3, E Søreide4, A Grejs5, C Duez5, A Jeppesen5, H Kirkegaard5

1University of Helsinki, Faculty of Medicine, Helsinki, Finland; 2Dept. of Emergency Care and Services, Helsinki University Hospital, Finland, Helsinki, Finland; 3Dept. of Intensive Care, Erasme Hospital, Belgium, Brussels, Belgium; 4Dept. of Anesthesiology and Intensive Care, Stavanger University Hospital, Stavanger, Norway; 5Dept. of Anesthesiology and Intensive Care Medicine, Aarhus University Hospital, Aarhus, Denmark

Introduction: Induction of hypothermia with cold fluids does not improve outcome in out- of- hospital cardiac arrest (OHCA) [1]. We hypothesized that this may be due to ineffective cooling and side effects.

Methods: A post hoc analysis of the randomised TTH48 trial (NCT01689077) comparing cooling for 48 or 24 hours after OHCA. Data collection included cardiac arrest factors, adverse effects, cooling methods and continuous core temperature measurements. The primary outcome endpoint was time to target temperature (TTT, <34°C) and prevalence of abnormal electrolyte levels and oxygen saturation within the first 24 hours. TTT was considered significantly reduced if it was at least 1 hour shorter with i.v. fluid cooling than with no or only surface pre-ICU cooling. We compared electrolyte and circulatory side effects between groups.

Results: Intervention group given pre-hospital cold fluids contained 110 patients and the control group 242 patients. The pre-ICU cold fluid cooling increased TTT (347min vs. 268min, p=0.010). Temperatures over time varied based on whether pre-ICU i.v. fluids were given (Figure 1). With linear multiple regression in a model including age, basic CPR, shockable or non-shockable rhythm, whether the CA was witnessed or not, time to ROSC, the use of cold fluids was positively associated with longer time to target temperature (coefficient 1.22, 95% CI 4.9 and 71.7, p=0.025). There was no difference in paO2 on ICU admission (mean 15.5 kPa compared to 16 kPa, p=0.787). Cold fluid use did not increase electrolyte abnormalities (Na+<130mmol 1.8% vs. 2.9% p=0.540, K+<3.0mmol 1.8% vs. 4.5% p=0.202).

Conclusions: Contrary to our hypothesis, the initiation of targeted temperature management with cold fluids before ICU arrival may increase TTT but does not cause other side effects.


1. Francis K et al. JAMA 311:45-52, 2014

Fig. 1 (abstract P213).

Temperature of patients given and not given pre-ICU fluids

P214 Targeted temperature management during extracorporeal life support after CPR

A Mecklenburg1, J Stamm2, F Angriman3, G Söffker2, H Reichenspurner3, L Del Sorbo1, E Fan1, S Kluge2, S Braune4

1Interdepartmental Division of Critical Care, University of Toronto, University of Toronto, Toronto, Canada; 2Department of Critical Care, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; 3Department of Cardiac Surgery, University Heart Center Hamburg, Hamburg, Germany; 4Department of Medical Intensive Care and Emergency Medicine, St. Franziskus-Hospital, Münster, Germany

Introduction: Extracorporeal life support (ECLS) is increasingly used after cardiopulmonary resuscitation. However, bleeding complications are common and challenging. Targeted temperature management (TTM) improves neurologic outcome after CPR but may compromise coagulation. The purpose of this study was to examine whether TTM in patients on ECLS post CPR increases risk for major bleeding.

Methods: Retrospective single-center study of patients on ECLS (vaECMO) post CPR±TTM from Jan 2009 to Dec 2015, using inverse probability weighting of a marginal structural modified Poisson regression model to estimate TTM effect. Bleeding was recorded within 36hrs post CPR (early observation) and between 36 to 72hrs post CPR (late observation) using the BARC score [1]. Secondary outcomes: ICU survival, 28d-mortality, organ dysfunction.

Results: 36 out of 78 patients received TTM in addition to ECLS after CPR. They were younger (48 [38.5-57] vs. 55.5 [48.3-65.5] yrs), had higher illness severity (SAPS2 66.5 [57.8-74.3] vs. 56 [43.3-66.3]), and, higher incidence of lung injury (83.3% vs. 23.8%), brain injury (58.3% vs. 28.6%) and multi-organ failure (88.9% vs. 64,3%). ICU survival and 28-d mortality did not differ. Overall bleeding frequency (BARC 1-5) was similar in both groups (TTM 86.1% vs. noTTM 92.9%). There was no difference in early bleeding incidences (Fig 1). During late observation, TTM patients had fewer minor bleeding (55.2% vs. 100%) and more intracranial bleeding (23.1% vs. 0%; Fig 2). Adjusted calculated risk ratio for major bleeding (including intracranial) for TTM was 0.24 (95%CI 0.10-0.54) at baseline and 1.45 (95%CI 1.18-1.77) over time.

Conclusions: Bleeding complications were common. Although the risk ratio for major bleeding increased over time in TTM patients, residual and unmeasured confounding in addition to selection and detection bias may limit the clinical relevance of this finding.


1) Mehran Ret al., Circulation, 123:2736–2747, 2011

Fig. 1 (abstract P214).

Bleeding frequencies in early observation period, ECMO vs. ECMO+TTM

Fig. 2 (abstract P214).

Bleeding frequencies in late observation period, ECMO vs. ECMO+TTM, p=0.05

P215 Features of cerebral thermal balance in the chronically critically ill patients

M Petrova1, O Shevelev2, S Shavkat1, R Mohan2

1Federal scientific and clinical center of resuscitation and rehabilitation, Moscow, Russia; 2Peoples´ Friendship University of Russia, Moscow, Russia

Introduction: In acute period of different cerebral events, often, there is fever & focal cerebral hyperthermia which can worsen the condition. Supposedly, the change in brain thermal balance (TB) can also develop in critical conditions (coma, vegetative & minimally conscious state) but there is inadequate information about variations in cerebral temperature (t°) thus, research was done to study the cerebral TB in the chronically critically ill (CCI) in comparison with healthy people (HP) & patients in acute period of ischemic stroke (IS).

Methods: Patients with neurological deficit > 10 by NHISS were included. The t° of the brain was recorded non-invasively using radiothermometer RTM-01-RES (Russia). We measured t° in 18 symmetric regions of left & right hemispheres, calculated the average t° of brain, Δ°C between warmest & coolest brain regions, made the correlative analysis of t° variation in symmetric regions (Pearson correlation) in HP (n=120), IS (n=65) & CCI (n=65).

Results: In HP, disposition of cold & warm regions doesn’t have typical localization, it was different in every individual. In IS, the region of focal hyperthermia & the location of penumbra coincided in 85% of cases (Table 1).

Conclusions: Observed moderate brain t° heterogenecity in HP, marked increase brain t° heterogenecity in IS & sharp decline of t° heterogenecity in CCI. Supposedly, correcting the impairment of cerebral TB (increase or decrease t°) through physical (selective cerebral hypothermia, magnetic stimulation etc.) or pharmacological (sedation) can contribute to positive therapeutic results in IS & CCI. Nonivasive radiothermometry of the brain can be an objective method of patients’ condition evaluation & their rehabilitation potential.

Table 1 (abstract P215). Results

P216 Angio-oedema or pseudobulbar palsy? The masquerades of basilar artery stroke

J Clarke, V Della Torre

West Suffolk Hospital NHS Foundation Trust, Department of Anaesthesia and Intensive Care, Bury St Edmunds, United Kingdom

Introduction: Basilar artery stroke has a multitude of different presentations and may not be captured on plain Computed Tomography (CT). It can progress to severe disability, locked in syndrome and death [1]. With the advent of thrombolytic and endovascular therapies, prompt diagnosis can change the outcome. We present a case of basilar artery stroke, which was heralded by tongue spasticity and dysarthria, indicative of pseudobulbar palsy.

Methods: Case reviewed with consent. A literature search was conducted using PubMed and Medline.

Results: A 53-year-old presented with pulmonary oedema and hypertension. He was transferred to our intensive care unit for treatment of a suspected anaphylaxis. His marked lingual swelling was associated with dysarthria. Glyceryl-trinitrate and labetalol infusions were started for hypertension. He developed left sided weakness and deteriorated over several days to the point that he could only move his right foot (Table 1). Magnetic Resonance Imaging (MRI) showed midbrain ischaemia and angiogram showed no flow in the basilar artery (Fig 1,2).

Conclusions: Common presenting features of basilar artery occlusion include dysarthria, vertigo, vomiting, headache and motor defects; these may evolve gradually or be intermittent [1,2]. Presentation with pseudobulbar palsy is described in early literature [2]. Delayed recognition of the stroke led to aggressive treatment of hypertension, potentially compromising perfusion to the penumbral area [2,3]. This case highlights the need for a wide index of suspicion with posterior strokes.

Consent: Informed consent to publish has been obtained from the patient


1 Ausman JI et al. Surg Neurol Int. 9:106, 2018

2 Silversides JL. Proc R Soc Med. 47:290-3, 1954

3 Jauch EC et al. Stroke. 44:889-891, 2013

Table 1 (abstract P216). Trends in blood pressure, neurological signs and imaging results
Fig. 1 (abstract P216).

MRI brain showing bilateral midbrain infarcts

Fig. 2 (abstract P216).

MRI Angiogram showing absence of flow in the basilar artery

P217 Burns intensive care (ICU) survivors show altered functional connectivity between sub-regions of the default mode network compared to healthy volunteers

EJ Watson1, S O´Connor1, ME Finnegan2, M Grech-Sollars3, CE Richards4, K Nenadlova5, L Honeyfield2, R Quest2, TL Edginton6, AD Waldman7, MP Vizcaychipi1

1Chelsea and Westminster Hospital NHS Foundation Trust, Magill Department of Anaesthesia, Intensive Care Medicine and Pain Management, London, United Kingdom; 2Imperial College Healthcare NHS Trust, Department of Imaging, London, United Kingdom; 3Imperial College London, Department of Surgery & Cancer, London, United Kingdom; 4Morriston Hospital, Department of Medicine, Swansea, United Kingdom; 5University of Westminster, Department of Psychology, London, United Kingdom; 6City University, Department of Psychology, London, United Kingdom; 7University of Edinburgh, Centre for Clinical Brain Sciences, Edinburgh, United Kingdom

Introduction: Up to two thirds of patients discharged from ICU experience long-term cognitive impairment (LTCI) and this can have a significant impact on quality of life. Little is currently known about its neurobiological basis. One area of interest is the Default Mode Network (DMN), a well-defined resting state functional network that is disrupted in pathologies including Alzheimer’s disease and Traumatic Brain Injury.

Methods: 15 patients, who had previously sustained a significant burns injury and had been admitted to burns ICU for invasive ventilation, and 15 volunteers, underwent resting state functional MRI (fMRI) scans. Data analysis was performed using the CONN toolbox in SPM8. Regions of Interest (ROIs) were the four major subdivisions of the DMN (anterior, posterior, l-TPJ and r-TPJ). The average BOLD time series for each ROI was correlated with all other ROIs. The correlation coefficients were compared between patients and volunteers using ANCOVA and correlated with cognitive performance (linear regression). Age and gender adjustment and multiple comparisons correction were undertaken.

Results: Reciprocal functional connectivity between r-TPJ and Posterior-DMN (F(1)(26) = 7.62, p=0.03) (Figure 1) and between r-TPJ and l-TPJ (F(1)(26) = 9.07, p=0.03) was lower in patients compared to volunteers. Functional connectivity between r-TPJ and Posterior-DMN positively correlated with cognitive performance across the cohort (r2=0.36, p=0.001), even after adjusting for predicted premorbid IQ levels (r2=0.37, p=0.002).

Conclusions: To our knowledge this is the first study demonstrating resting state network disruption in ICU patients. Since functional connectivity correlates with cognitive performance, further focussed research in this area may allow for targeted prevention or treatment of ICU related LTCI.

Fig. 1 (abstract P217).

Functional connectivity between RTPJ and posterior-DMN across volunteers and patients

P218 Ischemic stroke predictors in an Italian ICU

A Marudi1, G Bettelli1, G Branchetti2, E Marinangeli2, R Sandulli2, E Bertellini1

1Azienda Ospedaliero Universitaria di Modena, Struttura Complessa di Anestesia e Rianimazione, Modena, Italy; 2Azienda Ospedaliero Universitaria di Modena, Università degli Studi di Modena e Reggio Emilia, Modena, Italy

Introduction: Ischemic stroke is a rapidly evolving condition that may lead to death. The development of acute stroke network has facilitated recognition, diagnosis and rapid treatment improving outcome of ischemic stroke. The aim of this study is to highlight any predicting factors that facilitate to outline the prognosis of acute ischemic stroke.

Methods: We made a retrospective analysis of the patients admitted to Ospedale Civile of Modena’s NICU (Neurosciences Intensive Care Unit) with diagnosis of ischemic stroke. We selected a period of 22 months, from January 2017 to October 2018. At the admission we collected information about: Glasgow Coma Scale (GCS), worst GCS in the first 24 hours, SOFA score, SAPS II. Furthermore we analysed the following data: complications, surgical and non surgical procedures, GCS < or = 8 or mortality at discharge from NICU. We considered the last two parameters indicating the worst prognosis.

Results: 49 patients were studied, with average age of 70.8 sd ± 11.9. 34 (69.4%) patients were male. Although GCS on admission of 16 patients was missed, median of GCS was 8 iq 8. Median of the worst GCS in the first 24 hours was 9 iq 7. Median SOFA score and median SAPS II were respectively 5 iq 2.5 and 37 iq 22. During the recovery 21 patients (42.9%) developed neurological complications, of which 10 (47.6%) had malignant cerebral edema. Decompressive craniectomy was applied in 7 patients (14.3%). 33 patients (67.4%) underwent neuroradiology procedures. At discharge from NICU 19 patients had a GCS < or = 8 and were died (38.8%). The worst prognosis is related to GCS < or = 8 on admission (P = 0.003) and to malignant cerebral edema (P = 0.004).

Conclusions: Our study has shown some predictive factors closely related to mortality and morbidity in patients with acute ischemic stroke. GCS at admittance < or = 8 and onset of malignant cerebral edema lead to a worst prognosis at discharge from NICU.

P219 Coherence analysis of cerebral oxygenation using multichannel functional near-infrared spectroscopy evaluates cerebral perfusion in hemodynamic stroke

TJ Kim1, JM Kim2, SH Park1, HB Jeong1, HM Bae2, SB Ko1

1Seoul National University Hospital, Neurology, Seoul, South Korea; 2Korea Advanced Institute of Science and Technology (KAIST), Electrical Engineering, Daejeon, South Korea

Introduction: The functional near infrared spectroscopy (fNIRS) could evaluate brain function based on measuring changes in oxygenated and deoxygenated hemoglobin concentrations. The wavelet phase coherence (WPCO) can reveal the relationship of two signals between oscillations. We aimed to evaluate prefrontal functional connectivity using WPCO analyses of cerebral oxyhemoglobin (OxyHb) in patients with hemodynamic stroke.

Methods: A 35 consecutive patients with anterior circulation ischemic strokes and control patients (n=34) were included for analysis. The cerebral OxyHb data were collected using multichannel fNIRS (NIRSIT, OBELAB Inc., Republic of Korea). The coherences between eight segments of prefrontal OxyHb oscillations in five frequency intervals (I, 0.6–2Hz; II, 0.145–0.6Hz; III, 0.052–0.145Hz, IV, 0.021–0.052Hz, and V, 0.005–0.0095 Hz) were analyzed using wavelet coherence analysis. The patients were categorized into three groups with control groups, patients with perfusion defect and patients without perfusion defect. We compared the result of coherence analysis of OxyHb between three groups.

Results: Among the included 35 patients (age, 63.8; and male, 62.9%), 21 patients (60.0%) had perfusion defect in anterior circulation stroke. In patients with perfusion defect, phase coherence was significantly higher compared to patients without perfusion defect and control group, especially in interval III under the myogenic mechanism of cerebral autoregulation (0.70±0.17 vs. 0.57±0.20 vs. 0.58±0.15, P = 0.021) (Table 1). In addition, severe stroke patients were more likely to have higher phase coherence in interval III (P =0.078).

Conclusions: Our results demonstrated that the higher phase coherence of OxyHb in myogenic signal, which was originated locally from smooth muscle cells in brain was related to impaired cerebral perfusion. This suggests that monitoring cerebral oxygenation using fNIRS could be a useful noninvasive measuring tool for evaluating impaired cerebral autoregulation in stroke patients.

Table 1 (abstract P219). WPCO value according to perfusion

P220 Is esmolol associated with worse outcome at the acute phase of ischemic stroke that receives thrombolysis?

P Papamichalis1, E Neou1, O Triantafillou1, S Karagiannis1, E Dardiotis2, D Papadopoulos1, T Zafeiridis1, D Babalis1, A Pappa1, A Skoura1, N Ntafoulis1, A Komnos1

1General Hospital of Larissa, Department of Intensive Care, Larissa, Greece; 2University Hospital of Larissa, Department of Neurology, Larissa, Greece

Introduction: The aim of our study was to examine the hypothesis that esmolol has neuroprotective action at brain ischemia and test this assumption at strokes that receive Intravenous Thrombolysis (I.T).

Methods: Retrospective study from our department¢s thrombolysis database.We compared 36 patients [mean National Institutes of Health Stroke Scale (NIHSS) 11.5/range 3-23] who received I.T with alteplase for acute ischemic stroke and esmolol for rhythm / rate control within the first 24 hours from the stroke with 113 patients [mean NIHSS 11/range 2-28] who received alteplase but not esmolol. Severity scores, predisposing factors, complications, mortality / functional outcome at 3 months of the 2 groups were statistically analysed.

Results: Length of stay, 7 days NIHSS, complication rate and functional outcome at 3 months were worse at esmolol group. The esmolol group had higher Symplified Acute Physiology Score (SAPS) II and at higher rate atrial fibrillation (Table 1).

Conclusions: Our study failed to detect any neuroprotective effect of esmolol on ischemic brain. On the contrary esmolol was correlated with worse outcome and more complications. Further studies are needed to determine if this is a direct negative effect of esmolol or if it is just the result of greater severity index and more frequent presence of atrial fibrillation resulting at more severe strokes at patients receiving esmolol.

Table 1 (abstract P220). Comparison of the 2 groups: esmolol (n=36) versus control group (n=113)

P221 Neutrophil-to-lymphocyte ratio is associated with early neurological deterioration in acute ischemic stroke

TJ Kim, KW Nam, SHP ark, HB Jeong, SB Ko

Seoul National University Hospital, Neurology, Seoul, South Korea

Introduction: Ischemic stroke patients experienced frequent early neurological deterioration (END) events. Since ischemic stroke has also been shown as inflammatory disease, the neutrophil-to-lymphocyte ratio (NLR) may associated with END events. However, the direct study regarding this association has not been addressed. We evaluated the association between NLR and END in ischemic stroke patients.

Methods: We included ischemic stroke patients between 2010 and 2015. END was defined as an increase ≥ 2 on the total National Institutes of Health Stroke Scale (NIHSS) score or ≥ 1 on the motor NIHSS score within 72 hours of admission. The NLR was calculated as the ratio of the absolute neutrophil count to the absolute lymphocyte count on admission.

Results: A total of 1,152 patients were included. Among them, END occurred in 154 (13%) patients and the median NLR value was 2.47 [1.59-3.97]. In multivariate analysis, NLR remained an independent predictor of END [adjusted odds ratio (aOR) = 1.071, 95% confidence interval (CI) = 1.025 to 1.119, P = 0.002] (Table 1). These results were more prominent in large artery atherosclerosis group (aOR = 1.107, 95% CI = 1.031 to 1.188, P = 0.005), while stroke patients with small vessel occlusion (aOR = 1.114, 95% CI = 0.895 to 1.387, P = 0.335) or cardioembolic (aOR = 1.028, 95% CI = 0.912 to 1.159, P = 0.649) mechanisms did not show statistical significance.

Conclusions: A high NLR level was associated with END in ischemic stroke patients, especially in large artery atherosclerosis patients. The NLR may hep to identify high-risk patients in time and provide clues for further studies about inflammatory pathophysiology of ischemic stroke.

Table 1 (abstract P221). Multivariable analysis of possible predictors of early neurological deterioration

P222 The role of the transcranial Doppler in the neuromonitorization of the critically ill patient

J Higuera Lucas, S Gallego, G Narvaéz, A Caballero, D Cabestrero, C Soriano, R De Pablo

Hospital Universitario Ramón y Cajal, Medicina Intensiva, Madrid, Spain

Introduction: Hemodynamic, renal, hepatic or cardiac monitoring is widely described in the critically ill patient, however, the knowledge about neuromonitoring is scarcer. Especially in patients who are sedated and on mechanical ventilation. Our aim is to describe the role of the transcranial Doppler in this process.

Methods: A prospective, observational study was performed with all the patients admitted to an Intensive Care Service of a Tertiary, University Hospital. All patients analyzed required mechanical ventilation to support their pathology. All consecutive patients admitted to the unit under mechanical ventilation, noradrenaline dose <0.06 mcg/kg/min, were included, with the expectation of remaining under mechanical ventilation for at least 24 more hours after data collection. The severity indexe and, mortality rates were collected. As well, pulsatibility indexes of the middle cerebral artery were obtained through the left or right temporal window.

Results: A total of 30 patients were included. Mean age 63±14.3 years, SOFA 12.3±12.8 SAPS II 56±17.6; APACHE II 25.5±9.7; Total days on M.V. 7.5±13 days; Length of stay 29.5±32.5 days; Mean vasoactive drugs 0.003 mcg/kg/min. The average values of pulsatibility indexes in the group of survivors was 1.07, being 1.5 in the group of non-survivors (P=0.02). Values above 1.2 were associated in a statistically significant way with mortality (P = 0.013). The mortality rate in the group of patients which values were higher than 1.2 was 66.7%. On the other hand, patients with fewer values than 1.2 had a mortality rate of 20%. Patients with pulsatibility indexes greater than 1.2 presented higher severity indices. (P=0.02).

Conclusions: Transcranial Doppler is a useful tool in the neuromonitoring of critically ill patients. In our study, the pulsatibility indexes of the middle cerebral artery were associated in a statistically significant way with mortality. Patients with higher severity indexes presented altered pulsatility values.


P224 Predictors of mortality after subarachnoid hemorrhage: a restrospective multicenter cohort study

P Kurtz1, FS Taccone2, B Gonçalves3, M Soares3, F Bozza3, M Medeiros Machado4, M Maia5, M Ferez6, C Nassif7, C Shinotsuka8, JL Salluh3

1Paulo Niemeyer State Brain Institute, Neuro ICU, Rio de Janeiro, Brazil; 2Free University of Brussels, Brussels, Belgium; 3D´Or Institute for Research and Education, Rio de Janeiro, Brazil; 4Hospital Agenor Paiva, Salvador, Brazil; 5Hospital Santa Luzia, Viana do Castelo, Portugal; 6Hospital Sao Francisco, Brasília, Brazil; 7Hospital Nove de Julho, São Paulo, Brazil; 8Paulo Niemeyer State Brain Institute, Rio de Janeiro, Brazil

Introduction: Aneurysmal subarachnoid haemorrhage (SAH) is an acute, and often catastrophic, cerebrovascular event, with high mortality and morbidity. Data on predictors of mortality in low and middle-income countries is scarce. The purpose of this study was to investigate clinical predictors of hospital mortality in patients admitted with subarachnoid hemorrhage in a large sample of Brazilian ICUs.

Methods: We performed a retrospective cohort study of patients admitted with spontaneous (SAH) to 57 hospitals in Brazil, during 2014 and 2015. We retrieved patients’ clinical and outcome data from an electronic ICU quality registry. SAPS 3 non-Neuro and SOFA non-Neuro scores were calculated subtracting the Glasgow Coma Scale contribution from the original score values. We used mixed multivariable logistic regression analysis to identify factors associated with hospital mortality.

Results: A total of 1114 patients were included. Fifty five percent (n=610) of patients were female and 71% were admitted from the emergency room, while 18% were transfered from another hospital. Median age was 57 (interquartile range 45 – 72) and median ICU length of stay was 5 (IQR 2 – 10). Median SAPS 3 non-Neuro score was 46 (38 – 55) and SOFA non-Neuro score was 2 (0 – 5). A total of 446 (30%) patients presented with poor grade SAH (World Federation of Neurosurgeons grading scale IV and V) and hospital mortality was 35%. In univariate comparisons, nonsurvivors were older and had higher SAPS3 non-Neuro and SOFA non-Neuro scores (all P<0.001). Poor grade SAH, use of vasopressors, mechanical ventilation, intracranial pressure monitoring, external ventricular drainage, blood transfusions and renal replacement therapy were all more frequent among nonsurvivors (all P<0.001). Mortality was also higher with initial lactate above 2 mmol/L, in those admitted to public hospitals and when admission to ICU was delayed more than 24 hours after ictus. After adjusting for common predictors (age, gender and WFNS) SAPS 3 non-Neuro, SOFA non-Neuro, early vasopressor use and admission to a public hospital were independently associated with hospital mortality. Moreover, the area under the curve for prediction of mortality with SAPS3, SOFA and WFNS was 0.86 (Figure 1).

Conclusions: Mortality is elevated and highly variable in this large sample of SAH patients. Age, severity of clinical presentation, both systemic and neurological, as well as the presence of organ dysfunction were associated with mortality.

Fig. 1 (abstract P224).

Area under the curve (AUC) for hospital mortality for variables: SAP 3 non-Neuro; SOFA non-Neuro and the combination of both plus WFNS Poor Grade

P225 Incidence, risk factors & outcome from acute kidney injury in non-traumatic subarachnoid haemorrhage patients on the neuro-intensive care unit

S Muldoon1, G Bose2, V Luoma1, G Bird3

1The National Hospital for Neurology & Neurosurgery, London, United Kingdom; 2University Hospital North Midlands, Neuroanaesthesia, Stoke-On-Trent, United Kingdom; 3The National Hospital for Neurology & Neurosurgery, Neurocritical Care, London, United Kingdom

Introduction: The incidence of acute kidney injury (AKI) after non-traumatic subarachnoid haemorrhage (ntSAH) is poorly quantified, purportedly between 5-27% depending on diagnostic criteria used [1,2]. This observational study aimed to; quantify incidence, qualify risk factors and outcome following AKI in patients with ntSAH on the neuro-intensive care unit (NICU).

Methods: A retrospective electronic case-note review of patients admitted with ntSAH to NICU between 2014-17. Exclusion criteria: NICU stay <48 hours & >4 days ictus to NICU admission. AKI was defined by KDIGO criteria [3]. Data analysed included: demographics, pre-morbid risk factors for AKI, WFNS SAH grade, treatment, radio-contrast load, development of delayed cerebral ischaemia (DCI) (defined as decision to induce hypertension to treat new neurological deficit) and infection (defined as decision to prescribe antibiotics), NICU mortality, NICU and hospital length of stay (LOS).

Results: 204 patients included (Table 1), incidence of AKI was 26.5%. The majority were stage 1, with 4.4% [n=9] developing stage 2&3 AKI; only 2 required renal replacement therapy. The median [range] time from ictus to AKI was 4 [0-23] days. AKI was associated with poor grade ntSAH [P=0.0035], diabetes [p=0.0040], vasopressors [P=0.0001], infection [P=0.0001] & DCI [P=0.0253]. Hypertension & total contrast load were not associated with AKI. AKI was associated with increased NICU mortality [P=0.0008], but not LOS except for patients with stage 2 & 3 [p=0.018]. A trend towards increased incidence of AKI in males and patients treated by clipping was observed.

Conclusions: Incidence of AKI in our study was comparable with published data [2] and associated with poor-grade ntSAH, DCI and NICU mortality. Further work will focus on the cause of AKI in these patients, where early occurrence of AKI may be a consequence of early brain injury.


1. Kamar C et al. Ulus Travma Acil Cerrahi Derg. 23:39-45.2017

2. Bercker S et al. PLoS ONE 13,2018

3. KDIGO. Kidney Int. Suppl 2:1–138,2012

Table 1 (abstract P225.) Comparison of groups

P226 An acute hemodynamic effect of intra-arterial nimodipine injection for symptomatic vasospasm in patients with aneurysmal subarchnoid hemorrhage

SH Park, TJ Kim, HB Jeong, SB Ko

Seoul National University Hospital, Department of Neurology and Critical care, Seoul, South Korea

Introduction: Intra-arterial (IA) nimodipine injection has been used in the treatment of refractory vasospasm after subarachnoid hemorrhage (SAH). IA nimodipine promotes dilatation of cerebral vessels and improves cerebral blood flow (CBF). However, its effect of duration when injected intra-arterially still remained to be elucidated.

Methods: A 76-year-old woman with SAH (H-H scale 4, mFisher 3, 3mm X 3.5mm, ruptured left middle cerebral artery [MCA] bifurcation aneurysm) was admitted to the neuroICU. Aneurysm was secured with coil embolization. On hospital day 6, she became drowsy and developed new right hemiparesis (GCS decreased from 15 to 13). Mean flow velocity in the left MCA on Transcranial Doppler ultrasonography was increased to 182cm/sec compared to baseline (60cm/sec). Continuous CBF flow monitoring was monitored using cerebral flow index (CFI) in C-FLOW. This patient was treated with IA nimodipine infusion under refractory vasospasm.

Results: Before initiating IA nimodipine, hemodynamic parameters were as follows (mean arterial pressure [MAP], 110mmHg; heart rate [HR], 64/min; right CFI, 48; left CFI, 32). Two minutes after IA nimodipine (5mg/hr), MAP decreased 85mmHg, and CFI was increased (Rt, 54; Lt, 50) while HR did not change (62/min). After 15 minutes of IA nimodipine, MAP dropped to 80mmHg, and CFI started to decrease on the left (46) while right CFI did not change. The IA nimodipine rate was decreased to 1mg/hr and norepinephrine was increased to 10mcg/minute. IA nimodipine was infused over 75 minutes. After the procedure, MAP was 92mmHg, HR was 66/minute, CFI was 54 on the right, and 50 on the left. Neurological status was improved and patients became more alert. 1 hour after stopping IA nimodipine, CFI returned to baseline (Right, 44; Left, 33) (Figure 1)

Conclusions: Acute hemodynamic effect of IA nimodipine can be assessed using C-FLOW. The positive effect of IA nimodipine on CBF lasted only 1 hour on this observation. More studies are needed to confirm this observation.

Consent: Informed consent to publish has been obtained from the patient

Fig. 1 (abstract P226).

Variation of CFI after IA nimodipine injection

P227 Association of duration and intensity of intracranial hypertension insults with outcome in subarachnoid hemorrhage: an observational study of two cities

G Carra1, F Elli2, L Huber3, F Güiza4, B Ianosi3, V Rass5, B Depreitere4, G Meyfroidt4, G Citerio2, R Helbok5

1KU Leuven, Department and Laboratory of Intensive Care Medicine, Leuven, Belgium; 2University of Milan-Bicocca, Monza, Italy; 3University of Health Sciences, Medical Informatics and Technology, Hall in Tirol, Austria; 4KU Leuven, Leuven, Belgium; 5Medical University of Innsbruck, Innsbruck, Austria

Introduction: Some patients with severe-grade subarachnoid hemorrhage (SAH) require intracranial pressure (ICP) monitoring. Current ICP treatment thresholds in SAH are derived from studies in patients with traumatic brain injury (TBI). The purpose of the study was to assess the association of intensity and duration of episodes of intracranial hypertension with neurological outcome in adult patients with SAH.

Methods: Retrospective analysis of 2 prospectively collected datasets, including time series of ICP, from 52 patients at the San Gerardo University Hospital, Italy and from 46 patients at the Innsbruck University Hospital, Austria. The association of intensity and duration of intracranial hypertension episodes with 12-month Glasgow Outcome Score (GOS) was visualized using the methodology introduced by Güiza et al. [1].

Results: In both cohorts, it could be demonstrated that the combination of duration and intensity defined the tolerance to intracranial hypertension, and that a semi-exponential curve separated episodes associated with better outcomes from those associated with worse outcomes. The association with worse outcomes occurred at a lower pressure-time burden than what has been previously observed in patients with TBI. Nevertheless, the percentage of monitoring time spent by every patient in the zone associated with poor GOS was independently associated with worse 12-month neurological outcome, even after correcting for age and Fisher score (p-values of 0.001 and 0.02 in Monza and Innsbruck respectively). The pressure-time burden curve for the Monza patients was shifted to the left compared to Innsbruck, which could only partially be explained by differences in baseline characteristics between the cohorts.

Conclusions: In two cohorts of adult patients with SAH, an independent association of the pressure and time burden of ICP, and worse clinical outcomes, could be demonstrated. The association occurs at lower ICP to what was observed in TBI.


1. Güiza et al. Intensive Care Med 41:6, 2015

P228 Predisposing factors of failed apnea test during brain death determination in potential organ donor

JJ Kim, EY Kim

Seoul St. Mary´s hospital, Department of General Surgery, Division of Surgical Critical Care and Trauma, Seoul, South Korea

Introduction: Apnea test is an essential component in the clinical determination of brain death, but it may incur a significant risk of complications such as hypotension, hypoxia and even cardiac arrest [1]. We analyzed the risk factors associated with failed apnea test during brain death assessment in order to predict and avoid these adverse events.

Methods: Medical records of apnea tests performed for brain-dead donor between January 2009 and January 2016 in our institution, were reviewed retrospectively. Age, gender, etiology of brain death, use of catecholamine and results of arterial bleed gas analysis (ABGA), systolic/diastolic blood pressure (SBP/DBP), mean arterial pressure (MAP) and central venous pressure (CVP) prior to apnea test initiation were collected as variables. A-a gradient and PaO2/FiO2 were calculated for more precise assessment of the respiratory system. In total, 267 cases were divided into a group which was completed apnea test and the other which was failed the test.

Results: 13 cases failed the apnea test and the majority of reasons were severe hypotension (SBP < 60mmHg). In terms of hemodynamic state, SBP was significantly higher in the completed test group than the failed group (126.5 ± 23.9 vs. 103 ± 15.2, respectively; p = 0.001). In ABGA, the completed test group showed significantly higher PaO2/FiO2 (313.6 ± 229.8 vs. 141.5 ± 131.0, respectively; p = 0.008) and lower A-a gradient (278.2 ± 209.5 vs. 506.1 ± 173.1, respectively; p = 0.000). In multivariable analysis, low SBP (p = 0.040) and high A-a gradient (p = 0.002) were independent risk factors associated with failed apnea test.

Conclusions: Although the unexpected adverse events during apnea test for brain death determination do not occur frequently, they could be fatal. If a brain-dead patient shows low SBP and high A-a gradient, clinicians should pay more attentions and preparations prior to apnea test.


Murthy T. Medical Journal, Armed Forces India. 2009;65:155-160.

P229 Neuroinvasive West Nile disease (NIWND): a case series of the 2018 Tunisia outbreak

D Ben Braiek, K Meddeb, N Fraj, W Ammar, C El Marzougui, W Zarrougui, M Boussarssar

Farhat Hached University Hospital, Medical Intensive Care Unit, Sousse, Tunisia

Introduction: Tunisia has already suffered recurrent outbreaks since 1997. 2018 outbreak started relatively earlier this year. We were interpellated by the frequency of neuroinvasive presentation of the disease.

Methods: We report a case series of 11 patients presented to ICU with NIWND.

Results: We report 11 cases of NIWND with different severe presentations overlapping neurological manifestation including encephalitis (n=8/11), meningitis (n=10/11) and flaccid paralysis (n=8/11). Almost all patients live in the locality of Sousse. Six patients presented a long course of isolated fever before developing neurological signs. Cerebrospinal fluid was consistent with encephalitis within the 11 patients. Cerebromedullar MRI identified brain lesions (n=8/10), myelitis (n=1/10) and polyradiculoneuritis (n=1/10).Three patients had electromyography for flaccid paralysis showed diffuse axonal polyneuropathy with motoneuron involvement. Ten cases had a positive WNV IgM antibody and nine had a positive WNV IgG antibody in serum. Urine polymerase chain reaction was positive for WNV in 8/10 patients. Ten patients were mechanically ventilated. All patients were managed symptomatically. Two received high doses of methylprednisolone for 3 days, one patient received polyclonal immunoglobulin intravenous and one patient had plasmapheresis. Two patients died consecutive to brainstem lesions. Two patients recovered significantly and discharged with no complications. Five other patients evolved to persistent flaccid paralysis with a minimal consciousness state and weaning difficulties requiring tracheostomy. The last remaining patient is still evolving.

Conclusions: Modification of the regional climatic conditions accounted probably for the early 2018 outbreak of NIWND. This initial case series displays the severity and the poor outcomes of NIWND with higher incidence compared to past epidemics.

P230 Noninvasive estimation of intracranial pressure with transcranial Doppler: a prospective multicenter validation study

C Robba1, C Iaquaniello2, A Mazeraud3, M Czosnyka4, D Savo5, M Saini6, G Citerio7

1Policlinico San Martino IRCCS for Oncology, Anaesthesia and Intensive Care, Genoa, Italy; 2Università degli Studi “Milano Bicocca”, Scuola di specializzazione in Anestesia, Rianimazione, Terapia Intensiva e del Dolore, Milano, Italy; 3Tertiary Hospital Européen Georges Pompidou, Paris, France; 4University of Cambridge, Department of Clinical Neurosciences, Cambridge, United Kingdom; 5 ASST-Monza, NeuroIntensive Care Unit, Monza, Italy; 6 ASST-Monza, Neurointensive Care Unit, Monza, Italy; 7Università degli Studi “Milano Bicocca”, Milano, Italy

Introduction: Invasive intracranial pressure (ICP) monitoring through an intraventricular or parenchymal catheter is crucial in neurological critically ill patients. Non-invasive bedside ICP measurement techniques have been developed but results are questionable. Transcranial Doppler (TCD) Ultrasonography has shown promising results, as variation in the flow velocity waveform may reflect ICP changes. The aim of our study is to assess if a published formula [1] based on the TCD diastolic flow velocity correlates with invasive ICP in a cohort of brain injured patients.

Methods: We designed a prospective multicenter observational study of patients admitted in two tertiary neurocritical care units (Monza, Italy and Addenbrookes Hospital, Cambridge, UK) with a diagnosis of acute brain injury that required invasive ICP (ICPi) monitoring. Non-invasive ICP (ICPtcd) values were derived from the flow velocities measured by the TCD of the middle cerebral artery (MCA): MCA pulsatility index (PIa) and an estimator based on diastolic flow velocity (FVd). We applied the Bland-Altman method, and assessed sensitivity and specificity of the method through the ROC and the AUC analyses.

Results: 115 patients were enrolled, 314 paired ICPi and ICPtcd measure were performed. 25 patients (29%) had at least one episode of intracranial hypertension (ICP>20 mmHg). In the Bland-Altman approach [Fig. 1], mean bias was -3.24 mmHg (limits of agreement are ± 2 SD 24.6 mmHg). 7.5% measures were outside the limit of agreement in the overall population. However, when ICP was high, 43% of measures were out of the limit of agreement. The AUC [Fig. 2] was 0.344 and 0.362 for ICPtcd and for PIa respectively, with a mean bias of -3.24 mmHg (SD 12.28 mmHg with limits of agreement -27.8 - 21.32 mmHg).

Conclusions: According to our results, the method is not enough reliable to be used in clinical practice for substituting invasive ICP monitoring. Further studies are needed to confirm the hypothesis.


[1] Czosnyka M et al. J. Neurosurg., 88:802–808, 1998

Fig. 1 (abstract P230).

Bland-Altman analysis

Fig. 2 (abstract P230).

ROC curve

P231 Non-invasive intracranial pressure monitoring for ICU management of cardiovascular patients

P Travassos1, R Vale1, W Costa1, C Hayashi1, G Vilela2, V Veiga1, J Souza3, S Rojas1

1Hospital BP - A Beneficência Portuguesa de São Paulo, Neurological ICU, Sao Paulo, Brazil; 2Braincare Desenvolvimento e Inovacao Tecnologica S.A., Research Consultant, Sao Paulo, Brazil; 3Hospital BP - A Beneficência Portuguesa de São Paulo, Cardiovascular Surgery, Sao Paulo, Brazil

Introduction: Surgical treatment of aortic aneurysm needs extracorporeal circulation (ECC), aorta clamp and hypothermia, and it is often related to poor systemic perfusion and blood flow velocity. One of the main concerns of intensive care team is to prevent secondary neurological injury after long time without blood flow pulsatility, such as brain edema and seizure. The most common parameters for neuromonitoring would be intracranial pressure and EEG, however, for non-neurological patients this information is unusual and prevents optimal management.

Methods: We aimed to assess brain compliance and neurological condition of ICU patients on immediate post-operative recovery of Bentall-De Bono procedure and/or other aortic aneurysm surgical treatment using a novel non-invasive intracranial pressure (ICP) device. This device uses mechanical displacement sensor capturing extracranial continuous volumetric variation of the skull and this information proportionally reflects intracranial dynamic[1].

Results: Twenty patients were included in this study. ECC mean time was 111 minutes for 19 patients and only one did not need it. Eleven presented altered ICP curves with poor brain compliance (P2/P1 ratio > 1.0) assessed by ICP curve morphology analysis. Volemic optimization and neuroprotective measures were taken based on this ICP information for acute case management. Among these patients with altered ICP curves, eight were discharged from ICU with good clinical condition and Glasgow Coma Scale of 15. Overall mortality rate was six out of twenty (30%) and three of these had altered ICP curves.

Conclusions: Brain monitoring of cardiovascular post-operative patients is important to prevent secondary neurological complications and can be a helpful tool for neuroprotective acute management on ICU.


[1] Cabella et al. Acta Neurochirurgica Supplement 122:93-6, 2016.

P232 Functional electrical stimulation in the critically ill – what can we learn from healthy volunteers?

P Turton1, B Johnston1, S AlAidarous2, I Welters1

1Institute of Aging and Chronic Disease, Liverpool, United Kingdom; 2Institute of Global Health, Liverpool, United Kingdom

Introduction: Muscle loss is common in critical illness. Functional Electrical Stimulation (FES) is a potential method of preventing this. The technique supplies electrical current to muscle, combined with passive cycling. Prior to a clinical trial, we first investigated the effects of one session of FES in healthy volunteers.

Methods: Healthy male volunteers (n=15) were recruited. The participants had their postural sway assessed on a pressure sensitive board, and measurement of Maximal Inspiratory Pressure (MIP). Ultrasounds were taken assessing thickness of the quadriceps and rectus abdominis. They performed 20 minutes of supine passive cycling, with FES supplying the lower limbs and abdomen. After a 1 minute rest, the tests were repeated. A further 14 participants performed just the initial baseline tests, to help assess muscular factors affecting balance and sway.

Results: The current needed for palpable contraction was significantly correlated to weight in the abdomen (r=0.79, p<0.001) and quadriceps (r=0.82, p<0.001). Current required to stimulate the abdominal muscles was also correlated to depth of the subcutaneous fat layer (r=0.85, p<0.001) and echogenicity of the muscle (r=0.65, p=0.012). Pre-cycling, left and right vastus lateralis thickness inversely correlated to postural sway in the antero-posterior (r=-0.638, p<0.001) plane. Compared to pre-cycling, postural sway in the antero-posterior and lateral planes increased significantly after cycling. There was a significant decrease in MIP after cycling and greater reductions in MIP were found in participants who had thinner rectus abdomni.

Conclusions: Sway at baseline is related to quadriceps thickness, which atrophies during critical illness, and could worsen balance. MIP is reduced during FES and the severity of reduction is related to the thickness of the abdominal wall muscles at baseline, suggesting that FES can fatigue the diaphragm and abdominal muscles. In awake healthy volunteers, FES is a safe, comfortable technique.

P233 The prevalence of postoperative cognitive dysfunction after off-pump cardiac surgery

V Sharipova, A Valihanov, A Alimov

Republican Research Centre of Emergency Medicine, Uzbekistan, Anesthesiology, Tashkent, Uzbekistan

Introduction: In most cases postoperative cognitive dysfunction (POCD) is transient, but still some patients suffer from persistent cognitive impairment which is associated with increased length of hospital stay, early withdrawal from labor market and higher mortality. Available data on the prevalence of POCD after cardiac surgery is very diverse from 20% to 90% upon discharge and up 20% 3 months after surgery. We aimed to investigate the prevalence of short-term and long-term POCD after off-pump coronary artery bypass grafting (CABG) surgery.

Methods: Psychometric testing was performed in 230 (mean age 63.5±8.2) patients before, 10 days and 6 months after the surgery. We used following tests to assess cognitive capacity: auditory verbal learning test (AVLT), digit span test (DST), digit-letter substitution test (DLST), Stroop’s test and trail making test (TMT). A decline in comparison to preoperative test results for 20% or more in two or more tests was declared as POCD.

Results: The prevalence of POCD after 10 days was 31.7% (73 patients) and 9.1% (21 patients) after 6 months. When comparing patients who developed POCD with those who did not we found the former were older (69.2±8.7 vs 61.1±10.3 years; p<0.001), had lower education level (13.7±2.1 vs 10.6± 2.4 years; p<0.05) and had longer surgery duration (250.4±12.2 vs 232.1±15.9 minutes; p<0.05). The most affected cognitive domains were long term memory (AVLT) and executive function (TMT) and least affected – working memory (DST) and selective attention (Stroop’s test).

Conclusions: In our prospective study the prevalence of long-term POCD after cardiac surgery was slightly less (9.1%) in comparison to available data (from 9% to 20%). It might be due differences in psychometric testing and interpretation of its results among authors. Advanced age, low cognitive reserve and long duration surgeries are linked with higher incidences of POCD.

P234 Hospital length of stay increases postoperative cognitive dysfunction

V Sharipova, A Valihanov, A Alimov

Republican Research Centre of Emergency Medicine, Uzbekistan, Anesthesiology, Tashkent, Uzbekistan

Introduction: Postoperative cognitive dysfunction (POCD) is a common and widely described phenomenon in surgical patients. Advanced age, major surgery, certain general anesthetics, genetic factors, sleep deprivation and other factors were described as contributing factors to POCD. The hospital stay itself is a major ‘social’ trauma for patients; social isolation, sleep deprivation and changes in daily regimen may effect neurocognitive behavior of patients. In this trial we tried to assess the link between POCD and the length of hospital stay in cardiac surgery patients.

Methods: 94 patients who underwent ‘off-pump’ coronary artery bypass grafting (CABG) surgery selected for this trial. Neuropsychological testing was performed prior to the operation and upon discharge. We used auditory verbal learning test (AVLT), digit span test (DST), digit-letter substitution test (DLST), Stroop test and trail making test (TMT). A 20% or more decline in two or more tests in comparison to preoperative test results was declared as POCD. Patients were allocated into two groups according to the length of hospital stay: the SHORT-STAY group (group 1) included patients (n=36) who were discharged on the 8th day after surgery or earlier and the LONG-STAY (group 2) group consisted of patients (n=58) who were discharged on the 9th day after surgery or later. Patients received similar anesthesia, postoperative care and were operated by the same surgical team. Reasons for prolonged duration of hospital stay were mainly surgical.

Results: 11 patients (30.6%) in group 1 and 21 patients (36.2%) in group 2 had POCD upon discharge (p<0.05). Mean length of hospital stay were 7±1.2 and 10±1.4 days in group 1 and group 2 patients respectively (p<0.05).

Conclusions: Prolonged length of hospital stay increased the prevalence of POCD in our trial. Studies with various types of surgical procedures and larger patient populations needed to further understand the effect of length of hospital stay to POCD.

P235 The influence of multiple trauma with head trauma on posttraumatic meningitis: a nation-wide study with hospital-based trauma registry in Japan

Y Katayama1, T Kitamura2, T Hirose1, Y Nakagawa1, T Shimazu1

1Osaka University Graduate School of Medicine, Department of traumatology and acute critical medicine, SUITA, Japan; 2Osaka University Graduate School of Medicine, Division of Environmental Medicine and Population Sciences, Department of Social and Environmental Medicine, SUITA, Japan

Introduction: Posttraumatic meningitis is one of severe complications and results in increased mortality and longer hospital stay among head trauma patients. However, it remains unclear whether there is a difference in the incidence of post-traumatic meningitis due to single traumatic brain injury (TBI) and multiple trauma including head injury.

Methods: This study was a retrospective observational study during 12 years We included trauma patients registered in Japanese Trauma Data Bank whose head AIS score was >3 in this study. Multivariable logistic regression analysis was used to assess potential factors associated with posttraumatic meningitis such as CSF fistula, skull base fracture, type of injury that divided into single TBI and multiple trauma.

Results: Among 60,390 patients with severe head injury, 9415 (15.6%) patients were multiple trauma patients and 50,975 (85.4%) were single TBI patients (Table 1). The incidence of posttraumatic meningitis was 0.6% (61/9,415) in multiple trauma patients and 0.4% (223/50,975) in single TBI patients, respectively. Multiple trauma was associated with posttraumatic meningitis compared with single TBI (adjusted odds ratio [OR] 1.415 [95% confidence interval {CI}; 1.055-1.900]) (Table 2).

Conclusions: In this population, multiple trauma patients were associated with posttraumatic meningitis among trauma patients.


Sonig A, et al. Neurosurg Focus. 2012 ;32: E4

Adepoju A, Adamo MA. J Neurosurg Pediatr. 2017; 20: 598-603

Table 1 (abstract P235). Patient´s characteristics
Table 2 (abstract P235). Result

P236 Determination of NIRS and BIS values in traumatic braine injury patiens

C Balci1, E Haftaci1, H Aytuluk1, H Durmus1, R Sivaci2

1Kocaeli Derince Traning Hospital, Intensice Care Unit, Kocaeli, Turkey; 2AfyonKocatepe Universty, afyon, Turkey

Introduction: The aim of this study was to determine if regional cerebral oxygenation (rScO2) can be used as an indicator of tissue perfusion in ICU patients with TBI [1, 2], and to determine the prognostic value of cerebral oxygenation rScO2 in survival prediction.

Methods: Patients were enrolled retrospectively from January 2012 through July 2018 in the ICU of Derince Kocaeli Training Hospital. 250 patients with trauma patients and traumatic braine injury patients who were admitted to the ICU from the emergency room were included in the study. The sedation levels of the patients were followed up with BIS. The rScO2, BIS was taken as well as blood lactate level, mean arterial blood pressure and cardiac output at baseline time, 6, 12, 24, 48 and 72 hours.

Results: No significant difference was also detected between the value of rScO2 in all patients . It was Average ScO2 (right) 60.54 ±4.8 and Average rScO2 (left) 55.63± 5.4.

Conclusions: Cerebral regional oxygen saturation might be helpful as one of the perfusion parameters in patients with TBI but it could have no prognostic value in mortality prediction. However, further studies with larger sample size are still needed to validate these results.


1. Dunham CMet al. J Trauma. 52:40-46, 2002.

2. Balci C, et al. J Int Med Res 46:1130-1137, 2018.

P237 Traumatic brain injury in elderly: impact of frailty on outcome

L Zacchetti1, S Aresi1, R Zangari2, G Cavalleri1, L Fagnani1, L Longhi1, P Gritti1, F Ferri1, L Lorini1

1Ospedale Papa Giovanni XXIII, Dipartimento di Emergenza Urgenza e Area Critica, Bergamo, Italy; 2FROM, Fondazione per la Ricerca Ospedale di Bergamo, Bergamo, Italy

Introduction: TBI in elderly is an increasingly cause of admission in ICU. Data regarding management and prognosis of these patients are lacking. Validated prognostic models refer to younger patients and do not adequately consider the influence of pre-injury functional status, which often compromises with aging. Frailty has been defined as a state age-related of increased vulnerability and decline in autonomy of daily life activity. Aim of the study is to evaluate the impact of frailty on outcome in TBI elderly patients.

Methods: moderate and severe TBI patients >65years, admitted in NeuroICU from January 2017 to May 2018, were prospectively enrolled. Data of age, comorbidity, Glasgow Coma Scale (GCS), pupils’ reactivity, CT scan characteristics, neurosurgical intervention and GOSE (Extended Glasgow Outcome Scale) at 6-months were collected. Frailty status was measured by Clinical Frailty Scale (CFS) [1] and patients were divided as frail (CFS>4) and not frail (CFS<4). Bad outcome was defined as GOSE<4.

Results: 22(37%) of the 60 studied patients were frail. Frailty was not related to age. Frail patients had more comorbidities and worse pupils’ reactivity at admission (Table 1). Other variables did not differ between groups. In univariate analysis neurological diseases, GCS, tSAH (traumatic subarachnoid haemorrhage), compressed/absent basal cisterns, non-reactive pupils and CFS were significantly associated to bad outcome. In multivariate analysis only GCS and CFS remained associated to bad outcome (Table 2).

Conclusions: pre-injury frailty is strongly associated to outcome in TBI elderly patients.


1. Rockwood et al. CMAJ 173:489-95, 2005

Table 1 (abstract P237). Characteristics at admission
Table 2 (abstract P237). Multivariate analysis

P238 Efficiency of nerve block anesthesia of the scalp with low-volume local anesthetic as a component of anesthetic management of craniotomies

D Markevich1, A Marochkov2, V Sedin1, Y Zhloba1, N Raikova1, A Borozna1

1Mogilev city emergency hospital, department of anesthesiology and intensive care, Mogilev, Belarus; 2Mogilev Regional Hospital, Department of Anesthesiology and Intensive Care, Mogilev, Belarus

Introduction: Efficacy evaluation of the scalp nerve blocks with low-volume ropivacaine as a component of combined anesthesia for perioperative analgesia in patients with neurosurgical interventions

Methods: A randomized prospective study was conducted. The efficiency of 275 blockages of the peripheral nerves of the scalp in 54 patients with neurosurgical interventions on the head was evaluated. The age of the patients was 48.4±15.6 years. Patients were operated on for intracranial traumatic (39 cases) and non-traumatic hematomas (5), brain tumors (4) and the need for plastic of postoperative skull defects (6). General endotracheal total intravenous anesthesia with fentanyl, propofol, rocuronium, or tracrium was used. After tracheal intubation, 4-12 nerves were blocked (e.g., supraorbital, supratrochlear, zygomaticotemporal, auriculotemporal, great auricular, greater and lesser occipital nerves), depending on the surgical site. 0.75–1.0% ropivacaine was used. For blockade of one nerve used 0.5-2.0 ml of local anesthetic. Fentanyl was applied on section of a periosteum, dura matter and at inefficiency of blockade of nerves. Anesthesiology monitoring included HR, ECG, SpO2, NIB, respiratory parameters, EEG (CSI), body temperature, blood glucose and lactate levels. In 2 and 10-12 hours post-surgery, the intensity of pain was ranked by alert patients using VAS.

Results: the volume of local anesthetic for blockade in one patient was 7.7± 1.7 ml. In 5 (9.3%) from 54 patients, an additional fentanyl injection was required to skin incision due to an increase in blood pressure and heart rate by 25% of the baseline values, and an increase in CSI until 80 un. Patients available to productive contact in 2 hours post-surgery ranked the pain by VAS at 1 (0;2) point, and in 10-12 hours post-surgery ranked it at 2 (1;3) p.

Conclusions: At patients with craniotomies scalpe-block with low-volumes of a ropivacaine showed high efficiency (90.7%).

P239 Comprehension of oxidative stress in ICU patients

D Markopoulou1, A Karagianni2, K Venetsanou3, E Papadaki1, K Vlachos2, I Alamanos1

1KAT General Hospital, B ICU, Kifisia, Greece; 2KAT General Hospital, Neurosurgery, Kifisia, Greece; 3KAT General Hospital, Research Unit B ICU, Kifisia, Greece

Introduction: The oxidative-antioxidative balance (REDOX is critical in Intensive Care Unit (ICU) patients Oxidative stress disease development strongly depends on major REDOX molecules Malondialdehyde (MDA) and Superoxide Dismutase (SOD) which are activated under inflammatory conditions. Total antioxidant capacity TAC is the sum of all antioxidant circulating molecule. The aim of this study is to investigate early REDOX markers in ICU patients.

Methods: Forty two patients hospitalized in ICU enrolled in the study, 28 neurosurgical patients with CNST and 14 with traumatic brain injury (TBI). A single peripheral blood sample was collected from each patient before surgical intervention or upon admission to ICU. Major REDOX molecules, Malondialdehyde (MDA) and Superoxide Dismutase (SOD), as well as Total Antioxidant Capacity (TAC) were determined by spectrophotometry.

Results: TAC levels were lower in CNST patients when compared with the TBI group (p<0.05). On the opposite, MDA was found to be significantly increased in the neurosurgical group when compared with that of injured subjects (p<0.05). Although no significant differences found in SOD levels, distinctive alterations were revealed by studying SOD/MDA ratio.

Conclusions: REDOX markers, TAC and MDA were associated with patient diagnosis and were reversed between TBI and CNST individuals. Redox balance could be evaluated for ICU patient’s management.

P240 The microbiology of central nervous system infections of neurosurgical patients treated in the intensive care unit

G Bouboulis1, E Drosos2, M Nepka3, I Karaminas1, C Vrettou4

1Evangelismos Hospital, First Critical Care Department, School of Medicine, University of Athens, Athens, Greece; 2Evangelismos Hospital, First Department of Neurosurgery, School of Medicine, University of Athens, Athens, Greece; 3Evangelismos Hospital, Microbiology Department, Athens, Greece; 4Evangelismos Hospital, First Department of Intensive Care, Athens, Greece

Introduction: Our purpose is to identify the species and antibiotic resistance of the microorganisms that cause hospital acquired central nervous system (CNS) infections in neurosurgical patients treated in the intensive care unit. We also examine how often the CNS pathogens were previously isolated in samples other than the cerebrospinal fluid (CSF).

Methods: This is a retrospective epidemiological study. From the electronic medical files we retrieved the demographic clinical and microbiology data of the neurosurgical patients treated in our department from 1/1/2013 to 31/12/2017. The hospital’s ethics committee approved the study protocol.

Results: We found 42 positive CSF cultures from 31 neurosurgical patients with suspected CNS infection. The reasons for ICU admission and the presence of foreign bodies in the CNS are shown in Table 1. Figure 1 shows the isolated microorganisms. Resistant to Carbapenems were 77% of the gram negative pathogens and 44% of the gram positives. Resistant to Colistin were 30% of the gram negatives and there was one case of Vancomycine resistant enterococcus. In 50% of the cases the CSF isolates were previously cultured in the same patient, either in brochial secretions (37%), or in blood cultures (13%).

Conclusions: Our results are in agreement with recent published literature pointing that Acinetobacter and Klebsiella sps. are becoming more frequent pathogens of hospital acquired CNS infections than coagulase negative Staphylococci [1, 2]. The increasing emergence of Carbapenem resistance among these gram negatives complicates the empirical and definitive antibiotic treatment. The importance of previous infection and/or colonization with resistant pathogens in neurosurgical patients warrants further study.


1. O´Horo JC, et al. Neurocrit Care. 2017;27:458-467.

2. Tunkel AR et al. Clin Infect Dis 2017;00:1–32

Table 1 (abstract P240). Admission diagnosis and CNS foreign bodies for the study population
Fig. 1 (abstract P240).

Microorganisms cultured from the cerebrospinal fluid in the study population

P241 Association of electrolytic disturbances with increased mortality in the intensive care unit

J Gomez, MRRamos, A Tognon, R Gradaschi

Hospital São Vicente de Paulo, CTI Central, Passo Fundo - Rs, Brazil

Introduction: To assess the prevalence of electrolyte disorders at Intensive Care Unit (ICU) admission and its relation with mortality as an independent risk factor [1].

Methods: We performed a retrospective observational study including all patients >12 years-old admitted for more than 48 hours at the ICU of Hospital São Vicente de Paulo from January to August 2017. Data was entered using Microsoft Excel and the statistical analysis was performed using IBM SPSS Statistics version 22.0 for Windows. Mortality risk factors were identified using binary logistic regression models. Laboratory variables that significantly improved model’s predictive capacity compared to the model that considered only chronic kidney disease history and diagnosis of acute renal failure at admission were kept in the final model. Predictive capacity change was tested using likelihood test and was considered significant when P-value was < 0.15.

Results: 419 patients were admitted to the ICU, 405 were included in the study population and 14 (3.3%) were excluded due to missing data. The mean age was 53.7 ± 19.3 years and 237 (58.2%) of them male, chronic renal disease 23 (5.7%) and acute renal failure 63 (15.6%). Table 1 describes laboratory baseline characteristics of the study population. Regarding the outcome of ICU patients: 153 (37.8%) died, 197 (48.6%) were transferred to hospital ward or 53 (13.1%) to the center of intensive nursing care; 2 (0.5%) went to the surgical recovery room. Acute renal failure, hypernatremia and hyperphosphatemia were independent predictors of mortality as described in Table 2.

Conclusions: Hypernatremia and hyperphosphatemia were independent predictors of mortality in critically ill patients.


Bucley MS et al. Crit Care Med 2010; 38[Suppl.]:S253–S264

Table 1 (abstract P241). Serum electrolytes on ICU admission
Table 2 (abstract P241). Mortality predictors at ICU (n = 405)

P242 Relation between hydroelectrolytic disturbances and residence time in neurocritical patients

P Travassos, R Vale, W Geres, V Veiga, S Rojas

Hospital BP - A Beneficência Portuguesa De São Paulo, Neurocritical Care Unit, Sao Paulo, Brazil

Introduction: The objective of this study is to evaluate the relationship between electrolytic disturbances and length of stay in intensive care units in neurocritical patients.

Methods: All hospitalized patients were evaluated in a 3-month period in a large hospital neurological intensive care unit, considering sodium, potassium, magnesium and calcium analysis. During the period, 307 patients were admitted, of which 170 were female (55.4%), with a mean age of 59.85 years, 59.6% of which were surgical and the mean length of stay was 8.05 days. Statistical analysis was performed, using Cox regression, with time-dependent covariables, being considered a significance of 0.05.

Results: In the analyzed sample, in relation to sodium variations, RR was found to be 0.63, with p = 0.009 for hyponatremia and RR of 0.57, with p = 0.083. In the evaluation of magnesium, a RR of 0.54 (p = 0.06) was found for hypomagnesemia and RR of 1.51 (p = 0.415) for hypermagnesemia. In the potassium variations, RR of 0.26 (p = 0.001) was observed for hypokalemia and RR of 1.50 (p = 0.33) in hyperkalemia. In hypocalcemia, OR 0.85 was observed with p = 0.44 and for hypercalcemia, OR 0.611 (with p = 0.206).

Conclusions: In the analyzed sample, hyponatremia and hypokalemia were the electrolyte changes that correlated with increased residence time in ICU

P243 Strong ion difference assessment: point-of-care or central laboratory?

T Langer1, S Brusatori2, P Brambilla2, C Ferraris Fusarini2, R Maiavacca2, G Giudici1, A Zanella1, G Grasselli2, F Ceriotti2, A Pesenti1

1Università degli Studi di Milano, Critical Care and Emergency, Milan, Italy; 2Fondazione IRCCS Ca´ Granda Ospedale Maggiore Policlinico, Anesthesia, Critical Care and Emergency, Milan, Italy

Introduction: The Strong Ion Difference (SID) is essential for the assessment of acid-base equilibrium, thus requiring an accurate measurement of plasma electrolytes. Currently there is no gold standard for electrolyte measurements and SID computation. Differences in electrolyte values obtained with point-of-care (PoC) and central laboratory (Lab) analyzers have been reported [1, 2]. In previous studies [3, 4] we have shown that changes in PCO2 induce electrolyte shifts from red blood cells to plasma (and vice versa), yielding variations in SID. Aim of the present in-vitro study was to induce SID changes through acute changes in PCO2 and compare values of electrolytes and SID obtained with PoC and Lab techniques.

Methods: Blood samples from 10 healthy volunteers were tonometered (Equilibrator, RNA Medical) with 3 gas mixtures at fractions of CO2 (FCO2) of 2, 12, and 20%. Electrolytes were measured quasi-simultaneously with a PoC analyzer (ABL800 FLEX, Radiometer) and a routine Lab method (COBAS 8000 ISE, Roche). For both techniques a simplified SID was computed as sodium + potassium – chloride.

Results: Bland-Altman analysis of SID calculated with PoC and Lab showed a proportional bias (slope = 0.64, r2 = 0.55, p <0.0001), indicating a variable agreement between methods according to the average SID value (Fig.1). SID values measured with PoC and Lab at different FCO2 differed significantly (p<0.001, Fig.2). A similar discrepancy was observed for chloride (p <0.001, Fig.2), while sodium (p=0.439) and potassium (p=0.086) were similar.

Conclusions: SID measured with PoC and Lab differed significantly, mainly due to a variable discrepancy in chloride. Our findings suggest that our POC analyzer is superior to the Lab in measuring electrolytes and thus compute SID.


1. Morimatsu H et al. Anesthesiology 98:1077-1084, 2003;

2. Jain A et al. Int J Emerg Med 2:117-120, 2009;

3. Langer T et al. Crit Care 22(suppl 1):82, 2018;

4. Langer T et al. J Crit Care 30:2-6, 2015.

Fig. 1 (abstract P243).

Bland-Altman analysis of simplified Strong Ion Difference (SID) calculated with point-of-care (PoC) and central laboratory (Lab) techniques. X-axis represents the mean of the two measurements, while Y-axis represents their difference. Bias is represented as the horizontal solid line (-3.66 mEq/L); standard errors of the bias are represented as horizontal dotted lines; limits of agreement (±1.96 SD) are represented as horizontal dashed lines (-10.26 and 2.94 mEq/L)

Fig. 2 (abstract P243).

Variations in simplified SID and chloride concentration at incremental fractions of CO2 (FCO2) in gas mixture used for tonometry. Point-of-care (PoC) and central laboratory (Lab) data were compared via Two Way Repeated Measures ANOVA. * = p<0.05 as compared to Lab

P244 Comparison of two insulin regimens for the treatment of hyperkalemia in critically ill patients with renal insufficiency

J DeCleene, T Lam, J Jancik

Hennepin County Medical Center, Pharmacy, Minneapolis, United States

Introduction: This study evaluated the safety of half dose insulin (HDI) versus standard dose insulin (SDI) for the treatment of hyperkalemia in a medical intensive care unit (MICU) population with renal insufficiency. Recent emergency medicine data demonstrated a lower incidence of hypoglycemia in patients with renal insufficiency when HDI was used for the treatment of hyperkalemia [1]. There is limited data describing the safety of HDI in a MICU population with renal insufficiency.

Methods: This was a retrospective, chart review of patients admitted to the MICU with a diagnosis of AKI and/or CKD stage 3-5 with a serum potassium ≥ 5.8 mEq/L from January 2013 to September 2018. SDI is defined as 10 units of regular IV insulin and HDI as 5 units. The primary outcome was the incidence of hypoglycemia within 12 hours of insulin administration. Secondary outcomes included severe hypoglycemia and change of serum potassium after insulin administration.

Results: A total of 265 patients were screened and 98 were included for analysis. The incidence of hypoglycemia occurred in 6/45 patients (13.3%) and 6/53 patients (11.3%) who received SDI and HDI, respectively. One patient in the SDI group and two patients in the HDI group developed severe hypoglycemia. The mean decrease in serum potassium after insulin administration was 1.0 mEq/L in both groups. Patients in the HDI group who were re-dosed with 5 units of regular insulin did not have any hypoglycemic events.

Conclusions: In a MICU population with renal insufficiency, SDI and HDI regimens appear safe and effective for the treatment of hyperkalemia.


Mcnicholas BA, et al. Kidney Int Rep. 3:328-336, 2018.

P245 Hypernatremia in critically ill patients

EH Mestrom1, AJ Bindels1, JG Van der Hoeven2

1Catharina Hospital, Intensive Care Unit, Eindhoven, Netherlands; 2Radboud University Medical Center, Intensive Care Unit, Nijmegen, Netherlands

Introduction: This study aims to provide new insight into predictive variables in the development of hypernatremia in critically ill patients. Current literature lacks large scale retrospective studies or studies including more variables which are considered relevant in a patient with hypernatremia [1].

Methods: A retrospective case control analysis was performed using prospectively collected data in the Intensive Care Unit of the Catharina Hospital in Eindhoven, Netherlands. The dataset includes demographic details, clinical data such as fluid balances and laboratory parameters, medication prescriptions and ICU and hospital outcomes. Primary outcome was the development of mild (>145 mmol/L) or severe (>149 mmol/L) hypernatremia. Case-control matching was performed for age, length of stay and APACHE IV scores.

Results: A total number of 5836 ICU records were included for analysis. 1159 (19.9%) patients developed hypernatremia during their ICU stay, and 543 of these patients (9.3%) developed severe hypernatremia (Table 1). After case control matching, 910 matched patients were included in the final analysis. The multivariate analysis found no strong predictors for the development of mild or severe hypernatremia (Table 2). After case control matching, multivariate analysis for mortality in case of development of hypernatremia above 145mmol/L or 149mmol/L resulted in OR 2.84 (CI 1.51-5.33) and OR 2.2 (p 0.010, 95% CI 1.21-3.91) respectively.

Conclusions: Despite including more patients and more relevant variables than before, this study was not able to identify strong predictors for the development of hypernatremia. Nevertheless, the results showed both mild and severe hypernatremia during ICU admission as independent predictors of mortality.


Waite et al. J Crit Care. 2013;28(4):405-12.

Table 1 (abstract P245). Baseline characteristics
Table 2 (abstract P245). Results of multivariate analysis after case control matching

P246 Hypernatremia in septic patients: epidemiology and prognostic impact

M Shirazy1, K Bousselmi1, D Abduljabar2

1King Hamad University Hospital, Critical Care Department, Muharraq, Bahrain; 2King Hamad University Hospital, Internal Medicine Department, Muharraq, Bahrain

Introduction: Sepsis and septic shock are common causes of admission in the Intensive Care Unit with a high mortality rate [1,2]. Hence, electrolyte disturbances are common in this group of patients. Acute hypernatremia is one of the multiple features of homeostasis disturbances and available data in the literature suggest that its incidence can reach 47% [3,4].

Methods: A retrospective observational study from August 2017 to November 2018 on the patients with a primary diagnosis of sepsis admitted to the ICU in King Hamad University Hospital within 48 h after the onset of the disease. The database contained age, sex, diagnosis, initial APACHE II, initial and daily SOFA scores, Na levels, outcome, and ICU and hospital length of stay. All statistical tests were two-tailed, and the significance level was set at P<0.05. Data were analyzed with SPSS.

Results: Among 159 patient, 89 patients fulfilled the inclusion criteria. 32 of the included patients (35.9%) developed hypernatremia during the study period. Mortality rate was higher among the hypernatremic patients 22 (68.75%) compared to the eunatremic patients 11 (32.35%) (Fig 1, 2). The main source of sepsis was pneumonia with 37 affected patients (41.57%).

Conclusions: Hypernatremia is significantly associated with higher mortality in septic patients.


1. Levy MM et al. Intensive Care Med 2018;44:925-928.

2. Andre SF et al. Applied Soft Computing 2016 (42); 194-203.

3. Ni HB et al. Am J Med Sci 2016; 351; 601-605.

4. Bataille S et al. BMC Nephrol 2014; 15: 2-9

Fig. 1 (abstract P246).

The risk of hypernatremic patient´s mortality associated with specific APACHE II scores

Fig. 2 (abstract P246).

The outcome versus the sodium levels

P247 Calcium disorders in medical intensive care: a Moroccan experience, epidemiologic data and prognosis

H Ezzouine, A Karchi, A Raja, K Mediouni

Faculty of Medicine and Pharmacy; University Hassan II, Medical Intensive Care, Casablanca, Morocco

Introduction: Calcium disorders are frequent among intensive care patients. Our study aimed to determine the epidemiological and clinical profile of the patients who developed hypocalcemia or hypercalcemia during their hospitalization in our unit.

Methods: We conducted a retrospective study in the medical intensive care unit of the university teaching hospital Ibn Rushd of Casablanca-Morocco for one year.The epidemiological, clinical and therapeutic data were collected and analyzed as was the clinical evolution.

Results: The incidence of hypocalcemia was 24.12%,96 patients were identified, their average age was 42.15 years old and 55.21% were female.Their mean Charlson comorbidity index was 1.65, APACHE II 10.96, SAPSII 29.06 and OSF 3.94.Neurological pathology was the main reason for admission 36.46%. 83.33% developed hypocalcemia before the 5th day from admission.48.96% had mechanical ventilation, blood transfusion was necessary for 37.50% of the patients.The mortality rate was 31.25%. The incidence of hypercalcemia was 4.27%., 17 patients were identified, their mean age was 53.82 years old and 35.29% were female .The mean charlson comorbidity index was 4.41, APACHE II was at 10.05, SAPS II 30.05 and OSF 3.88. Metabolic diseases were the main reason of admission 47.06%. 100% of these patients had hypercalcemia in their admission .35.29% were intubated and ventilated, 58.82% received diuretics, 47.05% received biphosphonates.The mortality rate was 52.94%. The factors associated with mortality were, for hypocalcemic patients, APACHE II, SAPS II, early hypocalcemia, mechanical ventilation, vasoactives drugs, blood transfusion and length of hospitalization and for hypercalcemic patients the prognosis factors were APACHE II, low serum albumin and sedation.

Conclusions: Calcium disorders are associated with a high mortality risk. The main prognosis factors for these patients are APACHE II, mechanical ventilation, albumin rate and its timing.

P248 Prognostic value of serum phosphorus level in patients with multiple organ dysfunction syndrome after abdominal surgery

SA Tachyla, AL Lipnitski, AV Marachkou

Mogilev Regional Hospital, Department of Anesthesiology and Intensive Care, Mogilev, Belarus

Introduction: The aim of this study was to determine the prognostic significance of the dynamic measurement of serum phosphorus levels in patients with multiple organ dysfunction syndrome (MODS) after abdominal surgery.

Methods: 33 patients in the department of anesthesiology and intensive care unit with MODS after abdominal surgery were included in a prospective cohort study (20 men and 13 women, age 55.1±16.4 years, weight 78.9±18.4 kg.). All patients were divided into two groups according to hospital outcome: survivors (group 1, n=23) and nonsurvivors (group 2, n=10). Patients in the groups did not have statistical differences by sex, body weight and height.

Results: Mortality was 30.3%. In the group 2 the age of patients was higher 64.3±15.1 years vs 52.1±15.8 years (p=0.037). Apache III score was significantly higher in the group 2 – 55.4±19.4 vs 38.7±21.9 (p=0.01). The frequency of use of vasoactive drugs was significantly higher in the group 2 – 80% vs 39.1% (p=0.031). There were no significant differences between the groups in length of stay in the ICU. In group 1, there was an increase of serum phosphorus level and in the group 2 – the tendency to decrease. However, statistically significant differences were obtained only on the 2nd day after surgery 1.74±0.84 mmol/l (group 1) vs 0.88±0.26 mmol/l (group 2) (p=0.01). The ROC curve was constructed to assess the predictive significance of serum phosphorus levels (Fig. 1). AUC was 0.687; 95% CI 0.574-0.799; p=0.002; sensitivity 55.2%, specificity 83.9%. The Kaplan-Meier survival analysis (Fig. 2) showed that the survival rate was significantly higher in patients with serum phosphorus content >0.81 mmol/l (Gehan-Wilcoxon test, statistics=-3.92, p<0.001).

Conclusions: Constantly decreased levels of serum phosphorus indicate a poor prognosis in patients with multiple organ dysfunction syndrome after abdominal surgery.

Fig. 1 (abstract P248).

ROC-curve for serum phosphorus levels

Fig. 2 (abstract P248).

Kaplan-Meier survival analysis

P249 Obesity is not a risk factor for extubation failure in intensive care unit

A De Jong1, HQ Quintard2, RC Cinotti3, KA Asehnoune3, JM Arnal4, CG Guitton5, CP Paugam-Burtz6, PA Abback6, A Mekontso-Dessap7, KL Lakhal8, SL Lasocki9, GP Plantefeve10, BC Claud11, JP Pottecher12, PC Corne13, CI Ichai14, ZH Hajjej15, NM Molinari16, GC Chanques17, LP Papazian18, EA Azoulay19, SJ Jaber1

1PhyMed E