Nosocomial candidemia in an Italian tertiary care hospital

In a retrospective ongoing study, the incidence of nosocomial candidemia among patients (pts) admitted to a tertiary-care teaching hospital (bed capacity 2400; 78 beds for intensive care units, ICUs) was evaluated together with causative pathogens, potential risk factors, antifungal treatment, and crude mortality. During the primary study period (November 1991—October 1994) there were 136 episodes of candidemia occurring in 136 pts (median age 62 years, range 4–96). The overall incidence of candidemia was 11.3 episodes per 10 000 hospital admissions. The underlying diseases were the following: solid or hematological malignancies (43 pts), major abdominal surgery (30 pts), cardiovascular diseases (18 pts), trauma (17 pts), other diseases (28 pts). At the onset of candidemia 67 pts (49%) were located in ICUs, 48 and 21 pts respectively in surgical and medical wards. C. albicans (83 strains) accounted for 53% of all blood culture Candida isolates. Fifty-three pts (39%) received adequate antifungal treatment: 16 pts amphotericin-B and 37 pts fluconazole. The overall crude mortality was 54%. The number of positive fungal blood cultures (1 culture versus >1 culture) did not influence crude mortality (54% versus 63%). In addition, the mortality of pts infected by C. albicans (55%) was similar to that of pts infected by C. non-albicans species (45%). The mortality of pts located in ICUs (79%) was significantly higher (P < 0.001) than that of pts in surgical (29%) and medical wards (33%). Finally, the mortality of pts who did not receive adequate antifungal therapy (70%) was significantly higher (P < 0.001) than that of treated pts (30%). In conclusion, the incidence of nosocomial candidemia was high during the primary phase of the study; most of our pts with candidemia had severe underlying diseases and were hospitalized in ICUs. The number of fungal positive blood cultures did not influence the crude mortality, confirming that a single bood culture shuld not be dismissed as benign transient candidemia. On the contrary, about two-third of our pts did not receive an adequate antifungal treatment and the majority of them died.

R Re es su ul lt ts s: : Control of bleeding was achieved with 1 to 20 (m ± SEM: 5.86 ± 0.93) bronchoscopic interventions. Hemostasis was accomplished in a period of 0.5 h and 10 days. Cardiocirculatory instability was observed in five patients. One patient died because of persistent bleeding caused by severe aspergillosis. Six patients survived without further interventions. A high volume low pressure (HVLP) cuff does not protect the lower airway from contamination by material leaking along longitudinal folds within the cuff wall [1]. This is a major factor in the pathogenesis of ventilator associated pneumonia [2]. The combination of shape and high compliance of the Portex Soft-Seal cuff might eliminate the folds in the cuff walls circumferentially for a portion of the cuff and prevent leakage. We have tested the Soft-Seal cuff in a pig trachea model to establish whether protection against leakage is better than that afforded by standard HVLP cuffs. We wished to measure changes in lung volume (∆LV), airway pressures, and oxygenation during tracheal suctioning performed with a CSS and with an open suction system (OSS). We enrolled 7 adult patients, sedated and paralyzed, VCV ventilated by a SERVO 900C ventilator (Siemens, Sweden) with PEEP ≥5 cmH 2 O and FiO 2 ≥ 0.4. Keeping all remaining ventilatory settings unchanged, we set trigger sensitivity at -2 cmH 2 O, inspiratory time at 25%, inspiratory pause at 10%. We performed four suctioning manouvers at 20 min intervals using alternatively a CSS and an OSS. With both systems, we used 12 F size catheters. We performed no pre-oxygenation manouvers. Suction was applied for 20 s at a pressure of 100 cmH 2 O. We continuously recorded signals of respiratory inductance pletismography (RIP, Respitrace Plus, NIMS, FL), arterial oxygen saturation (O 2 Sat) by pulse oxymetry, and airway pressures. We obtained ∆LV as the change in the RIP signal measured during VCV and during suction. We measured Respiratory Rate (RR), peak inspiratory pressure (PIP), positive end-expiratory pressure (PEEP), and mean airway pressure (MAP) during VCV and during suction with the CSS. R Re es su ul lt ts s: : variables are reported as mean ± DS. C Co om mm me en nt t: : the use of the OSS resulted in discontinuation of ventilatory support with a loss in lung volume and in O 2 2 Sat. The CSS effectively preserved lung volume and oxygenation by maintaining airway pressures during the suction manouvre. The increase in RR observed with the CSS was due to activation of the trigger mechanism.  B Ba ac ck kg gr ro ou un nd d: : In trauma patients, rigid cervical collar placement reduces head extension (HE) during laryngoscopy [1]. In patients with difficult airway, upper teeth or gums may be traumatized by excessive laryngoscope blade levering motion (LBLM) needed for laryngeal visualization [2]. The current study aims to compare, under stimulated spine precautions, HE and LBLM upon maximum glottic exposure (MGE) achieved with #4 conventional Macintosh blade (CMB) and #4 modified Macintosh blade (MMB) carrying two 10 Foley catheters (Fig. 1).

VCV CSS OSS
M Me et th ho od ds s: : Anaesthesia was induced in 17 male, ASA I, Mallampati I, elective surgery patients. Spine precautions included rigid board placement under the shoulders and occiput and a rigid collar placement round the neck. Laryngoscopy was performed twice, changing between MMB and CMB. Before each laryngoscopy, the patients head was placed in the neutral position. MMB laryngoscopy technique consisted of MMB tip insertion into vallecula, right catheter balloon inflation with 2 ml air and MMB elevation until MGE achievement. The angles of laryngoscope handle axis (Fig. 2 AH) and of maxillary molars occlusal surface axis (OS) relative to horizontal (angles â 1 and â 2 in Fig. 2) were recorded upon MGE. Angles â 1 and â 2 were measured with an automatic angle finder (Fig. 1). The difference of 90°-â 2 was defined as HE angle and the difference â 1 -â 2 was defined as LBLM angle (angle â 3 in Fig. 2), He and LBLM angles were compared with paired t test; P < 0.05 was considered statistically significant.
R Re es su ul lt ts s: : MMB laryngoscopy resulted in significantly less HE and LBLM than CMB laryngoscopy (P < 0.001). Results and summarized statistics are presented in the Table. Values are shown as  I In nt tr ro od du uc ct ti io on n: : We describe our experience with the TLT technique, which is a purely dilatational PDT with low inherent risks. The technique has the additional benefit of maintained ventilation and airway protection.
T Te ec ch hn ni iq qu ue e: : The TLT consists of a reinforced tracheostomy tube, with an integral dilator, which is pulled out between tracheal rings following retrograde insertion through the larynx [2]. A cuffed oral 5mm-tracheal tube inserted past the proposed stoma site maintains ventilation and airway protection. We prospectively collected data in 103 consecutive patients, 56 males and 47 females, undergoing this technique. The authors (JWF & AK) performed tracheostomies on all patients (16 to 88 years old). Pre-existent coagulopathy was not corrected. Indications for tracheostomy were mainly for term ventilation (39) and weaning difficulties (44). R Re es su ul lt ts s: : 102 tracheostomies were performed successfully. One was converted to a Ciaglia technique after accidental decannulation. Mean duration of operative procedure was 13.9 min. The INR ranged from 0.8-2.6, (mean 1.3), platelets ranged from 23-667 × 10 9 (mean 184 × 10 9 ). There were six transient episodes of hypoxia (SpO 2 <90%), three cases of hypotension, two related to the anaesthetic technique and one following traumatic intubation. There were four episodes of accidental decannulation and one case of minor subcutaneous emphysema. There was one case of moderate blood loss (100-250 ml). There was one episode of loss of airway, in a patient who was difficult to intubate (Gr. III). We had two cases of wound infection associated with pre-existent systemic bacteremia. Total duration of the tracheostomy ranged from 1-65 days. Total closure of the stoma took a mean of 4 days (range 2-9 days). The resultant scar was minimal.
C Co on nc cl lu us si io on n: : This pure dilatational and bronchoscopically visualised method is easy to perform with training. It is worthy of consideration in patients with coagulation abnormalities. We feel it offers better control over the airway than other available techniques although there is a definite risk of decannulation while withdrawing the cannula over the obturator. The overall morbidity of this technique is low.
Independent lung ventilation (ILV) is effective for the patient who is suffering from unilateral lung disease. When we ventilate the patients with ILV, they should be intubated with a doublelumen endobronchial tube. While ILV is continued for some time a number of difficulties related to the management of the doublelumen endobronchial tube (DLT) arise. Movements of the patient and routine turning of the patient threaten the DLT position and can lead to loss of lung isolation or lobe occlusion. Nasal intubation is better suited for long-term intubation than oral intubation because it is safer for equipment attachment. We have ventilated six patients (Table) with ILV using the DT by nasotracheal I In nt tr ro od du uc ct ti io on n: : Postextubation stridor is a serious problem in children with an incidence of up to 33% in electively intubated children. Our aim was to determine whether steroids decreased reintubation rates and to identify other risk factors for reintubation.
M Me et th ho od ds s: : Retrospective analysis (1994)(1995)(1996) of the 82 children (72 received steroids). Steroids were categorized according to the type used and the time of administration. Recognized risk factors for postextubation stridor including age (<1 and >1 year) and duration of intubation (<120 and >120 h) were analyzed. R Re es su ul lt ts s: : There was no significant difference in either the preintubation grade or stridor (P = 0.67) in both outcome groups (reintubated 22/23 grade 3 and not-reintubated 50/59 grade 3) or in the postextubation grade of stridor between both groups (P = 0.1). Neither type of steroid (P = 0.32), nor time administered (P = 0.79), nor age (P = 0.22) nor duration of intubation (P = 0.35) was found to significantly influence reintubation rates. C Co on nc cl lu us si io on n: : The prophylactic use of corticosteroids in routine elective extubations for laryngotracheobronchitis cannot be rec-ommended, based on current findings. Overall, 28% of all patients needed to be reintubated. However, reintubation seems to be correlated best with atelectasis rather than the degree of postextubation stridor. ventilation. We evaluated the potential advantages of the AHME over a conventional active humidifier.
M Me et th ho od ds s: : The study included seven mechanically ventilated patients. In each patient, the AHME was used for 24 h and then substituted with a conventional active humidifier (F&P) (MR730, Fisher & Paykel) with a heated wire in the inspiratory limb, for the next 24 h. AHME was preset to keep the temperature of inspired gases at 37°C. The F&P was set to 37°C in the humidifier-chamber, and to 37°C at the Y piece. The AHME and the F&P were compared in terms of: humidity and temperature output, water consumption and condensate in the water traps.
The humidity output was evaluated on the basis of the condensate in the flex tube, which was scored from 0 (absent) to 3 (excessive).
R Re es su ul lt ts s: : Minute ventilation did not differ during application of AHME and of F&P. Both devices kept the set temperatures, and provided adequate humidification, as assessed by the condensate in the flex tube. However, when the F&P was used, there was formation of condensate in the ventilator tubings, and the water traps needed to be emptied on average eight times (range: 6-9) per day.
No condensation of water was found in the ventilator tubings with AHME. Compared with F&P, the AHME remarkably reduced the water usage.
C Co on nc cl lu us si io on n: : Compared to a conventional active humidifier, the AHME provides equivalent humidification, with the advantages of both reducing the time-expenditure for handling, and of eliminating the risk caused by water condensation in the ventilator tubings.

MY Yassin Libanese University School of Medicine, Department of Internal Medicine, Pulmonary and Critical Care Division, Hammoud Hospital, Sidon-Lebanon
Crit Care 1999, 3 3 ( (s su up pp pl l 1 1) ):P15 I In nt tr ro od du uc ct ti io on n: : VAP is a serious infection with a mortality rate exceeding 50%. It also leads to an increase in the duration of the treatment and adds to hospital costs. Bacteria, in intubated patients, may be directly inoculated into the endotracheal tube from the hands of medical personnel or from contaminated respiratory therapy equipment (i.e. humidifiers). We tried a heat and moisture exchanger to substitute the conventional ventilator humidifiers to prevent VAP in the ICU setting.
M Me et th ho od ds s: : Subjects were intubated and attached to the conventional respiratory assistance cascades in the first year of the study (July 1992-June 1993). Retrospectively, cases of VAP were calculated prospectively, during the following year (July 1993-June 1994), subjects were intubated and attached to respiratory assistance cascades; but PALL filter, a heat and moisture exchanger, was in-line and the machine humidifiers were bypassed. The cases of VAP were calculated.
S St tu ud dy y p po op pu ul la at ti io on n: : Intubated ICU patients with normal CXR on admission to the unit. R Re es su ul lt ts s: : VAP rates decreased in the group of HMEF dramatically in comparison to the conventional humidification method (see Table below). C Co on nc cl lu us si io on n: : We concluded that heat and moisture exchanger filters can prevent VAP in short term mechanically ventilated ICU patients, and can halve its rate in long term durations.  The adequacy of humidification of heat and moisture exchangers (HMEs) during long-term mechanical ventilation is still controversial. Recently, an active HME (AHME) (Humid-Heat, Gibeck) has been developed. This AHME combines a HME with a unit which adds water and heat between the patient and the HME.
The AHME automatically regulates the water and heat supply. The only user-set input for AHME is the minute ventilation (V'e) of the patient. We evaluated the AHME efficiency for humidification during long-term mechanical ventilation.
M Me et th ho od ds s: : The AHME was used for 5 days on seven patients which were mechanically ventilated in different modes. On each day we measured the number of tracheal aspirations, the secretions characteristics, the condensate in the flex tube and in the water traps, the airway temperature, the number of changes of the V'e setting on AHME. A chest X-ray and a bronchoscopy were performed on days 1, 3 and 5. We scored the secretions characteristics and the condensate in the flex tube from 0 (insufficient) to 3 (excessive), the atelectasis at chest X-ray from 0 (absent) to 2 (evident), and the bronchial occlusions at bronchoscopy from 1 (absent) to 4 (complete).
R Re es su ul lt ts s a an nd d c co on nc cl lu us si io on n: : AHME provided adequate humidification over the 5 days, as indicated by the secretions characteristics and by the absence of new atelectasis and of secretions accumulation in the bronchi. The temperature of inspired air was adequate. The value of V'e set on the AHME was changed on average twice (range: 0-8 times) per day, to maintain this setting close to the V'e of the patient. No water condensate was found in the water traps. The AHME is adequate for humidification in long-term mechanical ventilation, and eliminates the problem of condensation in the ventilator tubings. The humidification efficiency of AHME is not influenced by the mechanical ventilation mode, provided that the V'e setting of AHME is kept close to the V'e of the patient. Nr. of changes of V'e set on AHME 0.9 ± 1.2 2 ± 1.9 1.9 ± 1.3 1.9 ± 2. B Ba ac ck kg gr ro ou un nd d: : Heated humidification (HH) is commonly used with or without a heated wire circuit (HWC) to humidify inspired gases during mechanical ventilation (MV). We compared HH and HH with a HWC to a new active heat and moisture exchanger (AHME). The AHME (Humid Heat, Gibeck, Sweden) consists of a typical HME and a heat and water source delivered between the patient and the HME. The volume of water delivered and heat output are based on a set minute ventilation. A pre-set airway temperature of 37°C is used.
M Me et th ho od ds s: : Thirty patients requiring MV for >72 h were studied. Pts received humidification via a HH, HH + HWC (Fisher & Paykel), and AHME in random sequence for 24 h each. All devices were set to deliver 37°C at the proximal airway. During each period of ventilation, the following were measured; airway temperature, min and max body temperature, number of suctioning attempts, volume of secretions, consistency of secretions, number and volume of saline instilled, water usage, condensate, ventilator settings, minute volume, number of circuit disconnections. Water usage was measured by weighing the water bag before and after 24 h use. Consistency of secretions were judged as thin, moderate, or thick as previously described (Suzukawa: Respir Care 1989, 3 34 4:976). Condensate was measured by emptying fluid into a graduated container and sputum volume measured by collecting secretions in a Luken's trap. Airway temperature was measured at the ET tube using a rapid response thermistor. Resistance of the AHME was measured before and after use. R Re es su ul lt ts s: : There were no differences in any of the variables related to humidification efficiency (secretion volume and consistency, number of suctioning attempts, or volume of saline used). Water usage and volume of condensate were significantly different between devices, but delivered airway temperatures were not. Statistical analysis was done with ANOVA. *P < 0.05, see Table. M Me et th ho od ds s a an nd d r re es su ul lt ts s: : The Nottingham Physiology Simulator is a validated simulation of advanced, iterative physiological models [1]. The model was set up as a 70 kg adult with normal physiological values other than: pulmonary venous admixture 20%, alveolar deadspace fraction 20% of tidal volume and functional residual capacity 2 l. The patient's lungs were ventilated with 100% oxygen for 2 min and the patient was then apnoeic with an open airway exposed to 21, 50, 80 or 100% oxygen. Arterial oxygen and carbon dioxide tensions (PaO 2 , PaCO 2 ) were recorded continuously until arterial oxygen saturation fell to 50%. The changes in PaO 2 and PaCO 2 are shown in the figure on the previous page.
D Di is sc cu us ss si io on n: : Provision of very high ambient oxygen fractions greatly extends the safe duration of apnoea. As oxygen fraction is increased, increasingly large effects are achieved. The aim of the study was to evaluate the prevalence of OSA in hypercapnic ARF patients and its correlations with the severity and length of nocturnal arterial oxygen desaturation, diurnal arterial carbon dioxide (PaCO 2 ) and oxygen (PaO 2 ) tensions, diurnal oxygen saturation, sudden death and BMI. The polysomnography consisted of continuous polygraphic recording (by Compumedic Sleep PTYLTD Abbotsford) from surface leads for electroencephalography, electrooculography, electromyography and ECG, and from noninvasive sensor for nasal airflow, tracheal sounds, body position, thoracic and abdominal respiratory efforts, and oxymyoglobin level. The number and duration of nocturnal sleep apneas and hypopneas and the consequential oxygen desaturation were evaluated; sleep apnea was defined as more than five episodes of apnea or hypopnea per hour of sleep (apnea/hypopnea index = AHI >5). Furthermore BMI, basal diurnal PaCO 2 , PaO 2 and arterial oxygen saturation were also recorded.
R Re es su ul lt ts s: : With the PEEP valve set at 5 or 10 mbar, pressures within the nasal mask were 5.6 ± 0.8 and 9.4 ± 1.0 mbar, respectively. Mean intratracheal pressures increased in all patients and were significantly higher during 10 mbar mask pressure compared to 5 mbar (6.8 ± 0.3 vs. 2.9 ± 0.5 mbar; P < 0.007). The relative amount of mask pressure transmitted into the trachea was significantly higher with 10 compared to 5 mbar (P < 0.04) (Figure). With 5 mbar of nCPAP, in 50% of the subjects, significant negative pressure swings occured during inspiration. This was not the case with 10 mbar.  Table. C Co on nc cl lu us si io on n: : Subjective (dysnea Brog index), objective respiratory response (hypercapnia) and level of IPAP pressure applied during NMV were influenced by a specific design of element intercalates at mask-tube-ventilator line. A specific design of these elements as we showed with a home portable single tube ventilator could affect NMV efficacy in hypercapnic COLD exacerbations.   respirator through the mouthpiece with the nostrils clipped. Pressure trigger was set at -1 cmH 2 O in all respirators and in ES 900 flow trigger was also tested. After 1 min warm-up, 1 min breathing test was performed at the end of which volunteers were asked to classify their satisfaction with respirator. At first, 5 cmH 2 O CPAP was tested at random in all four settings (three respirators, in ES 300 for both pressure and flow triggering) and thereafter the evaluation continued similarly with CPAP 5 cmH 2 O + 10 cmH 2 O pressure support. Data are presented as means ± SD, Kruskal-Wallis test was used for statistical analysis, P < 0.05 was considered significant.

P23 Evaluation by volunteers of respirator characteristics in modes used in non-invasive ventilation
R Re es su ul lt ts s: : NPPV was successful in 14 of 18 COPD patients (77.7%), in all 7 patients with cAPE (100%) and in 3 of 7 patients with SP (42.8%). Failure in 4 COLD patients was due to mask intolerance in three cases and to sudden death in one case. Four patients with SP (three seriously immunocompromised) died. COLD patients were ventilated for 3 to 62 h (mean 21.5 h), cAPE patients for 4 to 15 h (mean 7.4 h) and SP patients for 12 to 148 h (mean 59.7 h). Ventilation was longer in SP patients who obtained a therapeutic benefit (mean 112 h) than in SP patients who did not (mean 23 h). C Co on nc cl lu us si io on n: : With the limits of this observational study, we conclude that NPPV has been shown to be an effective support therapy for COLD patients with acute exacerbation and for hypercapnic severe cAPE patients. The use of NPPV in patients with SP was less effective and warrants ulterior study to be validated, according to literature.  C Co on nc cl lu us si io on n: : NMV in asthma crisis refractory to (OMV) is a safe alternative to ETI, and could be avoided in selected patients (50%). Borg Dysnea score index and respiratory rate at 3 h: 38 ± 6 to 25 ± 6 rpm in NMV group are the best early clinical predictors.  The NIV in acute respiratory failure of a previously healthy lung is not much widespread but much discussed. We report the first data about four patients, who have been accepted in our ICU due to acute respiratory failure post-trauma and treated with ventilatory support via face mask like NIV. All patients were negative to pre-existing lung disease and got thoracic trauma with multiple costal fractures and bony fractures. We used the mechanical ventilator Adult Star (Infrasonic, Inc., San Diego USA). All the patients were co-operating and without neurological deficiency. The NIV has been applied for 2 days and alternated with spontaneous ventilation through Venturi mask after 24 h. R Re es su ul lt ts s: : the analyzed data show an improvement of PaO 2 in all patients, already after the first hours of treatment as well as a respiratory rate reduction.

P25 Noninvasive mechanical ventilation in asthma crisis: an alternative ventilatory therapy to endotracheal intubation
D Di is sc cu us ss si io on n: : The NIV has to be considered as a conventional ventilation's kind also by acute hypoxemic respiratory failure. The admission's criteria of the patients to this kind of ventilation is however important. In conclusion, we can affirm that the NIV has an important advantage compared to the conventional ventilation, that is a shorter stays in the intensive care unit, associated to a reduction of pneumonia related to endotracheal intubation.   M Me et th ho od ds s: : Inclusion criteria: acute respiratory insufficiency in a period (0-48 h): respiratory rate >30 rpm, increase accessory respiratory muscular activity, hypoxemia PaO 2 <60 mmHg at mask venturi (FIO 2 : 0.5) after a period of 'T' piece or PSV and almost four consecutive weaning failure trials. Excluded: hemodynamic instability (SAT <90 mmHg), uncooperative patients, and excessive secretions., IPAP/EPAP cmH 2 O to achieve: >10 ml/Kr and decrease in dysnea Borg score. Continuous cardiorespiratory monitoring.
C Co on nc cl lu us si io on n: : APRV may be used safely in patients with ALI/ARDS and decreases the need for paralysis and sedation compared to PCV. APRV increases cardiac performance with decreased pressor use and CVP in patients with ALI/ARDS. Further study of ARPV is warranted to discover its impact on resource utilization and patient outcome. C Co on nt te ex xt t: : Children with a history of neonatal respiratory disease that required mechanical ventilation, who develop subsequent bronchopulmonary dysplasia, often have abnormal pulmonary function. The extent to which the neonatal respiratory disease alone is involved is not clear.

P29 Pulmonary function in children who were on long-term mechanical ventilation due to neonatal respiratory disease
O Ob bj je ec ct ti iv ve e: : To evaluate the association between neonatal respiratory disease without bronchopulmonary dysplasia on discharge and pulmonary function later in childhood.
D De es si ig gn n: : Case-control study.
S Se et tt ti in ng g: : Ambulatory follow-up of former intensive care patients at a university medical centre. P Pa ar rt ti ic ci ip pa an nt ts s: : Eighteen children aged 11-15 years with a history of neonatal respiratory disease were randomly recruited, regardless of gestational age or cause of disease. Inclusion criteria: mechanical ventilation for >14 days; high inspired oxygen fraction for >2 days (FiO 2 >0.4). Exclusion criteria: presence of bronchopulmonary dysplasia or other acute or chronic pulmonary disease at the time of this investigation. Eighteen controls matched for age, sex and height were recruited from children of the hospital staff. All were healthy at birth and had no pulmonary disease at the time of this investigation. All parents gave informed consent.
P Pu ul lm mo on na ar ry y f fu un nc ct ti io on n t te es st ts s: : Vital capacity (VC); forced expiratory volume in the first second (FEV 1 ) with and without challenge by the bronchoconstrictor methacholine; diffusing capacity (D L CO); airway resistance (Rt) with and without methacholine challenge; and thoracic gas volume (TGV).
M Ma ai in n o ou ut tc co om me e m me ea as su ur re es s: : Variables of pulmonary function in the cases. Differences between the cases and controls were compared using the paired-sample t-test.
R Re es su ul lt ts s: : Both FEV 1 and Rt differed significantly (P < 0.01) between children who had had respiratory disease as neonates (cases) and controls. There were no significant differences in VC, D L CO and TGV (Table).
Differences in VC and FEV 1 between cases and controls after methacholine challenge were not significant; however, this analysis is of limited value because only eight or nine matched pairs underwent these tests.
C Co on nc cl lu us si io on n: : A mild degree of airway obstruction is apparent in children 11 to 15 years after neonatal respiratory disease, even in the absence of bronchopulmonary dysplasia or other pulmonary disease.

T Ta ab bl le e. . Pulmonary function tests: mean ± standard deviation
Group VC (n = 18) ml FEV 1 (n = 18) ml D L CO (n = 18) (mmol/min)/kPa Rt (n = 16) kPa/l/s TGV (n = 15) ml  3 3 ( (s su up pp pl l 1 1) ):P30 After major abdominal or thoracic surgery, the patient may develop rapid shallow ventilation because of splinting, pain or heavy sedation. This may lead to the development of post operative atelectasis and pneumonia. Hence, it seems reasonable to expect that the administration of large tidal volumes (VT) during post operative mechanical ventilation will prevent or decrease the incidence of post operative pulmonary complications including that of pneumonia. Whether or not this is true is yet to be determined. Therefore, we performed the following prospective study. We hypothesized that large VT mechanical ventilation after major operations resulted in a lower incidence of post operative pneumonia.
Adults admitted to the surgical intensive care unit for post operative mechanical ventilation after major abdominal or thoracic surgery were placed on one of two VT regimens: 9 ml/kg (group 1) or 14 ml/kg up to a maximum of 1000 ml (group 2). Patients who were not placed on the correct VT regimen and those whose tidal volumes were changed during the study were excluded. Standard ICU monitoring was instituted. In addition, ventilator performance, peak inspiratory pressures, blood gases and daily chest Xrays were monitored. The incidence of post operative pneumonia was recorded. Results were analyzed by SPSS statistical software. Results: Forty-nine patients completed the study, 29 in group 1 and 20 in group 2. Their mean age was 52.7 years. There were 28 males and 21 females. Thirteen of 49 patients (26.5%) developed post operative pneumonia. A comparison of the two groups is shown below: C Co on nc cl lu us si io on n: : Post operative ventilation with large tidal volumes does not reduce the incidence of pneumonia.  3 3 ( (s su up pp pl l 1 1) ):P31 I In nt tr ro od du uc ct ti io on n: : Bronchoscopy is an important diagnostic and therapeutic tool in modern intensive care medicine. In ventilated patients it can lead to hemodynamic instability and can compromise the gas exchange .
M Me et th ho od d: : We evaluated in a prospective study from 3/1997 to 9/1998 indications, complications and side effects of bronchoscopy in a 12 bed medical ICU. The vital signs of the patients were monitored continuously by ECG, invasive blood pressure and S a O 2 measurement. A BGA was performed 5 min before and 5/30/120 min after the examination.
After an interruption the ECG showed sinus rhythm and the hemodynamic stabilised again. In a 24 h period after bronchoscopy the patient died because of an acute myocardial infarction.
C Co on nc cl lu us si io on n: : Bronchoscopy is a safe procedure in critical ill mechanically ventilated patients. Even in patients with BAL, in the most cases only a slight decrease of the PFR could be observed. The lowest PFRs were observed 30 min after bronchoscopy, 90 min later the PFR was almost back to the starting level. Critical PaO 2 values were only seen in rare cases. Complications could be handled in all cases. The death of one patient in a 24 h range after bronchoscopy was probably caused by the underlying disease and is to be seen only in temporal coincidence. There was significant delay from FES presentation to ICU admission for subacute FES patients requiring ventilatory support than the patients improving with conservative therapy alone (P < 0.05). One patient in subacute (2.3%) and 10 patients in fulminant FES group (45.5%) died. The compliance of respiratory system (C rs ) at the start of intermittent positive pressure ventilation was significantly less in fulminant as compared to subacute FES patients (P < 0.05). Most of the ventilated patients had initial improvement in C rs with ventilation but only those patients who made continuous improvement in C rs beyond 48 h of ventilation ultimately maintained oxygenation and survived in both the groups. We conclude that early ICU admission and supportive therapy is important determinant of morbidity in FES. Patients with more number of abnormal laboratory parameters and those in whom C rs and oygenation index does not improve even after 48 h of adequate ventilatory support are unlikely to improve by conventional ventilatory support alone and need to be shifted to other modalities of maintaining oxygenation. Respiratory syncytial virus (RSV) is nowadays the leading cause of bronchiolitis and viral pneumonia in children. Although the course is often benign, some children need prolonged hospitalisation and mechanical ventilation or even ECMO if conventional mechanical ventilation (CMV) fails.

P33 Acute respiratory distress syndrome in a University Hospital ICU in Japan
HFOV is currently considered to be contraindicated in obstructive airway disease with prolonged time constants due to the risk of dynamic airtrapping. This could give rise to circulatory and ventilatory compromise and barotrauma. Nevertheless, bronchiolitis patients are sometimes put on HFOV after deterioration on CMV.
We report 9 patients with RSV bronchiolitis and pulmonary overdistention (small airway disease) successfully treated with HFOV after deterioration on CMV. Although marked hyperinflation was present in all our patients prior to transition, no airleaks developed during HFOV. In one patient the oxygenation index (OI) increased after start of HFOV. Nitric oxide was added and oxygenation improved immediately. All patients survived without residual lung disease.
In distinct to current opinion, we showed that small airway disease can safely and successfully be managed with HFOV. Ventilatory strategy should be directed to open up the small airways and keep them open with sufficiently high mean airway pressures ('the open airway strategy' similar to the 'open lung strategy' in diffuse alveolar disease). Permissive hypercapnia may be used to reduce pressure swings as much as possible, leading to less shear stress on lung tissue, without influencing airway recruitment. Further dynamic airtrapping can be prevented with the use of longer expiratory than inspiratory times and with prevention of spontaneous breathing. An increasing OI at 48 h may be an indicator of failure of HFOV and alternative treatments should be considered. NO might be such an option to avoid ECMO. as well as kinetic therapy (KT) and hemofiltration (HF) did not lead to a breakthrough in treatment of severe ARDS. M Me et th ho od ds s: : We studied 22 consecutive patients with severe ARDS (mean age 64 ± 11.16 [SE] years) in a clinical follow-up design. All patients received hFiO 2 , IRV and PEEP before starting prone position, while 15 obtained HF (Prisma ® , Hospal) and 3 KT (Rotorest ® ). Prone position was commenced 82 h median time (range 6 to 417 h) after onset of severe ARDS at a mean PaO 2 /FiO 2 ratio of 98.02 ± 6.11 (SEM) mmHg. We compared individual oxygenation index (PaO 2 /FiO 2 ) before and after start of prone position with linear regression analysis (Excel ® regression-procedure; SPSS ® T-test). R Re es su ul lt ts s: : In the stage of supine position neither treatment with hFiO 2 , IRV, PEEP nor HF and KT led to an improvement of oxygenation index. After starting prone position ventilation 20 of 22 patients showed a significant increase of the oxygenation index (responder: Y [ Renal failure of these two non-responders was not treated by HF. Improvement of oxygenation index was independent of duration in supine before the begin prone position (range 6 to 417 h). In one patient PP was started actually after 417 h of treatment at our Intensive Care Unit. C Co on nc cl lu us si io on n: : Starting prone position seems to mark a U-turn for oxygenation for the majority of patients with severe ARDS, while application of high fractional inspired oxygen, inverse ratio ventilation, positive end exspiratory pressure as well as kinetic therapy and hemofiltration do not necessarily improve oxygenation. The timing of this non invasive technique primarily depends on the decision to turn the patient from supine to prone. We recommend prone position in ARDS as soon as possible to reduce lung injury and complications resulting of mechanical ventilation. Supramaximal flow is characteristic of the cough manoeuvre and is thought to be the result of dynamic compression of collapsible airways. We investigated the effect of changing the pulmonary time constant on the peak flow rate produced by an in vitro cough manoeuvre. We used a prototype artificial cough generator and a simplified lung-airway model. The model consisted of a compliant bag with a resistor (internal diameter 3 mm to 7 mm) that emptied through a collapsible tube. The resulting range of emptying time constants (420 to 2800 ms) included those found in vivo (500 to 2000 ms). The lung-airway model was inflated to one of two pressures (31 cmH 2 O or 55 cmH 2 O) and then compressed within a glass container to a pressure of 45 cmH 2 O. A mechanically-operated glottis opened rapidly and the resultant flow was measured by a pneumotachograph. The cough peak flow rate (CPFR) was recorded for 20 cough manoeuvres for each configuration and the values of the mean and standard deviation are shown in the Table. The results from this bench-test model suggest that the pulmonary time constant has a profound effect on the magnitude of the cough peak flow rate. Mechanics and gas exchange can in be studied with a computercontrolled ventilator. The physiological profile obtained describes the Pel/V diagram, inspiratory and expiratory resistance versus volume and the expired volume of CO 2 versus tidal volume. When setting PEEP, frequency, I:E ratio, and minute ventilation or inspiratory pressure the physician needs to assimilate the information of the physiological profile and all clinical information to assure an adequate gas exchange at a non-traumatic ventilation. In ALI/ARDS harm can be caused both by ventilation at too low lung volumes and by ventilation at high volumes. In COPD the task is to ventilate at the lowest possible volume and airway pressure.

P36
The complexity of the physiology and ventilator settings makes it impossible to figure out what is the ideal pattern of ventilation in order to reach the immediate therapeutic goals defined by the physician. However, on the basis of an adequate mathematical physiological profile, a computer can by simulation prognosticate what would be the consequences of alternative modes of ventilation. Through repeated simulations the physician can search a mode of ventilation that leads to his goals.
Computer simulation can be used to: a) increase the understanding of various patterns of ventilation in disease. b) predict the consequences of alternative settings in a particular patient.
In left diagram the total pressure in the ventilator (Pvent), the tracheal pressure (Ptr) and the alveolar, i.e. the elastic pressure, Pel, M Me et th ho od ds s: : Ninteen post-operative pts requiring MV were studied. Vent settings by physician were noted and each pt was placed on those settings or ASV randomly. IBW was determined from standardized tables and %MinVol was set to 100%. PEEP and FiO 2 were determined by staff and held constant. ABG's and cardiopulmonary variables (f, V T , V E , T i , PIP, P aw , HR, MAP, and VCO 2 ) were measured and recorded after 30 min on each mode. Data were compared using student's t-test. R Re es su ul lt ts s: : 19 pts (14 male) were studied. Initial 'test breaths' during ASV were well tolerated. Mean IBW was 88.8 Kg. Mean age was 54.3 years. Table 1 reveals set and measured ventilator parameters for both study periods. PIP and V T were lower during ASV. Respiratory rate was higher during ASV. V E , T I , and P aw were unchanged between study periods. Mean values for PEEP and FiO 2 were 7.3 and 0.48, respectively. Table 2 reveals ABG measurements, CO 2 production, and V D /V T ratio. There were no clinically relevant differences in ABG's or VCO 2 between study periods. V D /V T was lower during ASV. No pt suffered any adverse events from derangements in ventilation or acid-base balance. One pt with ARDS receiving 17 cmH 2 O PEEP was hypoxemic during ASV (PaO 2 57.2). Table 3 reveals heart rate and mean arterial pressure during each study period. There were no clinical changes to any measured vital sign between the two study periods. D Di is sc cu us ss si io on n/ /c co on nc cl lu us si io on n: : Upon initiation of mechanical ventilation, the precise V E requirement of the pt may not be known. Clini-cians use rough estimates and clinical experience to determine V E , respiratory rate, V T , and T i . Determination of vent settings made by the machine have been suggested (Intern Care Med 1996, 2 22 2:199). Our results suggest that ASV as startup mode of ventilation is acceptable and comparable to physician determined ventilator settings. Gas exchange during ASV is equivalent to physician determined ventilation. V T during ASV is more consistent with 'lung protective' strategy (7.8 ml/kg) than was conventional V T (9.7 ml/kg). Mechanical ventilation with ASV is more efficient as evidenced by lower V D /V T and may be safer as a result of lower V T and PIP.  I In nt tr ro od du uc ct ti io on n: : Excess body weight increases the risk of death from any cause and from cardiovascular disease in adults [1]. In the majority of population studies, the relationship of BMI to mortality is a U-shaped curve, with increased risk in the lowest and highest percentiles of the distribution. In acutely ill patients however BMI below the 15th percentile remains an independent predictor of mortality whereas a high BMI (>85th percentile) was not significantly related to risk of mortality [2]. We wanted to study in a prospective clinical trial the relationship between IAP and lactate and BMI and their relationship to subsequent mortality in ICU patients. The results of an interim analysis are presented in this abstract. (8.6 ± 4.9) in medical versus 6.1% (6.9 ± 3.5) in scheduled surgical patients. The ICU and hospital mortality were respectively 18% and 27.2%. The IAP was significantly higher in patients who died in the ICU: 13.2 ± 5.2 versus 7 ± 3.6 (P < 0.0001) as well as in patients who died in the hospital: 11.5 ± 5.3 versus 6.9 ± 3.6 (P < 0.0001). The divided into the chest wall and the lung. Since the diaphragm is coupled to the abdominal wall any increase in IAP may therefore affect chest wall and lung compliance [1]. By calculation of static V-P curves it has been shown in animal and human studies that abdominal and subsequently chest wall compliance goes up after abdominal decompression and this correlates well with the volume recruited [1]. Recent studies looking at compliance in primary and secondary ARDS found that the latter presents with preserved lung but decreased chest wall compliance and PEEP allows to recruit lung units markedly [1,2]. In a previous study we found that in patients with secondary ARDS and raised IAP, PEEP-adjustment for IAP calculated at zero PEEP (ZEEP) resulted in significant better oxygenation at the expense of a significant increase in peak and plateau alveolar pressures but without the risk for early barotrauma [3]. In this pilot study we wanted to sort out if there is a correlation between IAP and Pflex. D Di is sc cu us ss si io on n: : It would thus appear that in pigs a dose of 5 ml/kg PFC enables satisfactory ventilation parameters to be restored, but the level remains lower than that obtained with 10 ml/kg. The fact that a low dose does not have long-lasting effects might be due to faster evaporation of the product and a drop in its efficiency. There is no beneficial effect with larger dose than 10 ml/kg.  T0  T1  T2  T3  T4  T5  T6  T7  T8   T I   O Ob bj je ec ct ti iv ve e: : To evaluate the effects of inhaled nitric oxide on gas exchange and hemodynamic data during acute hypercapnia with uncorrected and corrected blood-pH.
D De es si ig gn n: : Prospective, randomized, experimental study.
S Se et tt ti in ng g: : University research laboratory.
S Su ub bj je ec ct ts s: : Ten piglets weighing 9 to 13 kg.
I In nt te er rv ve en nt ti io on ns s: : After induction of anesthesia, tracheostomy and controlled mechanical ventilation animals were instrumented with two central venous catheters, a pulmonary artery and two arterial catheters, and an ultrasonic flow probe around the pulmonary artery. Acute respiratory failure was induced by the infusion of oleic acid (0.08 ml/kg) and repeated lung lavages with 0.9% NaCl (20 ml/kg M Me ea as su ur re em me en nt ts s a an nd d r re es su ul lt ts s: : Continuous hemodynamic monitoring included right atrial, mean pulmonary artery and mean systemic arterial pressures, and continuous flow recording at the pulmonary artery. In addition, airway pressures, tidal volumes, lung compliance and airway resistance, arterial and mixed venous blood gases were measured. Data were obtained with and without inhalation of nitric oxide at baseline, normocapnia and 2 levels of hypercapnia with and without pH correction and are given in the Table. C Co on nc cl lu us si io on n: : Acute hypercapnia resulted in a significant increase in pulmonary artery pressure and pulmonary vascular resistance without significantly influencing oxygenation and cardiac output. pH-correction at hypercapnic episodes decreased pulmonary artery pressure and pulmonary vascular resistance associated with a slight increase in cardiac output and oxygenation. Inhaled nitric oxide significantly reduced pulmonary hypertension induced by acute hypercapnia and significantly improved oxygenation during normocapnia and acute hypercapnia with and without acidosis. S Su ub bj je ec ct ts s: : Nine piglets weighing 10 to 14 kg.

P43 Cardiorespiratory effects of inhaled nitric oxide and moderate hypercapnia in an experimental model of single ventricle
I In nt te er rv ve en nt ti io on ns s: : After induction of anesthesia, tracheostomy and controlled mechanical ventilation animals were instrumented with two central venous catheters and two arterial catheters. After a midline sternotomy the animals were placed on cardiopulmonary bypass and subjected to atrial septectomy, patch closure of the tricuspid valve, and creation of a 4 mm systemic to pulmonary arterial shunt. Before weaning from cardiopulmonary bypass ultrasonic flow probes were placed around the pulmonary artery and the descending aorta. In addition a pulmonary artery catheter was inserted into the pulmonary artery via the right ventricle. The protocol consisted of randomly assigned periods with different PaCO 2 levels (Normocapnia = PaCO 2 40 torr, Hypercapnia I = PaCO 2 50 torr, Hypercapnia II = PaCO 2 60 torr,) and a period of inhaling nitric oxide (10 ppm) at normocapnia. Tidal volume was reduced to induce hypercapnia, inspiratory time and PEEP were adjusted to achieve constant mean airway pressures (Paw).
M Me ea as su ur re em me en nt ts s a an nd d r re es su ul lt ts s: : Continuous hemodynamic monitoring included right atrial, mean pulmonary artery and mean systemic arterial pressures, and continuous flow recordings at the pulmonary artery and the descending aorta. In addition, arterial and central venous blood gases were measured. Data were obtained at baseline, normocapnia with and without NO-inhalation and 2 levels of hypercapnia and are given in the Table. C Co on nc cl lu us si io on n: : The creation of this experimental model of single ventricle resulted in a significant decrease in oxygen saturations and mean arterial pressure. Moderate hypercapnia resulted only in minimal changes in pulmonary artery pressure, pulmonary vascular resistance, and oxygen saturations. Inhaled nitric oxide decreased pulmonary artery pressure and resistance associated a slight increase in oxygen saturations.  Figure). There was no correlation (neither linear nor exponential) between SPA/DPPC ratio and lung mechanics (compliance: r = 0.1, resistance r = 0.24).

BL
C Co on nc cl lu us si io on n: : Surfactant containing higher ratios of saturated phosphatidylcholine has a role in maintaining compliance and small airway patency in RSV infected infants. Although a certain minimal level of SPA is necessary for surfactant function, additional benefit is not seen with increasing SPA/DPPC ratios. These findings have implications for exogenous surfactant supplementation in this disease.  Twenty-one hybird dogs were randomly divided into acute lung injury (ALI) treated and control groups at simulated high altitude environment of 4000 m. All animals were sacrificed after 6 h. The tetramethylpyrazine (TMP) treated group showed that the WBC count was significantly higher at 15 min after fat tissue extract was given (P < 0.05), the edema of both capillary endothelial cells and alveolar epithelial cells was less serious, the number of leukocytes accumulated in the lungs was less. The increase in production of leukotriene B (LTB) by polymorphonuclear neutrophil (PMN), alveolar macrophage (AM), and the activity of platelet-activating factor (PAF) by PMN were partially inhibited (P < 0.01). Although the ratio of the pulmonary extravascular water volume/blood free dry lung weight (PEWV/BFDL) of TMP group was significantly elevated as compared with that of control group, no significant difference was seen (P > 0.05). However, the level of PaO 2 , the PMN and AM count, and the albumin level of the BALF in both groups had no significant difference. These results demonstrated that the early treatment with TMP could inhibit the decrease in the WBC count, and reduce the accumulation of leukocytes in the lungs. I In nt tr ro od du uc ct ti io on n: : The acute respiratory distress syndrome (ARDS) is characterized by diffuse injury to the endothelial and epithelial surfaces of the lung leading to severe respiratory failure. Alterations in the surfactant system and activation of the contact system of coagulation are major contributors to the pathophysiology of ARDS. C1-inhibitor (C1-INH) is the main inhibitor of contact activation and the only known inhibitor of classical pathway complement activation. The aim of this study was to investigate the effects of C1-inhibitor administration and rSP-C surfactant application on oxygenation and lung histology in an ARDS-model. M Me et th ho od ds s: : Thirty-six male Sprague Dawley rats were subjected to repetitive lung lavage with isotonic saline solution. Three experimental groups and two control groups were studied: Group 1 and 2 served as controls without any treatment. Animals of group 1 were sacrificed 60 min after the last lavage procedure (p.l.). Animals of group 3-5 received 200 U/kg body weight (b.w.) C1-INH (Centeon, Germany) intravenously (group 3), 25 mg/kg b.w. rSP-C Surfactant (Byk Gulden, Germany) intratracheally (group 4) or both (group 5) at 60 min p.l. Blood gases were determined 120, 150, 180 and 210 min p.l. All animals of group 2-5 were sacrificed at 210 min p.l. and the lungs were excised for histological examination. Hyaline membrane formation, distribution and severity of intraalveolar neutrophil (PMN) accumulation and the severity of intraalveolar and perivascular hemorrhage were graded semiquantitatively using a scale from 0 to 4+. R Re es su ul lt ts s: : At 210 min p.l. pO 2 values of group 4 (456 ± 74 mmHg) and group 5 (387 ± 155 mmHg) were significantly higher than in group 3 (120 ± 103 mmHg) or in controls (63 ± 12 mmHg). Hyaline membrane formation was significantly reduced in group 4 and 5. The grading for PMN infiltration was significantly lower in animals who received C1-INH (group 3 = 2.0, group 5 = 2.3) than in controls (group 2 = 2.7) or in animals treated with surfactant only (group 4 = 3.3). The severity of intraalveolar hemorrhage and edema were significantly reduced in group 3 and highest in group 4. C Co on nc cl lu us si io on n: : Surfactant application was effective in improving pO 2 which was related to the reduction of hyaline membrane formation. C1-INH administration had no significant effect on pO 2 and hyaline membrane formation but was effective in reducing PMN infiltration, intraalveolar hemorrhage and edema formation. HLf added to LMVEC at the time of stimulation did not influence chemokine production. However, when hLf was added prior to LPS stimulation, a significant inhibition of MCP-1 (P < 0.001) and ENA-78 (P < 0.01) but not of Groα production was observed.

P48 Inhibition of pulmonary microvascular chemokine production by human Laktoferrin and Phosphatidylethanolamine
In order to investigate if LPS induced chemokine production was dependent on PAF, Arachidonic acid (AA) or its metabolites, LMVEC were treated with PE, CoI and PAF-antagonist either M Me et th ho od ds s: : After institutional approval 40 critically ill patients were prospectively studied during ventilatory support and weaning, three patients due to death were excluded. All patients were weaned according to standard weaning protocol. Blood samples were drawn daily and collected until analysis. Malondialdehyde (MDA) serum levels, total glutathion (GSH), glutathion-peroxidase (GPX) and superoxid-dismutase (SOD) activity in erythrocytes and serum selenium levels were estimated at the time of admission to ICU (T1), on the last day of full ventilatory support (T2), on the day when weaning was started (T3) and on the first day of spontaneous ventilation (T4). After successful weaning patients were divided in two groups according to the length of weaning (W): group S (W ≤ 3 days, n = 15), group L (W > 3 days, n = 22). t-test or Mann Whitney Rank Sum test were used for statistical analysis (SigmaStat, Jandel Co., USA), values are expressed as mean (SD) or median (25%-75% percentiles), P < 0.05 was considered statistically significant. D Di is sc cu us ss si io on n: : Prolonged ventilatory support and weaning longer than 3 days were associated with higher MDA levels and lower GPX levels, also selenium levels were insignificantly lower in patients with prolonged ventilatory support. The clinical importance of these findings needs to be further studied. P Pa at ti ie en nt ts s: : All patients of ≥14 years admitted between January 1995 and January 1996 with a total body surface burn area of ≥20%. Exclusion criteria included immunosuppression, pregnancy, and length of stay less than 5 days or admission ≥48 h following burn trauma.
I In nt te er rv ve en nt ti io on n: : Collection of data on surveillance samples from throat and rectum on admission and afterwards twice weekly, and primary endogenous pneumonia during the intensive care unit stay. In the univariate and multivariate analysis the factors associated (P < 0.05) with mortality were primary endogenous pneumonia and full-thickness burn area. The risk factors associated (P < 0.05) with primary endogenous pneumonia were full-thickness burn area and inhalation injury. Increasing the number of cases (56 patients), both variables were statistically significant in the univariate analysis, but were not statistically significant in the multivariate analysis. At present we are continuing the study to know the factors associated with morbidity and mortality in severe burn patients.
C Co on nc cl lu us si io on n: : Half the patients developed a pneumonia (63% primary endogenous pneumonia). The isolated pathogens were predominantly Staphylococcus aureus. Primary endogenous pneumonia in severe burn patients may be associated with mortality, but is necessary collecting more cases to show it. The Gram-negative bacillus Stenotrophomonas maltophilia (SM) has emerged as an important pathogen associated with significant case/fatality ratio [1,2]. SM is a potentially dangerous organism because of its resistance to many antibiotics. We present here an outbreak of mucoid phenotype SM pneumonia (four cases) and respiratory tract colonisation (three cases). Our review of literature revealed only one case report of pneumonia characterised as mucoid phenotype [2]. The outbreak was caused following admission of a 65-year-old male patient with respiratory distress, fever, leukocytosis (24000/µl) in the ICU. Chest X-ray showed an infiltrative shadow in the right lower lobe and bilateral pleural effusion was detected on CT. Sputum cultures obtained before admission to ICU and subsequent days yielded mucoid phenotype SM. Treatment with ticarcilline plus clavulonic acid to which the isolates was susceptible was initiated. One-day later chest X-ray showed diffuse bilateral pneumonic infiltrates and the patient's condition rapidly deteriorated. Ciprofloxacine was added to the treatment. Subsequent SM isolates rapidly developed antimicrobial resistance to antibiotics. Four patients in the ICU were lost with SM pneumonia within 7-10 days. SM isolates were identified by standard Analytical Profile Index procedure (API 20E and API 20 NE). A significant number of both infected and colonised patients had severe systemic diseases and tracheotomy, they were mechanically ventilated and receiving broad spectrum antibiotics before isolation of SM. SM is emerging as an important nosocomial pathogen in critically ill ICU patients and should no longer be regarded as a harmless bacillus in ICU.  O Ob bj je ec ct ti iv ve es s: : Bacterial resistance to antimicrobial treatment is actually one of the most debated issue in medical field. Therefore, it is important to dispose a diagnostic procedure to allow an aimed antimicrobial treatment. Unprotected tracheal aspirate (UTA) is the most widely used sampling technique to assess pulmonary infection even though known to have a high sensivity and a low specificity [1]. Protected specimen brush bronchoscopically directed (PSB) is a procedure that purpose a higher specificity [2]. The aim of the study is to compare both methods for the diagnosis of pneumonia in ICU.

P51 Outbreak of nosocomial infection/colonisation caused by
M Ma at te er ri ia al ls s a an nd d m me et th ho od ds s: : 70 consecutive patients admitted to an 8bed general ICU over a period of 18 months, intubated and mechanically ventilated [3]. When patients met clinical and radiological criteria for suspicion of pneumonia both UTA and PSB were performed. R Re es su ul lt ts s: : Data collected from the two methods pointed out significative differences.
PSB vs UTA revealed complete negativeness or growth of different microorganisms in 29 patients (41.5%); this result had statistically significance (P < 0.01).
Above all it is to underline that microorganism most frequently represented in UTA and not detected by PSB were in sequence: Candida spp, Pseudomonas aeruginosas, Staphylococcus aureus. No complications were reported during the procedures.
Antimicrobial therapy based on PSB data was started, leading to a good clinical response and favourable outcome. C Co on nc cl lu us si io on n: : PSB is a reliable and safe method useful to investigate pulmonary infections. High specificity of the technique allows to aim antibiotic therapy, so reducing the risk of inducing resistance to molecule still effective with a consequent optimization of expenses. UTA and corrispective PSB were both negative in X case (y%) VPN = 1. In consideration of UTA elevated vpn, this method could represent a first diagnostic step followed in case of positiveness by PSB   I In nt tr ro od du uc ct ti io on n: : In last years the nosocomial ICU fungal infections are assuming a greater impact with increasing morbidity, mortality and cost. A better outcome is correlated with an earlier treatment.

P60 Biochemical and haematologic predictors of fungemia in previous colonised ICU patients
O Ob bj je ec ct ti iv ve es s: : The aim of the study was to assess the accuracy of simple not expensive biochemical and haematologic parameters to predict the change of fungal colonisation to a status of fungemia. Age, SAPSII in first 24 h, alkaline phosphatase, lactate dehydrogenase, blood lactate, pH, bicarbonate, leukocytes, and platelets were evaluated in all patients the 3 days preceding fungal documentation. We compared biochemical and haematologic results in group 1 and group 2 trying to identify a different profile and evaluated the predictive value of the different parameters. Results are presented as media and standard deviation. We applied t student test comparing the two groups and we considered a P < 0.05 to be significant.
R Re es su ul lt ts s: : (See Table). We found significant statistical difference with bicarbonate, and platelet count when comparing the two groups. We also observed an increasing blood level on alkaline phosphate in fungemic patients. C Co on nc cl lu us si io on n: : In our study, variations on bicarbonate, platelet count and alkaline phosphatase are predictive of fungemia in previous colonised patients. However, further observations, analyses, are needed and perhaps involving larger patient numbers to evaluate the clinical utility of these findings.  Table. C Co on nc cl lu us si io on n: : There were no significant differences in incidence of candida colonization and proportion of albicans v. non albicans species between both groups. Hospital and ICU stay and length of ventilatory support were nonsignificantly longer in group F. Clinical usefulness of early fluconazole prophylaxis needs to be further evaluated.

P61 Fluconazole prophylaxis of systemic candida infection in non
Group F Group C APACHE II 23.6 (3.8) 22.6 (4.7) ICU stay (days) 17.6 (8.9) 11.5 (4.1) Hospital stay (days) 21 (9) 15.7 (6.2) Ventilatory days 17.6 (11.9) 11.9 (4.9) Candida colonization 27% 33% and 37 pts fluconazole. The overall crude mortality was 54%. The number of positive fungal blood cultures (1 culture versus >1 culture) did not influence crude mortality (54% versus 63%). In addition, the mortality of pts infected by C. albicans (55%) was similar to that of pts infected by C. non-albicans species (45%). The mortality of pts located in ICUs (79%) was significantly higher (P < 0.001) than that of pts in surgical (29%) and medical wards (33%). Finally, the mortality of pts who did not receive adequate antifungal therapy (70%) was significantly higher (P < 0.001) than that of treated pts (30%). In conclusion, the incidence of nosocomial candidemia was high during the primary phase of the study; most of our pts with candidemia had severe underlying diseases and were hospitalized in ICUs. The number of fungal positive blood cultures did not influence the crude mortality, confirming that a single bood culture shuld not be dismissed as benign transient candidemia. On the contrary, about two-third of our pts did not receive an adequate antifungal treatment and the majority of them died. Herpes Simplex Virus (HSV) infection may cause different disorders in patients hospitalized in intensive care. Bronchoalveolar lavage (BAL) is a procedure performed almost as a routine in patients with unexplained respiratory insufficiency in our department. During the last 10 years, HSV has been isolated frequently from the respiratory tract at our 30 beds intensive care unit. The objectives of this retrospective study were to define risk factors of the population in whom HSV virus was isolated.
The study concerned patients with an isolation of HSV from either bronchial aspiration (BA) or BAL in the past 5 years (1992)(1993)(1994)(1995)(1996)(1997). HSV was isolated by culture on shell vials and identified by immunofluorescence after staining with monoclonal antibodies or by the conventional culture and cytopathogenic effect on Vero-cells. From the 64 cases observed, 47 HSV isolations originated from BA, 13 from BAL (of which 9 with simultaneously negative BA) and 4 from both BA and BAL. The mean age of the patients was 62 years (range from 16 to 82). Only 50% of the patients had fever at the time of the investigation. The majority of the patients (94.9%) was intubated before the isolation. The role of immunosuppression, previously recognized as a risk factor for herpes infection, was not confirmed in this study: only 20.4% had received either corticosteroids or immunosuppressive agents. Striking is that 73.4% had undergone a surgical procedure before the isolation, mainly coronary bypass grafting or other thoracic operations. Daily chest X-rays from 2 days before till 2 days after virus isolation were reviewed blindly by the same radiologist. There was no pathognomonic image at the chest X-ray: a localized infiltrate resembling pneumonia, diffuse alveolar infiltrates or an interstitial pattern were observed and 14% of the chest X-rays were even defined as normal. Lung injury was severe: almost 60% had a PaO 2 /FiO 2 less than 200. 28 patients received aciclovir therapy once herpes was isolated, without an effect on the outcome: 48.4% of all patients and 42.8% of those receiving aciclovir therapy (28) died.
Isolation of HSV in respiratory samples from critically ill patients is therefore more frequent than previously known. Whether these isolates contribute to illness and its evolution remains to be determined. S St tu ud di ie ed d g gr ro ou up p c ch ha ar ra ac ct te er ri is st ti ic cs s: : Unlike previous studies we studied now the group of patients with IMD treated in the whole West Bohemian region. Six patients from total number of 88 patients were excluded from the study due to incomplete data (n = 82). R Re es su ul lt ts s: : ( (1 1) ) Logistic regression model has validity coefficient = 0.469: P(Death=1)= 1/(1+EXP(Σβ 0 + β i X i )) ( (2 2) ) Receiver operating characteristic for logistic model -ROC report: (see Table overleaf).

P64 Preliminary validation of new prognostic scoring system in patients with invasive meningococcal disease
Set up of optimal value of P(Death=1) by logistic regression model was based on minimal number of the false death predictions (C) along with maximum correct death predictions (A), (D=survival prediction, B=false death prediction). We looked up in ROC report the value of P(Death=1) with maximum ratio between sensitivity and false positivity. B Ba ac ck kg gr ro ou un nd d: : A principle feature of Gram negative sepsis is the rapid onset of profound shock. The failure of anti-endotoxin antibodies to produce significant improvement in outcome [1] and the profound hypocalcaemia we have observed in meningococcal sepsis led us to re-evaluate the possible aetiologies of shock in Gram negative infection.
O Ob bj je ec ct ti iv ve e: : To test the hypothesis that Gram negative organisms directly or indirectly may be capable of proteolytic breakdown of albumin thus explaining in part the aetiology of shock and hypocalcaemia seen in severe Gram negative sepsis.
M Me et th ho od ds s: : Urine was collected from patients with severe meningococcal sepsis (11) and from controls including patients admitted to intensive care (2) and patients with known proteinuria (4). The urine was dialysed and subjected to polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting with a sheep antihuman albumin antibody.
Albumin was incubated with lipopolysaccharide (LPS) derived from various Gram negative organisms and the incubates were subjected to SDS-PAGE to ascertain the presence of albumin degradation products. Albumin was also incubated with homogenates of cultured Neisseria meningitidis and again the incubates were subjected to SDS-PAGE.
R Re es su ul lt ts s: : Multiple albumin fragments were detected in urine collected from patients with meningococcal sepsis. In vitro incubation of human albumin with crude LPS derived from gram negative organisms and subsequent SDS-PAGE also showed cleavage of albumin into multiple fragments. Similar in vitro studies with homogenates of N. meningitidis failed to show evidence of breakdown. No albumin cleavage products were detected in the urine of control patients. C Co on nc cl lu us si io on n: : This study suggests that in meningococcal sepsis there is release into the circulation of protease(s) which cleave albumin. We were not able to distinguish whether the protease action was of exogenous or endogenous origin. This may have profound significance for the treatment of meningococcal sepsis. This prospective crossover study compares the pharmacokinetics of meropenem administered by continuous infusion with intermittent administration in critically ill patients. Fifteen patients were randomized to receive meropenem either as a 2 g iv loading dose followed by a 3 g continuous infusion (CI) over 24 h or as intermit-tent administration (IA) of 2 g iv every 8 h (q8h). Each regimen was performed over a period of 2 days followed by a cross over to the alternative regimen for the same time. Pharmacokinetic parameters (mean ± SD) of CI included following: concentration at steady state (C SS ) was 11.9 ± 5.0 mg/l, area under the curve (AUC) was 117.5 ± 12.9 mg/l/h. Maximum and minimum serum concentration of meropenem (C max , C min ) and total meropenem clearance (Cl tot ) for IA were 110.1 ± 6.9 mg/l, 8.5 ± 1.0 mg/l and 9.4 ± 1.2 l/h, respectively. The AUC during IA regimen was larger than the AUC during CI (P < 0.001). In both treatment groups meropenem serum concentrations remained above the minimal inhibitory concentration for the most important bacterial strains all the time. We conclude that CI of meropenem is equivalent to the IA regimen and is therefore suitable for treating critically ill patients. Additionally, a CI regimen can save costs of antibiotic therapy as bactericidal serum levels can be achieved with only 50% of the amount of drug used for IA. Meropenem is a new carbapenem antibiotic with a broad spectrum of activity against Gram-positive and Gram-negative strains including β-lactamase producers. Thus, it is particularly useful in intensive care patients (pts) with septic complications due to unknown pathogens. The present study was conducted to evaluate the pharmacokinetic properties of meropenem in nine critically ill patients treated by continuous venovenous hemofiltration (CVVH).

Reference
All pts received one single dose of 1 g meropenem intravenously.
High-flux polysulfone membranes (Diafilter-30, Amicon, Ireland) were used as dialyzer. Meropenem serum concentrations as well as filtrate aliquots were determined by high-performance liquid chromatography.
Peak serum concentrations were 28.1 ± 2.7 µg/ml, trough levels 6.6 ± 1.5 µg/ml after 6 h CVVH. The post-to-pre hemodialysis ratio was 0.24 ± 0.06, total removal was 35. Often a new drug is licensed for use in the critically ill before its pharmacokinetic profile has been fully described. In particular information relating to the amount removed by continuous renal replacement therapies, such as haemofiltration, is sparse. This relates to the difficulties associated with patient recruitment and standardisation for in vivo studies. This study describes the removal of meropenem, a broad spectrum antibiotic, by an in vitro model of haemofiltration and compares the data with that obtained in a previous in vivo investigation [1]. The in vitro model incorporated a polyacrylonitrile membrane (Hospal, Multiflow 60) employing a blood pump (Hospal BSM22SC) to circulate carrier fluid (3.5% human albumin solution in Tyrode Ringer) around an extracorporeal circuit. Ultrafiltration rates were manipulated using a peristaltic pump attached to the ultrafiltration line. Pre-membrane, post-membrane and ultrafiltration samples were collected from the model at timed intervals, employing three different UFR rates. Meropenem concentrations were measured by HPLC and used to calculate the drugs' sieving coefficient (S) and filter clearance (FCL), using standard equations [2]. The results were then compared to values obtained from a previous in vivo study [1] employing a similar membrane (Hospal, Multiflow 100). A mean (± SD) S value of 0.99 ± 0.07 (n = 13) was calculated for the in vitro model for al UFR rates used which compared favourably with a mean ( ± SD) S of 0.95 ± 0.03 reported during the in vivo study, involving four patients. A significant linear correlation was seen between UFR and FCL for both in vitro and in vivo data (r = 0.98, P < 0.05 and r = 0.90, P < 0.05, respectively). The results of this study suggest that the in vitro model is capable of providing accurate meropenem filter clearance data. Although further validation of this model using a range of drugs is required, this preliminary work suggests that, in the absence of in vivo pharmacokinetic information, extracorporeal drug clearance determined using an in vitro model could be used to aid prescribing in patients receiving haemofiltration.   We observed that the administration of vancomycin (V) is sometimes accompanied by pruritus, a clinical sign which appears a few minutes after the beginning of V infusion and lasts soon after V administration has been stopped. Generally it is not associated with a cutaneous rash. The aim of this study was to evaluate the hemodynamic behaviour which follows the appearance of pruritus. significantly and no change was observed in MAP and HR at time 3 if compared with time 1. In patients of group B the hemodynamic data did not change significantly at the four times of the study (Table). No patient showed a cutaneous rash throughout the study.

References
D Di is sc cu us ss si io on n: : The analysis of our data points out that in patients who showed pruritus during the administration of vancomycin, SVRI went down. This vasodilating effect was offset by the increase in CI. As a result MSP was well maintained. Certainly this compensation was possible because the patients studied were normovolemic. But we would like to know what would happen if pruritus appears in patients with hypovolemia? Probably the compensatory mechanism would not be so effective and hypotension could occur. We conclude that pruritus which follows the administration of vancomycin can be considered an alarm-bell indicating a condition of peripheral vasodilatation, and must lead us to evaluate the patient in order to detect the hypovolemic state and to compensate for it before continuing the infusion of vancomycin. coli, Proteus and Bacteroides. There were no differences in haemodynamics or arterial blood gases between CLP or SH animals at different time points. However, there were significant differences in white blood cell count (WBC's), blood lactate and stopped capillary flow (CD stop ) density measurements between the groups (Table).

Reference
C Co on nc cl lu us si io on n: : In this acute model of sepsis remote organ damage occurs early and is equivalent to that seen at 24 h. Therapies aimed at MODS prevention need to be commenced at the first possible opportunity. O Ob bj je ec ct ti iv ve e: : We measured nitrite/nitrate (NOx) levels and transforming growth factor beta (TGF-β) levels in septic shock, and assessed these factors during the onset of shock.

References
P Pa at ti ie en nt ts s: Twenty-two patients with sepsis not complicated by shock and 23 patients with septic shock.
M Me ea as su ur re em me en nt ts s a an nd d m ma ai in n r re es su ul lt ts s: : NOx levels were significantly higher in the septic shock group than in the sepsis-alone group.
NOx levels were significantly higher in the group that died than in the group that survived. TGF-β levels were significantly higher in the sepsis-alone group than in the septic shock group. TGF-β levels were significantly higher in the group that survived than in the group that died. Twenty-one (80.8 %) of the 26 patients with NOx levels of 92.9 µmol/l or more (mean + standard deviation in the sepsis group without shock) had sepsis complicated by shock, as opposed to only 2 (10.5%) of the 19 patients with NOx levels below 92.9 µmol/l, and the rate of occurrence of shock as a complication of sepsis was significantly higher when the NOx level was 92.9 µmol/l or more. Two (12.5%) of the 16 patients with TGF-β levels of 19.3 ng/ml or more (mean + standard deviation in the septic shock group) had sepsis complicated by shock, versus 21 (72.4%) of the 29 patients with TGF-β levels below 19.3 ng/ml, and the rate of occurrence of shock as a complication was significantly higher among the patients with TGF-β levels below 19.3 ng/ml. There was a significant negative correlation between NOx levels and TGF-β levels.
C Co on nc cl lu us si io on n: : NO is involved in the pathogenesis of septic shock.
TGF-β appears to inhibit NO production, and may act to prevent septic shock.
P73 Albumin clearance in the endotoxemic rat after administration of N ω ω -nitro-L-arginine methyl ester (L-NAME) K  I In nt tr ro od du uc ct ti io on n: : Endotoxin (LPS) is a powerful activator of the inducible nitric oxide (NO) synthase. Whereas NO seems to be one factor behind the decreased responsiveness of the circulation to adrenergic stimulation in septic shock, the role of NO in increased vascular permeability is less clear. In a former study [1] we have shown that although NO production increased after LPS there was no increased extravasation of albumin in a wide variety of rat tissues examined; on the contrary clearance was decreased in the entire gastrointestinal tract. In this study tissue extravasation was examined after administration of the nitric oxide synthase inhibitor L-NAME.
M Me et th ho od ds s: : Anaesthetised Wistar rats were given E. coli lipopolysaccharide (LPS) 3 mg/kg i.v. and were studied for 5 h. Mean arterial pressure (MAP) and heart rate (HR) were recorded. As an indicator of NO production methemoglobin (metHb) was measured in the beginning and end of experiments. 2 h after LPS a bolus of L-NAME 100 mg/kg, or saline, was given i.v. The tissue clearance of albumin was studied over the last 2 h of the experiment by means of a double isotope method [2]. R Re es su ul lt ts s: : In response to LPS all rats had a drop in MAP. After administration of L-NAME (n = 7) MAP increased significantly as compared to controls (n = 8). MetHb increased during experiments in controls but not in NAME-treated rats. Tissue plasma clearance for albumin increased in the NAME-group in skin, skeletal muscle and heart and decreased in testes as compared to controls.
D Di is sc cu us ss si io on n: : We have shown an increased production of NO after LPS and the dose of L-NAME administered abolished this. No differences in gastrointestinal albumin clearance were detected between groups, however in heart, skeletal muscle and skin albumin extravasation was increased. We conclude that this is most likely due to changes in regional hemodynamics with locally increased capillary pressures leading to increased albumin filtration in certain tissues only. In the majority of tissues no differences were found. I In nt tr ro od du uc ct ti io on n: : 'Rigid' red cells in sepsis are thought to play a role in multiorgan failure by plugging the microvasculature and compromising oxygen delivery. During sepsis endogenous nitric oxide (NO) production is increased. What effect this has on erythrocyte deformability (RBCd) is unclear. We report the effects on RBCd and capillary blood flow when NO overproduction was prevented in septic rats.

P74 Loss of erythrocyte deformability during systemic sepsis is prevented by nitric oxide synthase inhibition
M Me et th ho od ds s: : Acute sepsis was induced in Sprague-Dawley rats via cecal ligation and perforation (CLP). At 2 h post CLP, aminoguanidine (AG), a selective inducible nitric oxide synthase inhibitor was infused (i.v. 60 mg/kg/h) to maintain baseline NOx levels. Capillary blood flow in the EDL skeletal muscle was filmed using intravital video microscopy. Plasma NOx (NO 2 -/NO 3 -) levels were measured by chemiluminescence and deformability was assessed by membrane displacement, using the micropipette aspiration technique. R Re es su ul lt ts s: : At 6 h, an increase in plasma NOx of 260% ± 46 SEM in CLP animals was associated with a 12% ± 1.6 SEM loss in red cell deformability and a twofold increase in stopped flow capillaries (P < 0.05, relative to Sham). Infusion of AG prevented the increase in NOx, the loss of deformability and the increase in stopped capillary flow, (P < 0.05). In sham rats, AG augmented RBCd (P < 0.05), but had no effect on stopped flow. C Co on nc cl lu us si io on n: : Eliminating nitric oxide overproduction in septic rats was associated with preventing 1) the loss of red cell deformability and 2) the increase in stopped capillary blood flow, resulting in the maintenance of the microvascular circulation. I In nt tr ro od du uc ct ti io on n: : The bulk of data that links inducible nitric oxide synthase (iNOS) activity to the pathophysiology of sepsis originates in animal studies. However, the role of iNOS in human sepsis is controversial. Therefore, we measured in this pilot study iNOS activity in inflammatory cells from septic ICU patients compared to normal controls. M Me et th ho od ds s: : Blood samples from 5 ICU patients with clinically and bacteriologically documented sepsis, and from four healthy volunteers were centrifuged to separate the plasma/buffy coat. The buffy coat was layered onto Histopaque 1077 and centrifuged at 400 g to finally isolate white blood cells (WBCs). Constitutive (cNOS) and iNOS activities were analyzed in WBCs by the [ 3 H] L-arginine-L-citrulline assay and measured in Units (pmol L-cit-rulline evolved/min/mg protein). The metabolic end-products of nitric oxide (nitrite/nitrate; NO x -) were also determined in plasma from these subjects by chemiluminescence. R Re es su ul lt ts s: : Plasma NO x levels were elevated in septic compared to control subjects (208 ± 107 vs 26 ± 7 µmol/l, respectively).WBCs from septic patients exhibited low cNOS activities (0.1 ± 0.1 vs 1.0 ± 0.6 Units for controls). iNOS activity from the septic WBCs was elevated, compared to controls (3.1 ± 1.8 vs 0.5 ± 0.3 Units, respectively). C Co on nc cl lu us si io on n: : This pilot data suggests, that consistent with the plasma accumulation of nitric oxide metabolites, inflammatory cells of septic humans produce high levels of iNOS compared to healthy controls while cNOS production is suppressed. These findings support the theory that iNOS has an important role in the pathogenesis of human sepsis. Phospholipase A 2 (PLA 2 ) regulates eicosanoids and platelet activating factor production and plays an important role in regulating critical mediators in inflammatory diseases. PLA 2 activity is significantly enhanced during sepsis and multiple organ failure and therefore offers an intriguing target in developing anti-inflammatory drugs. We have identified several kinds of biflavonoids with inhibition of PLA 2 activity, which are isolated from plant sources, as potential putative anti-inflammatory and anti-septic agents. Two of them (bilobetin and ginkgetin) potently inhibit several kinds of type II 14 kDa PLA 2 but exhibits a weaker inhibition of type I 14 kDa PLA 2 using 2-linol-[1-14 C]PE as substrate. These inhibitors have been tested for their ability to inhibit the production of TNF-α and the formation of two enzymes, inducible NO synthase (iNOS) and inducible cyclooxygenase (COX-2) using LPS-stimulated Raw264.7 macrophages as assay systems. In the Raw264.7 cells, bacterial LPS induced the protein of COX-2 and iNOS as well as TNF-α release. The inhibitors consistently inhibited the production of TNF-α in a dose-dependent manner. The inhibitory effect of TNF-α was observed at concentrations similar to those related by PLA 2 . Moreover, treatment of cells with bilobetin and ginkgetin inhibited nitrite production, one of the stable end products of NO production measured in culture supernatants. The inhibition of NO products is caused by decreased iNOS protein levels as assessed by immunoblotting using a specific anti-iNOS antibody. The inhibitors treatment also reduce the expression of COX-2 protein level to about 80% in LPS-stimulated cells, which coincided with reduction of the iNOS protein. These results suggest that inhibition of PLA 2 and subsequent metabolism of arachidonic acid by COX-2 contribute to LPS-induced NOS pathway including TNF-α in Raw264.7 cells and these two inhibitors may develop as useful agents for anti-inflammation. A Ai im m: : The present pilot study was carried out to test the hypothesis that fusidin downregulates the production of pro-and antiinflammatory cytokines in septic patients. The study was approved by the Regional Ethical Committee and informed consent was obtained from each patient or a close relative.

P77 Fusidin down-regulates the production of IL-6 in septic patients: a pilot study
M Ma at te er ri ia al l a an nd d m me et th ho od ds s: : Five consecutive septic patients received fusidin 500 mg × 3 i.v. for 1 day. Blood samples were drawn two times before fusidin administration, six times during the 24 h where fusidin was given and 24 h after the last dose. The pro-inflammatory cytokines IL-1a, IL-1b, TNFα and IL-6 and the antiinflammatory cytokines IL-10 were analysed using ELISA.
R Re es su ul lt ts s: : Three females and two males were included. age 21-72 years (range). APACHE II score 13-24 (range). Two patients died in the ICU. No clinical or biochemical side effects were seen in relation to fusidin administration.
The proinflammatory cytokines IL-1b, TNFα and IL-6 and the antiinflammatory cytokine IL-10 were detectable in peripheral blood in al patients while IL-1a was undetectable. Treatment with fusidin was associated with a decline in plasma concentrations of IL-6 from 183 (78-293) to 116 (67-406) pg/ml 12 h later (median values with range) (P < 0.05). No changes occurred in the other cytokine levels. The measured cytokines were characterized by large interindividual variations.
D Di is sc cu us ss si io on n: : The results from this pilot study provide further in vivo evidence for the antiinflammatory properties of fusidin. Fusidin may be useful in the management of the systemic inflammatory response in septic patients. Human umbilical vein endothelial cells (HUVECs) were incubated with interferon-γ (IFNγ) interleukin β (IL-β), and LPS, or their different combinations for 2-48 h. TNF-α was measured by time-resolved immunofluorometric assay. Unstimulated HUVECs did not produce detectable amounts of TNF-α but IFNγ IL-β and LPS in combination induced TNF-α production in a timedependent manner. Immunofluorescent staining confirmed that the TNF-α was synthesized by endothelial cells. IFNγ IL-β or LPS alone did not induce TNF-α production, whereas IFNγ and IL-β in combination induced TNF-α production, which was further increased with LPS. TNF-α messenger-RNA expression was detected with RT-PCR in stimulated, but not in unstimulated HUVECs.

P78 Human vascular endothelial cells produce TNF-α α after stimulation with proinflammatory cytokines
Human vascular endothelial cells are capable of producing TNFα after proinflammatory cytokine stimulation, and may therefore contribute to the increased amount of TNF-α found in states like cachexia and septic shock. I In nt tr ro od du uc ct ti io on n: : Poor muscle functions play a pivotal role in developing ventilator dependency after long term ventilatory support and studies have shown that sepsis may be associated with decreased muscle contractility. The aim of the study was to evaluate plasma levels of TNF alpha, IL-8 and sIL-2R during ventilatory support and weaning.

P79 Cytokine plasma levels during weaning in critically ill patients with sepsis
M Me et th ho od ds s: : After institutional approval 40 critically ill patients were prospectively studied during ventilatory support and weaning, three patients due to death were excluded. All patients were weaned according to standard weaning protocol. Blood samples were drawn daily and collected until analysis. Apache II score, organ failure score (Goris), sepis organ failure assessment score (SOFA), ventilatory days and 'weaning' days were recorded. After successful weaning patients were divided in two groups according to the length of weaning (W): group S (W ≤3 days, n = 15), group L (W >3 days, n = 22). TNFα, IL-8 and sIL-2R serum levels were selected and measured at the time of admission (T1), on the last day of full ventilatory support (T2), on the day when weaning was started (T3) and on the first day of spontaneous ventilation (T4).
Values are expressed as a mean ± SD (or median and 25 th and 75 th percentiles), t-test or Mann Whitney Rank Sum test were used for statistical analysis (SigmaStat, Jandel Co., USA), P < 0.05 was considered statistically significant.
R Re es su ul lt ts s: : Total ventilatory and weaning days were 9.6 ± 4.8 resp. 1.7 ± 0.7 in group S and 24.6 ± 11.3 resp. 9.0 ± 3.7 in group L. Selected results (TNFα and IL-8 in pg/ml) are presented in the Table:   T1  T2  T3  T4 Group S -TNFα 11.6(5.6) 7.72 (7.7 R Re es su ul lt ts s: : Amongst the 24 patients studied, 16 presented IL-6 >300 pg/ml. High concentrations of IL-8 and TNFα were also observed, but these were not uniformly coincident with the former. Two of these patients survived, being those in whom we were able to interfere with the cytokine profile. The 7 patients with SIRS, presented relatively low concentrations of cytokines, having one of them died. TNFα and IL-8, sometimes in very high concentrations, do not correlate with any particular organic dysfunction. IL-1β and TGF-1β always presented low values, close to the detection limits, in all of the patients. D Di is sc cu us ss si io on n: The high concentrations of IL-6 (the 'black smoke'?), revealed a homogenous correlation with the clinical severity, thus making it a useful diagnostic and prognostic serum marker. The rapid knowledge of the cytokine profile is important for intervention in the mechanisms, which lead to multiple organ failure. The relation of some cytokines with particular organ dysfunction, such as ARDS and cardiac failure, as well as the influence of presently known and future anti-cytokine strategies, remains to be evaluated. IL-6 levels increased from t1 to t2 (P < 0.01), whereas C1-INH functional levels declined tendencially and antigenic levels dropped (P = 0.024). Levels of C1-INH at t3 returned to preoperative values and IL-6 declined.
C Co on nc cl lu us si io on n: : As expected (postoperative agression syndrome) IL-6 increased significantly. Surprisingly, plasma levels of the antiinflammatory acute phase protein C1-INH remained normal or even declined. On the first postoperative day C1-INH and IL-6 levels tended to return to preoperative values. This was associated with uncomplicated clinical course. We suggest, that this short period of disproportion between pro-and anti-inflammatory mediators may increase the 'second hit' risk, if it is longer lasting. Crit Care 1999, 3 3 ( (s su up pp pl l 1 1) ):P82 B Ba ac ck kg gr ro ou un nd d: : Recent data from our laboratory demonstrated a significant decrease in the number of circulating T helper cells and monocytes as well as in the expression of IL-2 receptors on T cells during induced whole body hyperthermia [1,2]. The aim of further investigations was to analyze the influence of these effects on the function of TH1 cells, TH2 cells and monocytes. Therefore, we measured the levels of IL-2, IL-4, IL-6 and IFN-γ in the blood of cancer-patients, undergoing whole body hyperthermia of 42°C. This is used as part of so called 'systemische Krebs-Mehrschritt-Therapie' (sKMT) in our clinic.
M Me et th ho od ds s: : Cytokine levels of 9 patients were measured by an ELISA technique. Blood samples were obtained before beginning of therapy at 37°C, at 40°C, at the end of the plateau of 42°C, at 37°C again, as well as after 20 h (on the next morning). Time between the first four investigations was about 2 h. Cytokine levels were compared by using a Wilcoxon rank sum test.
R Re es su ul lt ts s: : We found a reversible, significant decrease (P = 0.017) of IL-2 at 42°C (Fig. 1), whereas the levels of IL-4 and IFN-γ decreased slightly (data not shown). In contrast, IL-6 showed a sustained increase (P = 0.008) during and after therapy which returned to baseline after 20 h (Fig. 2). C Co on nc cl lu us si io on n: : Despite a similar decrease in the number of both circulating T helper cells and monocytes, there seems to be a different change in the function of these cells during whole body hyperthermia up to 42°C. IL-2, which is postulated to be mainly produced in TH1 cells, decreased significantly; IL-4 and IFN-γ, mainly produced in TH2 cells, decreased slightly and IL-6, one of the main products of monocytes, showed a significant increase. Further investigations are necessary to verify these results.  I In nt tr ro od du uc ct ti io on n: : Clinical studies demonstrated that moderate hypothermia may improve neurological outcome after severe head injury. On the other hand hypothermia is associated with increased incidence of infection complications. Recent studies suggest that mild hypothermia directly impaires natural host defenses -leukocyte mobility, phagocytosis and reactive oxygen species production and antibody production. R Re es su ul lt ts s: : Our findings demonstrated that moderate hypothermia (33°C) did not alter the basic levels of IFN-γ, TNF-α, IL-1α, IL-2 and IL-10 mRNA expression in PBMC, but significantly inhibited increase in IL-2 mRNA expression caused by PHA stimulation. C Co on nc cl lu us si io on n: : These data strongly suggest that cytokine expression in stimulated human leukocytes can be affected by hypothermia. IL-2 is one of the key cytokines of immune response. It is known to stimulate growth and differentiation of T cells, B cells, NK cells, LAK cells, monocytes and macrophages. Hence, inhibition of IL-2 mRNA expression in PHA-stimulated PBMC by hypothermia can partially explain increased risk of infections in hypothermic patients. M Me et th ho od ds s: : Eighteen pigs were assigned to a T° group during CPB: 37°, 28° and 20°C, respectively (n = 6 each). Duration of CPB was 120 min and aortic clamping 60. Cardioplegia was achieved with a single dose of Bretschneider solution (4°C; 30 ml/kg). TNF-α was determined by a pig specific ELISA. Six hours post-CPB, tissue probes of the heart were taken for standard-and immunohistochemistry examinations. Apoptotic cells were detected by an in situ apoptosis detection kit (TUNEL). R Re es su ul lt ts s: : TNF-α production during and after CPB was significantly higher in group 37°C than in group 20°C. There was no TNFα production in group 28°C. Histological examination showed that the most important myocardial tissue damage in terms of intertitial edema, leukostasis and necrosis was seen in group 37°C followed by group 20°C while the least important damage was present in group 28°C. There was significantly lesser degree of apoptosis of myocardial cells in group 37°C than in both hypothermic groups. C Co on nc cl lu us si io on n: : Hypothermia during CPB induces a reduction of the systemic release of TNFα production and also of myocardial tissue damage. This could be due to increased apoptosis seen in the animals operated on in hypothermia. Apoptosis during cardiac operations could be in part responsible for the protective role of hypothermia. R Re es su ul lt ts s: : Intestinal I/R resulted in intestinal barrier dysfunction with pronounced plasma leakage to the intestinal lumen. A protely plasma activity was evident. MPO content significantly increased as did levels of interleukines. Treatment with the PAF inhibitor partly, though not fully, restored the changes caused by I/R. C Co on nc cl lu us si io on n: : PAF seems involved in the release of cytokines and consumption of protease inhibitors following intestinal I/R and the associated impairment of intestinal barrier integrity. Treatment with a PAF antagonist was effective in restoring changes caused by I/R, though not reaching normal levels. I In nt tr ro od du uc ct ti io on n: : Platelet-activating factor (PAF) and its inactivating enzyme PAF-acetyl-hydrolase (PAF-AH) are implicated in the development of sepsis and its sequela septic shock. It has been shown that the administration of rPAF-AH has a beneficial effect on the outcome of sepsis in animals as well as in humans.

P86 Platelets and platelet-activating factor acetylhydrolase in septic patients
M Me et th ho od ds s: : We measured PAF-AH activity daily in 2586 plasma samples that were obtained from 240 patients admitted to our intensive care unit. Patients were screened daily for sepsis according to ACCP/SCCM criteria, and PAF-AH activities were analysed in relation to severity of sepsis and to whole blood platelet count as an indicator of platelet activation and consumption. R Re es su ul lt ts s: : PAF-AH activity was positively correlated to the severity of disease, but was proved to be a poor sepsis marker, when compared to others, such as neopterin, TNFα, procalcitonin etc. In patients with septic shock a low PAF-AH activity (<2.00 µmol/ml/h) could indicate a high mortality risk. Only 4 patients met these criteria, but all died. Platelet count was highest in patients with uncomplicateted sepsis, but dropped dramatically in septic shock. The overall correlation between PAF-AH and platelet count was relatively poor (r = 0.266), but remarkable differences were observed between patients with PAF-AH activities <2.00 or >5.00 µmol/ml/h, resp.: 125 (70/112) × 10 8 /ml versus 280 (206/388) × 10 8 /ml; P < 0.0001. C Co on nc cl lu us si io on n: : Our data provide further evidence that PAF-AH has a beneficial effect in sepsis and that it can prevent platelet activation and sequestration which is known to contribute to multiple organ failure. Hemorrhagic shock predisposes to adult respiratory distress syndrome, which frequently results in prolonged ICU stay and carries a 50% mortality. We have previously shown that resuscitation with hypertonic saline (NaCl 7.5%) attenuates the post-hemorrhage lung injury by preventing neutrophil (PMN) sequestration. This beneficial effect was due to multiple effects on PMN function and included shedding of the PMN adhesion molecule L-selectin. The aim of the present study was to investigate the signalling pathway underlying this immunological effect. Isolated human PMN were treated with either iso (290 mOsm) or hypertonic (500 mOsm) medium, for up to 2 h. Hypertonicity induced extensive tyrosine phosphorylation in multiple bands. The broad-spectrum inhibitor genistein, abrogated this effect and concomitantly prevented the hypertonic (HT) shedding of L-selectin (graph). In order to characterize the tyrosine kinases involved in this process, we investigated which kinases were phosphorylated upon shrinkage, and then whether pharmacological inhibition prevented shedding. We found that the non-receptor tyrosine kinases Syk, Pyk-2 and the Src-family kinase Hck were strongly phosphorylated upon shrinkage. However, PP1, a Src-family inhibitor, prevented their phosphorylation but not the HT shedding of L-selectin, suggesting that this effect is independent of Src activation. Next, we found that the p38 was activated upon hypertonic shrinkage in a genistein sensitive but PP1 insensitive way.

No infection SIRS
Moreover, the inhibition of p38 activation by SB203580 significantly reduced the HT shedding suggesting that p38 is involved in this process (graph). The LPS-and FMLP-induced shedding of L-selectin was also abrogated by SB203580. THUS: Hypertonicity induces a unique pattern of tyrosine phosphorylation in human neutrophils, involving a variety of kinases, most Src-dependent. However, the hypertonic shedding of L-selectin seems to be selectively coupled to p38 activation. In fact, p38 appears to be a central mediator of L-selectin shedding induced by various stimuli. Hypertonicity-induced, p38-mediated L-selectin shedding appears to have an important role in the beneficial immune modulatory effect of hypertonicity, preventing neutrophil lung sequestration and cell-mediated tissue damage. I In nt tr ro od du uc ct ti io on n: : Acute renal dysfunction is a common postoperative complication of CABG. Extracorporeal circulation induces an inflammatory response, causing the release of adhesion molecules by endothelial cells and leukocytes. These adhesion molecules are incriminated in the pathophysiology of renal dysfunction, but their relative importance is unknown. We investigated the relationship between levels of s-ICAM and renal dysfunction following CABG.  R Re es su ul lt ts s: : As verified by Western blotting and secondary immunofluorescence human monocytes and HepG2 hepatoma cells express PCT in association with cytoskeleton. The content of PCT seems to be higher in monocytes than in HepG2 cells. No PCT was found in cytoplasmic fraction. C Co on nc cl lu us si io on ns s: : Our data suggest a downregulation of PCT levels in critically ill patients. Lacking rise of PCT serum levels with recurrent severe infections seems to be associated with high mortality. The predictive value of PCT for severe infections might be impaired by this mechanism. Further studies are required to verify these findings and to explain the potential reasons for failure of PCT in detection of recurrent infections of some patients in ICU. However, there were no significant differences in the levels of Creactive protein (CRP), interleukin 6 (IL-6) or tumor necrosis factor-α (TNF-α) between the two groups. PCT levels in SIRS patients with severe sepsis and septic shock (172.2 ± 276.3 ng/ml) were significantly higher than those in SIRS patients with sepsis. Levels of CRP, IL-6 and TNF-α were also significantly higher in the patients with sepsis compared to those in patients with local infection. Significant correlations were observed between the levels of PCT and those of CRP, IL-6 and TNF-α in SIRS patients. It was suggested that to measure the levels of procalcitonin in patients with SIRS is useful to diagnose the infection and severity of illness. I In nt tr ro od du uc ct ti io on n: : A high serum level of procalcitonin (PCT), an inflammatory mediator precursor of human calcitonin, has been detected in patients with inflammatory conditions from bacterial infection. The formation and release of PCT seems to be a selective induced response to bacterial inflammation or sepsis and it is sustained during a prolonged period of time compared with other inflammatory mediators. In relation to this, PCT could be an important parameter to evaluate patients with AP as systemic involvement and infectious complications that influence the antibiotic use, CT scan indication, invasive hemodynamic monitoring, and surgical intervention are frequently. . Neither CRP, cell counts nor the degree of fever showed significant differences between sepsis and severe sepsis. White blood cell count and platelet count differed significantly between severe sepsis and septic shock. C Co on nc cl lu us si io on n: : In contrast to AP-II, PCT appears to be a useful early marker to discriminate between sepsis and severe sepsis. We used PMX-20R in the treatment of 20 septic shock patients who developed complicated multiple organ failure caused by intestinal perforation. Extracorporeal circulation was performed for 120 min with blood flow of 80-120 ml/min. In this treatment, we assessed the changes of the patient's hemodynamics and the levels of cytokine levels (TNF-α, IL-6, IL-1ra and IL-10), soluble ICAM-1, thrombomodurin, NOx and PAI-1 during and after hemoperfusion using PMX-20R. Before the treatment, the mean APACHE-2 score of the 20 patients was 24.3, mean septic severity score was 46.9, and Goris score was 4.7. Mean arterial BP and LVSWI after the treatment increased significantly compared to the values in similar patients who did not receive this treatment, while our patients showed slight decrease in platelet counts. The values of the endospecy test (the new PCA method) apparently declined at the end of PMX treatment. The two inflammatory cytokines, TNF-α and IL-6, and the anti-inflammatory cytokines, IL-1ra and IL-10, decreased immediately after PMX and at 24 h after completing PMX. On the other hand, the levels of ICAM-1, thrombomodurin, NOx did not decline. PAI-1 decreased remarkably at the end of PMX treatment and 24 h after PMX. As a result, 15 of the 20 patients had good prognosis and 5 had a poor progno-sis. Therefore, hemoperfusion with PMX-20R could be a useful therapeutic measure for patients with septic shock caused by intestinal perforation, and it is recommended that this treatment shold be begun as quickly as possible after emergency surgical treatment. I In nt tr ro od du uc ct ti io on n: : Antithrombin III is a physiological inhibitor of thrombin, a central procoagulatory factor with pleiotropic activities. Decompensated disseminated intravascular coagulation in septic patients is associated with a rapid consumption of AT III. Therefore the anti-inflammatory effects of AT III is a main point of interest in the pathway of sepsis. To determine whether AT III concentration has beneficial effects on the severity of immunological function in sepsis, the present study investigated the association between AT III and the DR-expression on monocytes. R Re es su ul lt ts s: : There was a significant correlation between AT III and DR-expression on monocytes (P < 0.002). The substitution of AT III in a standard dose was associated with higher level of DRexpression on monocytes. Also the AT III level shows a linear correlation to IL-6 and TNF-α (P < 0.05; P < 0.03).

P99 Antithrombin III (AT III) prevents increased permeability and leukocyte adhesion in
C Co on nc cl lu us si io on n: : The results indicate that the AT III level is not only a marker of the disseminated intravascular coagulation in septic patients. Also there is a relationship to the process of inflammation. Higher levels of AT III were associated with a higher amount of DR-expression. Thus, the study confirmed the effect of AT III on the immunomodulation. During disseminated intravascular coagulation (DIC), the extrinsic tissue factor (TF)-dependent pathway has been implicated as the dominant route to thrombin generation and the production of IL-6 has been shown to correlate positively with the severity of sepsis-induced DIC. Pharmacological doses of AT have been shown to reduce mortality and morbidity in patients with DIC and there is increasing evidence to suggest that AT possesses antiinflammatory properties in addition to its anticoagulant properties.

P101 Effect of antithrombin III (AT) on lipopolysaccharide (LPS)-induced production of tissue factor and interleukin-6 (IL-6) by human umbilical vein endothelial cells (HUVECs), mononuclear cells (MNCs) and whole blood
In the present study, we have investigated the effect of AT on LPS-induced TF and IL-6 production in three different in vitro systems. Citrated whole blood, HUVECs and MNCs were stimulated with LPS for 4-6 h in the presence or absence of AT. TF activity was estimated by a TF-dependent clotting or chromogenic assay and IL-6 was measured by ELISA. Our results show a dose-dependent inhibition of TF and IL-6 production by AT, EC50 -~36 and 20-35 iu/ml respectively in MNCs and HUVECs, but ~14 and <10 iu/ml in whole blood. Immuopurifica-tion experiments confirmed that the inhibitory activity was attributable to the AT and not to components that may have co-purified with the clinical product. In addition, up to 40 µM of hirudin, a specific thrombin inhibitor, did not inhibit the production of TF and IL-6 in either of the three cell systems, suggesting that the observed inhibition by AT was not due solely to the inhibition of thrombin. Our investigation has shown that, apart from the inhibition of thrombin and other activated clotting factors, AT may also down-regulate the cellular expression of proinflammatory cytokines. Consequently, AT concentrates may have value in the treatment of sepsis-induced DIC. The purpose of this report is to (1) study whether PAI-1 related markers including tPA-PAI are parameters related to coagulopathy closely related to MODS, and (2) analyse whether the coagulopathy indicated by the elevation of the PAI-1 related markers is the cause of MODS in seriously ill acute septic patients with glucose intolerance. Patients under strict blood glucose control by artificial pancreas (AP) were selected in order to sample reliable PAI-1 values because fluctuation of blood glucose and serum fat levels are believed to influence the blood levels of PAI-1 related parameters. AP used was STG-22, manufactured by NIKKISOH corporation in Japan.
M Ma at te er ri ia al ls s: : Nine severe septic patients with glucose intolerance without NIDDM, aged 27-83 years were investigated. Primary diseases were, four patients with hepatobiliary diseases, two with gangrene of lower extremities, two with ARDS, and one with burn.
Analyzed items were (1) 7) The marked change of tPA-PAI levels apparently preceded those of MOF score in three out of eight patients and were parallel to them in four out of eight.
I In nt te er rp pr re et ta at ti io on n a an nd d c co on nc cl lu us si io on ns s: : Several important relationships between tPA-PAI and DIC, hypercoagulability, endothelial cell injury, and MODS became evident. These analyses were thought to be possible because strict blood glucose control was performed by using AP. The degree of MODS correlated with that of DIC and/or hypercoagulability. Among parameters related to coagulopathy, tPA-PAI was not only a sensitive marker of DIC and hypercoabulability, but also correlated well with the severity of MODS and the endothelial cell injury. Moreover, hypercoagulable state indicated by the elevation of tPA-PAI was thought to be one of the causes of MODS, and treatment for the hypercoabulability may be justified as an important method.
septic multiple organ dysfunction in clinical and experimental studies. This study investigates the AT III effect on leukocyte/endothelial cell interaction and microvascular perfusion. In the skin fold preparation of the hamster severe endotoxinemia was induced by repeated administration of endotoxin (LPS, 2 mg/kg), at t 0 = 0 h and t 3 = 48 h. AT III (250 U/kg) was substituted intravenously at t 0 , t 2 = 24 h, and t 3 (n = 6 animals, AT III group). In control animals (n = 5, controls) LPS was given without AT III substitution. By intravital fluorescence microscopy (FITC dextrane, rhodamine 6G) venular leukocyte adherence was determined at t 0 , t 1 = 8 h, t 2 , t 3 , t 4 = 56 h, and t 5 = 72 h. Functional capillary density (FCD) served as a measure of capillary perfusion. AT III resulted in a significant modulation of LPS-induced leukocyte adherence and in a modulation of the LPS-induced depression in FCD (P < 0.01, MANOVA). Thus, the number of sticking leukocytes after induction of endotoxinemia was significantly lower in the AT III group compared with control animals (AT III: t 1 = 182 ± 35 cells/mm 2 , t 2 = 176 ± 21, t 3 = 210 ± 51, t 4 = 243 ± 48, t 5 = 144 ± 29; control: t 1 = 630 ± 105, t 2 = 465 ± 113, t 3 = 404 ± 50, t 4 = 542 ± 93, t 5 = 356 ± 102; P < 0.05). AT III downregulated LPS-induced leukocyte/endothelial cell interaction and prevented the depression in FCD which served as a measure of capillary perfusion. Both mechanisms may explain beneficial AT III effects in patients with severe sepsis. A renaissance of the 'glucocorticoid discussion' emerged during the last years with reports of 'stress dose' or 'low-dose' hydrocortisone (HC) replacement therapy in patients with septic shock, assuming relative adrenal insufficiency, improving hemodynamic stability, modulating the inflammatory response, and probably improving outcome [1,2]. Here we present results from an interim analysis for the first 20 patients enrolled in a double blinded, randomized, placebo controlled, cross-over study to investigate the effects of HC infusion on 40 patients in septic shock [3] who needed norepinephrine (NE) for hemodynamic support. Patients were randomized to receive either HC 10 mg/h after an initial bolus of 100 mg, or placebo (PL). After 3 days, the medication was switched, i.e. patients who had HC for the first 3 days received PL for another 3 days, and vice versa. Plasma nitrite/nitrate (Griess reaction) was measured before the study and daily for 6 days, hemodynamic monitoring was performed before and every 8 h throughout the study period. No differences between the two group were found for age, sex, cause of sepsis, and severity of illness at time of study entry established by SAPS II and SOFA. HC treatment allowed marked reduction of NE infusion within 48 h after study began (Fig. 1). Systemic vascular resistance (SVR) increased with HC infusion but remained unchanged in the PLgroup during the first study period (Fig. 2). When HC was switched to the other group, SVR decreased despite increased NE requirement in patients who received HC before, whereas in the other group SVR increased and NE could be reduced. Mean arterial blood pressure, but not cardiac index, paralleled changes of SVR. Plasma nitrite/nitrate decreased with HC infusion, indicating suppression of endogenous nitric oxide (NO) production (Fig. 3). Interestingly, rebound phenomenon after cessation of HC was not accompanied by increased nitrite/nitrate concentrations.
C Co on nc cl lu us si io on ns s: : In patients with septic shock, stress-dose HC infusion improves hemodynamic stability, reduces NE requirement and increases SVR. Improvement of SVR may be due to HCinduced suppression of inducible NO synthases (iNOS) and/or suppression of the synthesis of iNOS stimulating cytokines. Cessation of HC infusion induces rebound effects which seem to be NO-independent.  C Co on nc cl lu us si io on n: : We could show that stress-dose hydrocortisone given to patients in septic shock markedly decreases soluble E-selectin and IL-6 levels. In addition to the hemodynamic effects of this therapy, there also seems to be an immunmodulating and antiinflammatory effect, which might be organ protective.  R Re es su ul lt ts s: : There were no striking differences in monocyte HLA-DR expression between patients who received HC or placebo ( Fig. 1). However, compared to baseline values, a transient decrease of HLA-DR expression was observed in the group which received HC early. INF-γ increased in both groups after start of the study, but returned to baseline in the placebo-group on day 3 (Fig. 2). In the follow-up, INF-γ did not further increase in the placebo-group but noticeably in the HC-group. C Co on nc cl lu us si io on n: : Stress-dose HC treatment did not induce immunoparalysis in patients with septic shock during the study period. HLA-DR expression remained almost constant over the period of the trial which we postulate to be due to HC-induced increase of INF-γ synthesis. Experimentally it was recently well established that gender differences lead to an increased susceptibility to sepsis in males.
In a prospective clinical study gender differences were evaluated in patients of a surgical ICU in terms of survival, sex hormones and cytokine response. Fifty-two critically ill patients (19 females and 33 males) were included in this study -there was no difference in the characteristics of patients concerning the age, cause of sepsis and severity of their disease.
Mean age was 55.4 years for females and 53.1 for males. APACHE II score was 17.3 for females and 18.5 for males at entry of the study, MOD-score 9.9 versus 10.8 respectively. Biactivity of TNF and Il-6 were measured for 14 days, as well as Il-10 (ELISA), total testosteron and 17β-estradiol (RIA).
Though clinical assessment did not reveal any difference, prognosis and outcome of sepsis was significantly different in males and females: MOD-score was always similar in both groups, however, hospital mortality was significantly different with 70% (23/33) in male and 26% (5/19) in female patients (P < 0.01, log-rank test). Evaluation of cytokine response revealed significantly elevated TNF levels on day 10 in males (P < 0.05 Mann-Whitney U-test) while no difference was found for Il-6 levels. Females, however, displayed enhanced Il-10 levels compared to males from day 1 to day 10 which reached significant levels of P < 0.05 on day 3 and day 5. Total testosterone levels were below the normal range for males and estradiol levels were initially increased both in men and postmenopausal women with higher levels for women.
Sex dismorphism, as shown, with a significant better prognosis and outcome of sepsis in women should be considered as a novel therapeutic approach (testosterone receptor blockade) in sepsis. Unit of a large Canadian University affiliated teaching hospital were screened for study eligibility into one of three multi-center sepsis trials. The screen log defines patients who meet inclusion criteria as eligible. Reasons for non-enrollment are divided as follows: 1, study specific exclusion criteria; 2, hopeless prognosis; and 3, enrolled in another trial. Truly eligible patients were those who did not meet above criteria 1-3 and were not enrolled because informed consent could not be obtained or because the window of eligibility was missed by study personnel. Recruitment efficiency was calculated as the proportion of patients enrolled of those who were truly eligible. R Re es su ul lt ts s: : During the 23-month period of screening, 559 patients were admitted with sepsis or presumed sepsis. The inclusion criteria were met for 273/559 (48.8%) patients in least one of these three sepsis studies. Only 37/559 (6.6%) were enrolled into a sepsis trial. The Table contains the number of eligible, excluded (with reasons for exclusion) and enrolled patients screened for entry into the three trials. C Co on nc cl lu us si io on n: : Within an institution that actively participates in sepsis clinical trials, only a minority of patients with sepsis are treated in the context of a trial. The impact of study-specific exclusion criteria is to create very different study populations. Such differences may account in part for the discordant results seen in Phase II and Phase III trials and raise important questions regarding the external validity of conclusions from trials with low inclusion or recruitment efficiencies.  The cardiodepressant IL-1β effect was documented by a lacking response in pulsation amplitude to the isoproterenol-challenge (control: n = 20, 148% ± 20 versus IL-1β: n = 27, 103% ± 3*, P < 0.05), which was preserved by co-incubation with Dex. (control: n = 21, 131% ± 10 versus IL-1β: n = 27, 130% ± 8). Arrhythmias were regularly elicited in controls upon α-adrenoceptor-stimulation (16/17), even if the duration of the electrical pulse was increased to keep the cells in pace. In contrast, recordings of IL-1β-treated CM (n = 11) did not display beating irregularity. If, however, Dex. was added to the incubation medium, arrhythmias occurred both in the groups without IL-1β (9/11) and with IL-1β (9/10). C Co on nc cl lu us si io on n: : A potentially beneficial antiarrhythmic effect of IL-1β may go along with its cardiodepressant action in vivo.  An APACHE-II-score ≥24 on the 1 po day is a prospectively validated parameter to identify cardiac surgery patients with an escalating 'post pump' systemic inflammatory response syndrome at high risk of multiple organ dysfunction syndrome. We investigated the impact of cardiopulmonary bypass-assisted cardiac surgery on monocyte markers in a prospectively conducted study for up to 5 days (group 1: APACHE II ≥24, group 2: APACHE II <24), compared to septic non-surgical patients (Elebute sepsis score ≥12, APACHE II score ≥24). C Co on nc cl lu us si io on n: : Extracorporeal circulation is associated with the excessive release of IL-10 and lactoferrin and the kinetic of production of these both factors is very similar. This may suggest that they participate in the limitation of the inflammatory process during ECC.  R Re es su ul lt ts s: : 1) The IL-6 plasma levels increased in all three groups to a significantly different degree with maximal IL-6 levels between 3 and 24 h after intervention.

P110 Decreased beating rate variability of cultured cardiomyocytes by endotoxin
2) The levels of the three collectives were significantly different at 3, 6, and 24 h (Table).
3) The corre-lation of IL-6 peak levels and duration of CPB was stronger in CPB-supported PTCA than in CPB-CABG. R Re es su ul lt ts s: : We detected no significant levels of IL-1α, IL-1β, IL-8 and TNF-α during CPB support. IL-1ra levels were increased. IL-6 levels increased measurable starting 30 min after begin of CPB with peak values of 20-60 pg/ml between 3 and 12 h. One patient showed significant levels of IL-10, this patient expressed the lowest level and shortest kinetic of IL-6 production and more pronounced TNFα-receptor levels, although TNF-receptor levels increased in all patients. sCD14 raised continously in all 3 patients to a maximum of 7 ng/ml followed by a plateau for more than 5 days. C Co on nc cl lu us si io on n: : In this study we compared the cytokine levels of patients undergoing high risk coronary angioplasty with CPBsupport. There were no findings to show a significant relation of IL-1α, IL-1β, IL-8 and TNF-α to the inflammatory response after CPB-PTCA -this might be a sign for other mechanisms than systemic activation of monocytes by endotoxin may be involved. IL-6 as a marker of the degree of systemic inflammatory reaction increased significantly. We suggest, the mayor source of this increase of IL-6 levels is the CPB-support. IL-6 release might be inhibited by IL-10 production. Direct coronary angioplasty of the infarct related artery is well accepted as one of most important therapeutic options for cardiogenic shock (CS) complicating acute myocardial infarction (AMI). However, in-hospital still remains high. The aim of the following study was to analyse which clinical and procedural factors were associated with high or low in-hospital mortality when primary PTCA is applied systematically to all patients with CS within 12 h of symptom-onset. Patient chararcteristics: n = 78, age 60 ± 14 years, male 67%, primary venticular fibrillation, mechanical ventillation 59%, total branch block 24%.

P116 tPA-lysis leads to reduced levels of sL-selectin in patients with acute myocardial infarction
P Pr ro oc ce ed du ur ra al l d da at ta a: : Single-vessel-disease 41%, ejection fraction (acute biplane) 0.51 ± 0.16, infarct-related artery: LAD 38%, LCX 8%, RCA 54%; intra-aortic-balloon-pumping 17%, coronary stents 28%, successful angioplasty 87% (TIMI 3, residual stenosis <50%), in-hospital mortality (0-30 days) 49%. The most important predictors for a high in-hospital mortality rate were: acute ejection fraction <40% (P = 0.0035), unsuccessful PTCA (P < 0.05) and patient age >75 years (P < 0.05). A high in-hospital mortality rate was also seen in patients requiring mechanical ventillation. Mortality did not depend on the infarct location (inferior versus anterior), patient sex, ventricullar fibrillation or total branch block prior to intervention, single-or multi-vessel-disease. Furthermore mortality was independent of the time between onset of symptoms and PTCA and was also not affected by the employment of coronary stents or intra-aortic balloon conterpulsation.
C Co on nc cl lu us si io on n: : Systematic primary-PTCA results in a lower in-hospital mortality rate when compared to conservative therapy of AMI with CS. However, mortality remains extremely high if angioplasty is unsuccesful. But even if myocardial perfusion is able to be re-established, patients initially requiring mechanical ventillation or with a low acute ejection-fraction as well as the very elderly >75 years of age maintain a poor prognosis. R Re es su ul lt ts s: : The storage-induced decline in ATP and 2,3 DPG in human RBCs were consistent with the literature. These changes, however, occurred more rapidly in rat RBCs; ATP levels after 7 days of storage declined to the same extent as human RBCs after 4 weeks (40% decrease). DPG levels in rat and human RBCs fell by 60% and 90% after 7 days of storage. By day 7 of storage the mean membrane deformability had dropped 45% (P < 0.001). RBCs exposed to the rejuvenation protocol at day 7 had ATP levels returned to baseline while the mean RBCd showed almost complete recovery to baseline levels. Significantly, 12% of the population of rejuvenated cells still showed compromised membrane deformability (i.e. membrane displacement less than 80% of baseline).

P119 The storage-lesion in murine red blood cells: comparison to stored human red blood cells and applications
C Co on nc cl lu us si io on n: : The biochemical data from this study suggest that rat RBCs stored for 7 days develop a storage-lesion similar to that of human RBCs stored for 29 days. Rejuvenation of RBCs improves RBCd and may be related to improved ATP levels. Using rat RBCs stored for 7 days gives researchers a valuable tool to assess blood storage and the consequences on transfusion efficacy and tissue oxygen availability. After 60 min, however, intravascular survival of septic and ETX treated RBC started to fall in healthy rats and was significantly lower than survival of naive RBC at 240 min. In septic rats, not only the E and S-RBC but also the naive RBC were cleared by 15% at 240 min. In healthy animals, sequestration of transfused S-and E-RBC in liver and spleen was higher than sequestration of N-RBC. In septic animals, no difference in sequestration was found between N-, E-, and S-RBC. In skeletal muscle, lungs, intestine, femur, diaphragm and skin, sequestration was not different between healthy or septic recipients nor between the different groups of transfused RBC. In our previous study [1] we demonstrated that the ozonized autohemotransfusion increases erythrocyte filterability and reduces plasma and total blood viscosity, thus supporting the theory that the ozone-induced improvement of peripheral vascular disorders (PVD) might be related to its influence on the hemorheologic properties of blood.
Given that oxygen delivery to tissues is dependent on its affinity for hemoglobin, in the present study we evaluated the effect of oxygen-ozone treatment on hemoglobin oxygen affinity in 15 patients suffering from PVD (clinical stage IV according to Fontaine).
Before and 30 min after slow reinfusion of 150 ml of autologous venous blood exposed in a glass box to an O 2 -O 3 mixture (3.6 mg of total ozone erogation with a Multiossigen Medical 93 Multi Tech Milano, Italy) we evaluated hemoglobin oxygen affinity using p50 STD value. This value is defined as oxygen tension in mmHg at 50% oxygen saturation, at pH 7.4, at 37°C and at pCO 2 40 ± 2 mmHg and is calculated according to the formula: p50 std = antilog [log p02 (log s02/2.7) -0.4 (7.4-pH)].
The value of 2,3 diphsophoglycerate (2,3 DPG), an important regulator of oxygen unloading were also evaluated before and after ozone treatment. Control studies in the same patients were done in other occasions and random order to test the influence of blood manipulation (non-ozonized autotransfusion). The results ( Table) show that after ozone treatment, p50 value increased (P < 0.005), whereas plasma value of 2,3 DPG did not significantly change. No significant changes were induced by control tests (non-ozonized autohemotransfusions).
These results strengthen the conclusions of our previous 'in vitro' study [2] proving that the ozone treatment shifts to right the oxygen hemoglobin dissociation, ultimately resulting in improved tissue oxygenation. I In nt tr ro od du uc ct ti io on n: : Severe metabolic alkalosis is a common problem in ICU due to high gastric reflux, diuretic drugs or parenteral nutrition. The mortality is considerable. Therapy of severe metabolic alkalosis with hydrochloric acid is widely accepted and described as save and effective.
P Pa at ti ie en nt ts s: : We report five unexplained cases of acute cardiac arrest on our ICU in 1998 following infusion of hydrochloric acid in mechanically ventilated patients with severe metabolic alkalosis. Treatment with HCl (0.2 N, 10-15 ml/h) was commenced at pH 7.55 and BE +10 mmol/l. All patients suddenly showed a marked fall of oxygen saturation followed by bradycardia, hypotension and cardiac arrest 30 to 140 min after onset of HCl-infusion. CPR and high dose adrenalin (up to 5 mg in bolus) showed no effect. Oxygenation could not be improved by    Pulmonary edema develops when the movement of liquid from the blood to the interstitium exceed the return of the liquid to the blood. The diagnosis of this interstitial expansion is based on chest X-ray and the basic clinical signs appear when the pulmonary compliance is reduced (aspecific dyspnea), a large mismatch exists between ventilation and perfusion or with the onset of alveolar flooding (moist and fine crepitant rales, wheezes or extreme breathlessness). The detection of an interstitial edema is a crucial step in the diagnostic procedure in a dyspneic patient and the intensivist or the emergency physician must make daily therapeutic decisions on the basis of a bedside clinical examination, often difficult, and chest X-ray, wich is known to be often technically deficient.
M Me et th ho od ds s a an nd d p pa at ti ie en nt ts s: : During a 4-month period, 83 patients (49 males and 34 females, mean age 74 years) admitted to our Emegency Room (ER) were included in a prospective study.They showed dyspnea (>25 breaths/min) and/or discomfort and signs of augmented work of breath (inspiratory retraction of the intercostal spaces and supraclavicular fossa) or orthopnea. Immediately after the clinical examination, all patients underwent chest sonography. Longitudinal scans of the anterior, lateral and posterolateral (or posterior in the sitting patient) chest wall were taken using a Toshiba SSA250A portable unit equipped with a 3.75 MHz convex transducer. We particularly studied the respiratory motion of the pleuropulmonary surface (gliding sign) and the comet tail artifacts arising from the lung surface. These artifacts are roughly vertical narrow based projections spreading up to the edge of the screen and appear when there is a marked difference in acoustic impedance between subpleural septa thick-ened by edema and the alveolar air (alveolar-interstitial syndrome). Chest radiographs of all patients performed during the same period of ER evaluation were interpreted by radiologists unaware the sonography findings and classified on the basis of widely accepted criteria. Once the diagnosis of wet lungs was sonographically confirmed, the patients were considered for hearth failure treatment (diuretics and vasodilators) in absence of other diagnostic possibilities or particulary controindications, while the patients with dry lungs underwent advanced diagnostic work up. Finally we evaluated the effect of the diuretic therapy (in 3 h), the correlations between radiologic and sonographic patterns and the usefulness of ultrasonography in the diagnostic approach to the critically ill patients.
R Re es su ul lt ts s: : All patients were successfully and quickly (<5 min) analysed using ultrasound (feasibility 100%). Twenty-nine subjects (34%) showed 'wet lungs' with diffuse bilateral comet tail artifacts. Of these, 21 (72%) had associated pleural effusions (bilateral in 13 cases), with water levels (WL) between 1 and 8 cm. Chest X ray discovered congestion or edema in 30 pts. (flow inversion, enlarged/iperdense ila: 5 pts. blurred ila, perivascular/peribronchial cuffs, Kerley B lines: 13 pts.; patchy alveolar edema: 7 pts. and confluent alveolar edema: 5 pts.), 29 of them exibiting diffuse artifacts. Pleural effusions were shown radiologically in 13 subjects with eight missed diagnoses (effusions with WL <2 cm). One discordant case was noted (sonographic false negative) but none of 51 patients with normal X rays had significant comet tail artifacts. Sonographic imaging led to a change in the initial diagnosis (hearth failure) in 11 pts. (13% of the whole group studied), six with COPD, three with pulmonary embolism and one with important anemia, these subjects showed normal chest X rays and sonographic 'dry lungs'; one patient had the diagnosis changed from COPD to hearth failure. C Co on nc cl lu us si io on n: : We think that echography offers a new method for the diagnosis of alveolar-interstitial syndrome at bedside and may provide vital informations when a radiograph is not readily available or undesiderable. Moreover it may be valuable for differentating cardiogenic pulmonary edema from decompensated COPD or pulmonary embolism showing, in our experience, a sensibility of 96% and a specificity near to 100% for diagnosing 'wet lungs'. M Me et th ho od ds s: : Ten patients on the ICU admitted pre-operatively for haemodynamic optimization were prospectively studied. Each patient had measurement of cardiac output by a thermodilution right heart catheter technique and also by IP. Data sets were obtained during the procedure of haemodynamic optimization. Data was analyzed using regression analysis for the differences between cardiac output measurements between the two techniques, depending on both the patient and the absolute level of cardiac output.

References
R Re es su ul lt ts s: : Ten patients were analyzed with a total of 51 pairs of cardiac output measurements. Thermodilution cardiac outputs were obtained between 4.1 l/min and 10.5 l/min. Regression analysis revealed a significant difference between cardiac output measurements for the two techniques (P < 0.0001). Differences existed when the data was analyzed between patients and also when looking at different measurements for individual patients.
D Di is sc cu us ss si io on n: : IP would be ideal non-invasive tool for the measurement of cardiac output in ICU patients. This study suggests, however, that there are major differences between the cardiac outputs obtained from thermodilution and IP for pre-operative patients on the ICU.  Their effect is usually controlled by physical examination. The dosage in daily routine occasionally is not adapted to the patients needs in the circardian course. Evidence exists that classical neurological and hemodynamic parameters do not always reflect the level of sedoanalgesia. Neuromonitoring with heart-rate-variability (HRV) is a new opportunity to evaluate the patients neurological status. HRV is a window for usually invisible central autonomic regulation. This phenomenon is caused by oscillation in the interval between consecutive heart beats. It represents a quantitative marker of autonomic activity. Currently monitoring of autonomic nervous system is no routine tool for mechanically ventilated patients. Our study presents first results of continous neuromonitoring of autonomic nervous system with heart-rate-variability in the setting of an intensive care unit.

P126 Hemodynamic monitoring by double indicator dilution technique in patients after orthotopic heart transplantation
M Me et th ho od ds s: : We studied 10 mechanically ventilated patients (5 male, 5 female) without any cardiovascular diseases in case history who received analgesics (Fentanyl ® ) and sedatives (Dormicum ® ) continuously. Heart-rate-variability was recorded with a flash-memory recorder, the analysis of HRV was performed by a special software (both elamedical, Munich). We investigated over a period of 24h in each case. B Ba ac ck kg gr ro ou un nd d: : We have previously described a simple indicator dilution method of measuring cardiac output in which lithium chloride is injected via a central venous catheter and its plasma concentration-time curve measured in arterial blood using a lithium selective electrode [1]. This technique has the advantage of requiring only central venous and arterial cannulation and therefore avoids pulmonary artery catheterisation.
There are many patients who have arterial cannulae and peripheral, rather than central, venous access in whom cardiac output measurements would greatly assist management.
We have therefore explored the feasibility of measuring cardiac output using a peripheral rather than central venous injection of lithium chloride. R Re es su ul lt ts s: : There was good agreement between the two methods (r 2 = 0.98) and the results are summarised in the Table. C Co on nc cl lu us si io on n: : This study shows that LiDCO can be measured using peripheral or central venous injections of lithium chloride. Safe, quick and reliable cardiac output measurements can therefore be obtained in patients with arterial and antecubital venous access.    Table. D Di is sc cu us ss si io on n: : Other authors reported GEDV modifications correlated to cardiac index rather than standard cardiac filling pressures such as CVP and PAWP, so that GEDV is considered a 'pure' volume indicator. EVLWI can be considered a potential indicator of lung damage and pulmonary function, and it is hypothesized to be a better endpoint than PAWP during fluid management.
In conclusion monitoring EVLWI, as pulmonary edema indicator and GEDV, as cardiac preload value, lead our diagnostic and terapeutic management during major surgical procedures. I In nt tr ro od du uc ct ti io on n: : In order to minimize complications of invasive techniques for measurement of cardiac output, non-invasive methods will be of growing importance in anesthesiology and intensive care. The validity of non-invasive measurements has frequently been questioned. In our study the validity of non-invasive measurements was assessed in supranormal cardiac output.
M Me et th ho od ds s: : In 12 patients, who underwent whole body hyperthermia (WBH) in general anesthesia, aortal blood flow was measured using an esophageal Doppler probe (DYNEMO 3000, Sometec, France). Measurements were performed at 37°C, 40°C, 42°C and 39°C body core temperature. At the same time cardiac output was determined by the invasive thermodilution method using a pulmonary artery catheter (Swan-Ganz-Catheter) as well an arterial catheter (PULSION COLD ® ). Blood flow in the descending aorta is assumed to represent approximately 70% of total cardiac output, accordingly 70% of the value measured with the invasive technique was compared to the non-invasive measurement. For cardiac output measured by Doppler ultrasound median values were evaluated over a period of 5 min. Statistics were performed using the Mann-Whitney-U-Test.
R Re es su ul lt ts s: : There were no significant differences between values obtained with the two different invasive techniques. Values measured with Doppler ultrasound were significantly lower at 40°C (P = 0.04) and 42°C (P = 0.01) compared to invasive measurements, despite a growing tendency with rising temperature (see Figure).  M Ma at te er ri ia al l a an nd d m me et th ho od ds s: : Eighty-four medical or surgical patients were studied (mean age 51, SD 17; mean SAPS II (1 st day) 56, SD 9). After 6 h of ICU stay, a 7.5 F pulmonary artery catheter and a 4 F femoral artery catheter, with thermistor and fiberoptics were inserted and connected to 'COLD System', an integrated monitoring system which uses the double indicator technique for studying blood volumes. All patients were in CMV (PEEP <8 cmH 2 O); haemodynamic management was realized in order to optimize cardiac output (CO, l/min/m 2 BS) and systemic oxygen delivery. Infusion of crystalloids and colloids was guided by measurements of CVP and PCWP. All data were recorded at the beginning of the study (T0) and after 6 (T1), 12 (T2), 24 (T3) and 36 (T4) h. Statistical analysis of data was performed using Manova Test, considering the significant differences in the times of study between group A (38 pts., ITBV in T0 <1 l/m 2 BS) and group B (46 pts., ITBV in T0 >1 l/m 2 BS) and analysing the variance of repeated measures. Levels of P < 0.05 were accepted. R Re es su ul lt t a an nd d c co on nc cl lu us si io on ns s: : The Table shows the trends of parameters in the times of study (data are expressed as mean and (SD); A vs B, $ P < 0.0001; *P < 0.05; T vs T0, § P < 0.05).

P133 Trends of volemic indicators in a group of critically ill patients
When preload is the main determinant of CO, CVP and PCWP may be misleading in management of volemia in mechanically ventilated patients, on the contrary ITBV may be useful to optimize central filling and haemodynamic conditions.   C Co on nc cl lu us si io on n: : At least for patients on a surgical intensive care unit, single transpulmonary thermodilution is sufficiently accurate for the estimation of intrathoracic blood volume and extravascular lung water.  C Co on nc cl lu us si io on n: : These results suggest that measurements of phi and TBV using BIA combined with a difference of COP levels before and after administration of albumin are promising approaches to discriminate CLS from non-CLS patients at the bed-side. R Re es su ul lt ts s: : There was a clear linear relationship between impedance (ohms) plotted against body mass index (r = 0.95, P < 0.0001) in the control babies. Using control data for BMI and Z, an equation was developed using linear regression to predict Z from BMI values. This equation was used (Z=19.04BMI + 114.6) to predict values for Z expected in the babies with bronchiolitis on admission. These values were compared with those at admission (Z 1 ), where using paired t-test a significant difference existed (P < 0.0008) and at discharge (Z 2 ) where there was no significant difference. There was no significant relationship between changes in body impedance and changes in fluid balance (both positive or negative) although impedance changed appropriately in association with fluid boluses or with diuretics. C Co on nc cl lu us si io on n: : Regular impedance measurements give a guide to the state of hydration of babies requiring intensive care and help determine whether the baby is adequately hydrated or not. There is a poor relationship between measured changes in fluid balance and changes in impedance and it may not be used to calculate absolute values of fluid required.  [1,2]. We designed an experimental situation where external cardiac pressure conditions were controlled and adjusted to physiological extremes to mimic clinically relevant situations, while cardiac performance was assessed using left ventricular pressure-volume relationships (LVPVR) which are relatively preload and afterload independent.

P138 External cardiac pressures and the left ventricular pressure-volume relationship
M Me et th ho od ds s: : Healthy adult pigs (n = 4) were anesthetized, received central vascular catheters, a pericardial catheter, and bilateral pleura drains. Left ventricular volume was assessed by the conductance technique [3]. External cardiac pressures were manipulated: pneumothorax (20 ml/kg air injected in the pleura), and pericardial infusion (mean pressure of +6 mmHg). End sytolic elastance (Ees), preload recruitable stroke work (PRSW), preloadadjusted maximal power (PWRmax/EDV2) [4]. R Re es su ul lt ts s: : During pericardial infusion, where the end-systolic pressure was low and limited in beat-to-beat decrement during the preload reduction, only elastance increased while the other derived systolic parameters decreased. During pleura insufflation, all the systolic function parameters increased.
D Di is sc cu us ss si io on n: : These data suggest that relatively load-independent means are needed to assess cardiac function in the setting of extreme extracardiac pressure. LVPVR provides beat-to-beat insight into heart function at wide ranges of loading conditions. Further work is warranted to validate clinically applicable means to implement this type of assessment, as well as to further develop reference methodology for experimental and clinical heart volume assessment.  D Di is sc cu us ss si io on n: : The software simulates realistically the altering of preload (Ved) and afterload (Ea, HR, SVR), contractility (Ves, Vd) and the corresponding modifications of ME. LVF and ME evolve according to theoretic and experimental expectations, i.e. ME = 0.28 to Ved = 250 ml; ME max to Ea/Ees = 0.5. R Re es su ul lt ts s: : We interrupted the study as a significative difference in mortality rate was seen. The mortality rate in control group (50%) was significantly higher (P < 0.05) when compared with protocol group (15.7%). The incidence of clinical and infectious complications was higher in control group (P < 0.05) and organ dysfunction evaluated by SOFA score occurred more frequently in non-achievers. I In nt tr ro od du uc ct ti io on n: : The physiological assumption of oxygen debt elimination is a spontaneous or therapeutically induced increase of DO 2 in relation to the values of DO 2 recorded before the occurrence that led to the oxygen debt formation. The target of this research was to analyse the relation between post-operative achievement of DO 2 I >600 ml/min/m 2 and the dynamics of changes of DO 2 during the perioperative period and in that way to give an answer to the question if the achievement of so called 'supranormal values' of DO 2 is equal to the physiological principle of oxygen debt elimination.

Pleural intervention Pericardial infusion
M Me et th ho od ds s: : There were included 36 high-risk surgical patients in the prospective research (age 53 ± 15 years, 28 male and 8 female, 58% extensive ablative surgery for carcinoma). PA catheter and arterial catheter were inserted 12 h before surgery in the average. The target of therapeutical approach was to reach DO 2 I >600 ml/min/m 2 within 12 h from the end of surgery in every patient and then to keep these values during following 36 h. Haemodynamic measurements and laboratory analyses of blood samples, including arterial lactate, were analysed during the first 48 h after an interval of 6 h. While the data were analysed, we were comparing the dynamics of changes of DO 2 and arterial lactate in the peri-perative period in relation to the real achievement of the therapeutical target (DO 2 I >600 ml/min/m 2 within 12 h after the end of surgery) and the survival rate of the patients. R Re es su ul lt ts s: : The 28-day mortality was in the whole group of patients 31% (11/36). We achieved therapeutical target in 22 patients (61%). The mortality was 23% (5/22) in this group which was not statistically lower compared to the group of patients where the target was not achieved (43%, 6/14). When we compared the dynamics of changes of DO 2 and arterial lactate during the perioperative period in relation to the real achievement of the therapeutical target and surviving of patients, we found out in group of patients which achieved target the results which are demonstrated on the following figures. The results which were found out in group of patients which did not achieve target were the same. C Co on nc cl lu us si io on n: : Regardless of post-operative achievement of DO 2 I >600 ml/min/m 2 , in survived patients there was observed that they were achieving statistically higher values of DO 2 in comparison with pre-operative values of DO 2 and this process was accompanied by a decrease of arterial lactate level. We suppose that supranormal values of DO 2 should be define in relation to the pre-operative (i.e.normal) values of DO 2 and not in relation to the 'magic number' 600 ml/min/m 2 .   R Re es su ul lt ts s: : There was a significant positive correlation (P < 0.001) between the three methods.

P142 Correlation between three methods of calculating oxygen extraction ratio (OER) B Prasad, S Giles and F Gao Smith
D Di is sc cu us ss si io on n: : These three methods can be used to calculate OER. However, the conventional method (OER1) appears to have more variability than OER2 and OER3. OER3 (SpO 2 and SvO 2 -CCO), in contrast, is the simplest and most accurate method for continuous monitoring of OER in intensive care management. I In nt tr ro od du uc ct ti io on n: : A pathologic dependency between oxygen consumption (VO 2 ) and oxygen transport (DO 2 ) is characteristic of septic patients. Septic state appears to be associated with a defect in oxygen extraction (OER = VO 2 /DO 2 ), causes possible cellular hypoxia, mitochondrial dysfunction and development of multiple organ failure. We studied the relationship between the mean OER and age in septic patients in a 12-bed ICU.  While the AKBR reflects the glycolysis function of liver mitochondria, we hypothesized similar relationships between the blood concentrations of lactate and the AKBR in lactic acidosis because when the function of mitochondria is low, aerobic glycolysis in the Krebs cycle is disordered and hyperlactemia is caused.

P145 AKBR (arterial ketone body ratio) associates with lactate in the lactic acidosis
M Me et th ho od ds s: : We studied 20 MODS patients in our ICU. We surveyed their blood concentrations of lactate and AKBR and analyzed arterial blood gas when they were in a state of shock. Then we studied the correlation between the lactate concentrations and AKBR in lactic acidosis and non-acidosis. The correlation coefficient was found by regressional analysis each in lactic acidosis and non-lactic acidosis.
R Re es su ul lt ts s: : Among 20 MODS patients, in 10 lactic acidosis patients who were all type A lactic acidosis the blood levels of lactate and the AKBR were interrelated. The correlation coefficient was 0.75 and that showed the blood levels of lactate correlated closely with the AKBR in lactic acidosis. When the AKBR was low, the blood level of lactate rose, but in the other 10 non-lactic acidosis patients, the blood levels of lactate had no relation with the AKBR.
C Co on nc cl lu us si io on n: : We conclude that the AKBR has a strong relation to the concentration of lactate in lactic acidosis, but in non-lactic acidosis there was no relation. This study clinically proves that in lactic acidosis the blood concentration of lactate reflects the AKBR which is shown by the malfunction of liver mitochondria. Now right at the bed side, we can survey lactate concentrations easier and at a lower cost than the AKBR. Admission lactate did not increase with increasing MOSF score (P = 0.5). However mortality increased with increasing MOSF score (P = 0.005).

P146 Blood lactate: an excellent prognostic indicator in high-risk surgical patients
C Co on nc cl lu us si io on n: : Early hyperlactataemia is associated with a high mortality in critically ill children. Organ failure and peak lactate levels may distinguish nonsurvivors in this group. I In nt tr ro od du uc ct ti io on n: : Monocytes are within the first line of an organisms immune defense. However in the course of sepsis they undergo apoptotic cell death [1,2]. It is unclear whether this serves to protect the organism from a hyperreactive inflammatory response or is a sign for immune dysfunction.

Survivors
The artificial colloids hydroxyethylstarch (HES), dextran (DEX) and gelatine (GEL) are essential in perioperative volume replacement as well as in the treatment of trauma, shock and sepsis. In this study we investigated whether artificial colloids influence survival or apoptosis of human monocytes in vitro.
M Me et th ho od ds s: : Monocytes were isolated from buffy coats of healthy donors by gradient centrifugation and adherence to plastic culture dishes. They were incubated for 8 h with HES, DEX and GEL at 10 to 40 mg/ml. Staurosporine was added to induce cell death. Alive, apoptotic and necrotic cells were identified by Annexin V/Propidium Iodide staining and 10 000 cells were analysed by flow cytometry. Presence of apoptotic cell death was confirmed by electron microscopy, TUNEL, and cell death detection ELISA. Regression analysis of colloid concentration against cell status was performed. Slope values were tested with Students't-test against 0. Significance was assumed for P < 0.05. R Re es su ul lt ts s: : All artificial colloids reduced the fraction of viable cells in a concentration dependent manner. This effect was significant with DEX. Apoptotic cells, which were calculated as a fraction of dead cells were reduced with DEX more than with HES, but increased significantly with GEL.
Incubation with staurosporine reduced cell viability and increased the fraction of apoptotic cells. Neither colloid nor concentration had additional influence. Only the results for HES are shown. C Co on nc cl lu us si io on n: : DEX, HES, and GEL promote monocyte death in vitro. This effect is concentration dependent, but most obvious beyond concentrations found in clinical practice. Cell viability is reduced most by DEX, whereas GEL seems to delay the course of cell death, as apoptotic cells undergo secondary necrosis in vitro. Staurosporine induced cell death is not blocked.  Polygeline (a polymer prepared from heat-hydrolyzed gelatine) is a plasma substitute used by infusion as a 3.5% solution in the management of hypovolaemic shock. TSEs (transmissible spongiform encephalopathies) are a group of fatal neurodegenerative (CNS) diseases affecting humans (e.g. Kuru, CJD) and animals (e.g. ovine scrapie, bovine BSE), all caused by a common class of agents, Prions. The link between TSEs and polygeline lies in its precursor, the gelatine which derives from bovine bones; bones may be at risk due to adjacent CNS (skull and vertebrae) contamination. In this experiment the main steps of the manufacturing process of polygeline were validated in order to see if the process is able to reduce the risk of iatrogenic transmission of the infectious agent, if present. It is the first time that results of a validation study on a gelatine-derived product have been published. Three steps of the process were validated separately: in step 1, gelatine was subjected to three alternative autoclaving schedules (1A: 121°C for 1.5 h; 1B: 121°C for 3 h; 1C: 133°C for 40 min).

References
Step 2 was the crosslinking and distillation phase, and Step 3 the final sterilization at 121°C for 45 min. The hamster-adapted 263K strain of scrapie was used as the TSE model. The infective spike was added to each material before being processed and titrated in hamsters. Each assay was performed in duplicate, and animals were monitored for 1 year. The initial hamster-titrated infectivity of the spike resulted in 10 9.0 LD50/2 ml. From the preliminary results of the experiment, only based on symptomatology (histological results of all brains expected till February 1999), the average step-clearance of infectivity (mean of two replicates) was (LD50/2 ml): 10 6.0 (1A), 10 6.9 (1B) and ≥10 7.4 (1C), 10 2.4 (2) and 10 4.6 (3). It is clear that heating the gelatine (step 1) was very effective in reducing the infectivity of TSE agents. Taking also into account that the initial experimental contamination level adopted was extremely high, that raw materials used in the real production are carefully selected from a BSE-free country (USA) and exclude the skull and spinal cord, that the starting material -gelatine -is already produced by BSE-reducing procedures, and that steps 2 and 3 also contribute to lowering the infectivity, if any, it may be concluded that the polygeline manufacturing process is capable of inactivating BSE agents to a very high extent. I In nt tr ro od du uc ct ti io on n: : It was noticed that severe persistent pruritus was a common complaint in patients attending our nurse-led ICU follow up clinic. Pruritus is a known adverse effect after hydroxyethyl starch (HES) infusions [1]. We therefore undertook this retrospective study to clarify any association between pruritus and HES infusions.

P150 Persistent pruritus after hydroxyethyl starch (HES) infusions in critically ill patients
M Me et th ho od d: : Questionnaires were sent to all surviving patients who, over a 6 month period, had been on ICU for greater than 24 h (n = 100). The 19 questions covered a wide range of areas including general well-being, quality of life, mood and memories of intensive care. Two questions asked about itching. Respondents complaining of pruritus and non respondents were telephoned. Standardised questions were asked to identify incidence, severity, duration, triggering or relieving factors and the parts of body affected. For patients, the volume of HES received in the ICU was identified from ICU charts. Statistical analysis was by Mann-Whitney U test, with significance determined by P < 0.05. R Re es su ul lt ts s: : Details were obtained from 73 patients. 34% had experienced pruritus since their discharge from ICU. Of these 44% had severe persistent pruritus, which had not resolved with conventional treatments. In patients with pruritus, the total volume of HES infused ranged from 0-27350 ml, (median 2000 ml), infused over a mode of 2 days. The 'non pruritus' group, total HES volume ranged from 0-13350 ml, (median 500 ml), infused over a mode of 1 day. There was a significant relationship between the volume of HES and the occurrence of pruritus (P = 0.003).
C Co on nc cl lu us si io on n: : This retrospective study shows that HES infusions may be associated with persistent pruritus. This may seem a trivial problem after a life-threatening illness, but our experience suggests that it significantly detracts from quality of life in survivors. R Re es su ul lt ts s: : There were no significant intergroup differences in any of the hemodynamic variables MAP, CVP and HR. There was no significant difference among the three groups in the plasma viscosity after surgery. Mediastinal blood flow increased significantly in the HES group 2-6 h after surgery (P < 0.05) compared with the geline and Ringer's group. The two colloid groups needed significantly less fluid compared with the Ringer's group (P < 0.05). There were no differences in diuresis. There were no significant intergroup differences in the oxygenation index. The need for postoperative ventilatory support did not vary between the three groups. C Co on nc cl lu us si io on n: : This randomized comparison of two colloid and a Ringer's solution fluid regimens after cardiac surgery shows that there is no difference in hemodynamic stability, plasma viscosity, oxygenation index and duration of intubation. The HES group has a significantly higher blood loss. The Ringer's group has a significant higher volume uptake. The colloid-free regimen did not affect the pulmonary function.

Reference
The colloid-free Ringer's solution regimen is clinically fully acceptable and economically more favourable than the two colloid fluid regimens studied. The controversy regarding the use of crystaloids or colloids for fluid resuscitation of critically ill patients continues. Fluid remaining in the vascular compartment would have an obvious advantage. We used bioimpedance, a technique allowing assessment of body cell mass and extracellular water, to compare NS and HES, a relatively new colloid.

P152 Extracellular fluid variations during a fluid challenge: a comparison of normal saline (NS) and hydroxyethyl starch (HES) in stressed patients
M Me et th ho od ds s: : Twenty-two critically ill patients requiring a fluid challenge were randomized to receive either 500 ml of NS or 500 ml of HES 10% (Fresenius, Germany). Vital signs (heart rate, systolic blood pressure, central venous pressure (CVP) and urine output) were noted before and immediately after the challenge. Bioimpedance changes, using a tetra-polar system working on 800 microamperes and 50 Khz (BIA-109), Ackern) were measured before and after the fluid challenge. Body cell mass (BCM) and extracellular water (ECW) were then derived. Results are expressed as the mean ± SD. R Re es su ul lt ts s: : Ten patients (mean age 57 ± 18.6 years) received HES 10% and twelve (mean age 56.9 ± 13.9 years) received NS. There were no significant differences between the two groups regarding pre-and post-challenge hemodynamic parameters, in particular change in CVP. Bioimpedance measurements before and after fluid challenge were as follows: C Co on nc cl lu us si io on n: : We showed that there is an increase in extracellular water in critically patients receiving a fluid challenge with normal saline but not with HES. This could indicate a beneficial effect of HES on extravascular extravasation of water in stressed patients. Hydroxyethyl starches (HES) are widely used in volume replacement therapy. One point of concern is tissue storage after repetetive dosing [1]. A controlled, multi-dose study was performed in 48 female rats. Daily

Time after HES infusion [h] Plasma Concentrations [mg\mL]
On Day 1 On Day 10 LPS-infusion resulted in hypodynamic shock after 60 min with no intergroup differences. Resuscitation with HYPER improved the mesenteric and specifically the mucosal circulation, whereas ISO was ineffective in this respect. HYPER also tended to improve CO while this effect failed to gain statistical significance (P = 0.11).
C Co on nc cl lu us si io on n: : Volume resuscitation with hypertonic colloid proved superior to isotonic colloids to support intestinal and in particular mucosal perfusion during hypodynamic septic shock. The results indicate that hypertonic colloids might be of special value to support mucosal perfusion and thereby possibly barrier function in sepsis.  C Co on nc cl lu us si io on n: : In our study, dopamine caused a significant diuresis without significant changes in creatinine or urea levels. Low-dose dopamine was not associated with a higher incidence of adverse cardio-circulatory reactions. We observed no protective effect on the incidence of ARF and our results indicate that low-dose dopamine may be associated with unfavourable outcomes, in particular higher mortality.   O Ob bj je ec ct ti iv ve e: : In patients after cardiac surgery dopexamine is known to increase both global and hepato splanchnic blood flow. Sepsis per se and the infusion of noradrenaline may alter the response to the onset of dopexamine. Therefore we determined the changes in global and regional hemodynamics in patients with septic shock. P Pa at ti ie en nt ts s/ /m me et th ho od ds s: : Twelve patients with septic shock were studied. All patients had a cardiac index (CI) ≥3 l/min/m 2 and needed noradrenaline ≥ 0.04 µg/kg×min -1 to maintain mean arterial pressure (MAP ≥60 mmHg). In addition to routine systemic hemodynamics and gas exchange we inserted a balloon-tipped Swan Ganz catheter into a hepatic vein to determine splanchnic blood flow (Qspl), hepatic venous pressure (HVP) and the hepatic venous occlusion pressure (HVOP) as an estimate of portal venous pressure. Splanchnic blood flow was measured using the steadystate indocyanine green (ICG) infusion technique. Measurements were done before, during and after dopexamine infusion. Data were always obtained after at least 90 min of hemodynamic steady-state. Dopexamine was titrated (1-4 µg/kg×min -1 ) to obtain a 30% increase in CI.

P158 Effect of a dopexamine induced increase in cardiac output on splanchnic hemodynamics in septic shock
R Re es su ul lt ts s: : See Table. C Co on nc cl lu us si io on n: : The dopexamine induced increase in Qspl paralleled that of CI. A preferential effect on splanchnic circulation could not be detected. The increase in Qspl was due to a decreased prehepatic resistance. O Ob bj je ec ct ti iv ve e: : Septic shock is characterized by enhanced hepatosplanchnic blood flow resulting from increased metabolic activity. This hypermetabolism may lead to mismatch of regional O 2 supply and demand reflected by increased CO 2 -gradients and lactate/pyruvate ratios. In patients after cardiac surgery dopexamine is known to increase splanchnic perfusion. Therefore, we studied the effect of dopexamine on regional CO 2 -gradients and energy balance in patients with septic shock. P Pa at ti ie en nt ts s a an nd d m me et th ho od ds s: : Twelve patients with septic shock were studied. Cardiac index (CI) was ≥3 l/min/m 2 and noradrenaline ≥0.04 µg/kg×min -1 was infused to maintain mean arterial pressure (MAP) ≥60 mmHg. In addition to routine systemic hemodynamics and gas exchange we inserted a Swan Ganz catheter into a hepatic vein to measure splanchnic blood flow (Qspl) using primed continuous infusion of indocyanine green (ICG) dye. Moreover, we assessed splanchnic lactate uptake (Fick principle), hepatic venous lactate/pyruvate ratio as well as PCO 2 (PCO 2 hv), splanchnic O 2 delivery (DO 2 spl) and consumption (VO 2 spl). The gastric mucosal PCO 2 (PCO 2 gm) was determined via a nasogastric tube. Measurements were done before, during and after dopexamine infusion. Data were obtained after 90 min of hemodynamic steady-state. Dopexamine was titrated (1-4 µg/kg×min -1 ) to obtain a 30% increase in CI.
R Re es su ul lt ts s: : See Table. C Co on nc cl lu us si io on n: : The unpredictable changes in the metabolic state and regional PCO 2 gradients after a dopexamine-induced increase in DO 2 spl, underscores the independent response of hepatosplanchnic perfusion and metabolism to therapeutic interventions. The study was prospective and double-blind. Fourteen patients with ileostomies were randomised into treatment (n = 7) and placebo groups (n = 7). The stomas were exposed to the air for a period of 20 min whilst the laser Doppler scanner (Moore Instruments, Axminster, Devon, UK) was positioned above the patient at a distance of 32 cm. A laser scan was then made, and the stoma outlined on the photographic image. This equated to over 2500 individual perfusion measurements on the corresponding perfusion image, allowing calculation of mean perfusion units (PUs) within the stomal mucosa. Heart rate and mean arterial pressure were recorded. An intravenous infusion of either dopexamine (2 µg/kg/min) or of a placebo was then commenced and after 30 min the recordings were repeated. The infusion was then stopped and a final set of recordings made after 30 min. The results were analysed using the Mann-Whitney test for non-parametric data. An increase in splanchnic perfusion in general or gastric mucosal perfusion in particular following oesophageal resection may potentially reduce the incidence of anastomotic leaks and strictures as these complications are thought to be caused by hypoperfusion and consequent tissue hypoxia at the gastric end of the oesophagogastric anastomosis. This study assessed the effect of dopexamine on gastric tube and jejunal mucosa pHi measured tonometrically following oesophagectomy.

Pre infusion
M Me et th ho od ds s: : Twelve patients undergoing oesophageal resection for carcinoma and reconstitution of gastrointestinal continuity using a gastric tube were randomised into dopexamine and control groups. During surgery tonometer balloons (Tonometric Division, Instrumentarium Division, Helsinki, Finland) were placed 5 cm distal to the anastomosis within the stomach and 10 cm during the duodeno-jejunal flexure within the jejunum. These were con-nected to separate 'Tonocap' analysers (Datex, Helsinki, Finland). 24 h following surgery all the patients were sedated, ventilated and cardiovascularly stable. Three measurements of heart rate, mean arterial pressure, central venous pressure as well as gastric and jejunal pHi were made at 30 min intervals prior to the commencement of an intravenous infusion of either dopexamine (2 µg/kg/min) or of a placebo. Four further sets of measurements were made at 30 min intervals during the infusion, and after 2 h it was stopped and three measurements over the next 90 min were made. The results were analysed using the Mann-Whitney test for non-parametric data.
R Re es su ul lt ts s: : There were no significant changes in systemic arterial or central venous pressure in either group during the study. However, in the dopexamine group there was a significant increase in mean (SD) heart rate from 85 (12) to 104 (10) beats per minute during the infusion and a subsequent fall to 94 (10) beats per minute after its cessation (in both cases P < 0.005). There were no significant changes in either gastric or jejunal pHi during dopexamine or placebo infusion (in all cases P > 0.05).
C Co on nc cl lu us si io on n: : Dopexamine hydrochloride does not increase gastric tube pHi following oesophagectomy. Furthermore there is no evidence from this study that dopexamine is capable of influencing jejunal mucosal perfusion, and its potential role not only in protecting gastrointestinal anastomoses but also in reducing mortality due to MODS by directly influencing splanchnic perfusion is not supported by the findings of this study.  R Re es su ul lt ts s: : Colonic ischaemia was noted in 9 (30%) patients based on microscopic findings. Endoscopy alone had a sensitivity of 55.5%. There was a significantly lower incidence of colonic ischaemia in patients receiving dopexamine compared to placebo (P < 0.05). One death resulted from colonic infarction in the placebo group 11 days post-operatively. There was increased MPO and MCT expression in patients with histological evidence of ischaemia (P < 0.05). iNOS staining within the vascular (P = 0.001) and lamina propria (P < 0.05) components of the mucosa was also significantly greater. No association was found with eNOS. C Co on nc cl lu us si io on ns s: : Peri-operative dopexamine infusion confers a degree of protection to colonic mucosa following aortic surgery, possibly through an anti-inflammatory effect. Baxter Healthcare Corp., Irvine, USA) and a polarographic intramyocardial oxygen catheter (Licox; Kiel, Germany). Documentation of standard parameters followed every 15 resp. 60 min. intra-and postoperative until extubation. Retrospectively patients were devided into two groups (2×12) by the amount of postoperative colloid volume replacement: Group 1: <750 ml colloids during the first 3 postoperative hours; Group 2: >1000 ml colloids during the first 3 postoperative hours. Statistics were done by using Mann-Whitney-U and Friedman-test.

P164 Endothelin-1 (ET-1) blockade improves mesenteric perfusion in a porcine low cardiac output model
R Re es su ul lt ts s: : There were no significant differences between group 1 and 2 with regard to age, ejection fraction, duration of extracorporal circulation, number of bypasses, arterial or mixed venous blood gas analyses, arterial or pulmonary arterial hemodynamics, lactate, heart rate and central venous pressure. Regional oxygenation differed significantly between the groups. Group 1 (<750 ml/3 h) showed a small aiDCO 2 during the first 5 postoperative hours. Afterwards splanchnicus perfusion impaired (aiDCO 2 >20 mmHg). In group 2 (>1000 ml/3 h) an initially higher aiDCO 2 was lowered by volume therapy. Group 2 developed a significantly lower postoperative increase in intramyocardial oxygen then group 1. More than 5 h after extracorporal circulation there was a significantly increasing need of epinephrine in group 2 and cardiac index was lower in group 2 without reaching significance. C Co on nc cl lu us si io on n: : Volume replacement after coronary bypass surgery should be monitored by gastric tonometry since hemodynamic parameters are less sensitive. High volume replacement without need (aiDCO 2 <20 mmHg) can improve splanchnicus perfusion but might impair myocardial oxygenation and myocardial function. Measurements of systemic perfusion are often extrapolated clinically to reflect regional perfusion, including the gastrointestinal organs. This extrapolation may introduce errors in the evaluation of hemodynamic status. Furthermore, the complex variable of perfusion involves movement of blood and erythrocytes as well as the exchange of carbon dioxide and oxygen.

P166 Evaluation of intestinal perfusion monitoring techniques
A Ai im ms s: : To investigate the relationship between clinically available techniques of measuring mucosal perfusion in relation to mesenteric and central blood flow during acute circulatory failure.
M Ma at te er ri ia al ls s a an nd d m me et th ho od d: : Thirteen fasted, anesthetized (pentobarbital) mechanically ventilated, normovolemic pigs (28-35 kg) were instrumented to monitor cardiac output (CO), portal blood flow (QPV, Transonic Systems), jejunal, mucosal laser-Doppler flowmetry (LDF, Perimed AB), jejunal CO 2 -tonometry (TONO, Tonocap, Datex Instr) and jejunal, mucosal oxygen tension (tO 2 , Licox, GMS). Acute reduction of CO by 40% from baseline was established by intrapericardial infusion of dextran and maintained for 90 minutes. Correlations between monitored variables were analyzed by ANOVA and linear regression (*P < 0.05) and differences were analyzed by Wilcoxon's test ( § P < 0.05). R Re es su ul lt ts s: : The best regressions coefficients were found between variables relating to measurements of movement of volume (QPV) or erythrocytes (LDF). Second to best regressions were obtained for TONO (measuring the exchange of CO 2 ). Notably, tPO 2 (measuring the exchange of O 2 ) did not correlate to variables of flow or CO 2 exchange. C Co on nc cl lu us si io on n: : In the setting of acute circulatory failure in pigs, cardiac output approximates mesenteric as well as intestinal mucosal perfusion. Importantly, the mucosal oxygen tension might vary independent from flow, which probably reflects the complexity of the counter current circulation within the mucosa. Oxygenation, being the pivotal variable determining tissue function, is thus not assessed even by techniques specifically directed towards the mucosal circulation. B Ba ac ck kg gr ro ou un nd d a an nd d a ai im ms s: : Intestinal NO production has been attributed a central role in the maintenance of the intestinal mucosal barrier. Hypofunction of this barrier has been suggested to be one important factor behind the initiation of the multiple organ dysfunction syndrome. Jejunal NO formation, as we previously have reported, has been shown to be impaired during mucosal hypoperfusion [1]. This study was undertaken to investigate if the impaired jejunal NO levels could be due to restricted mucosal availability of NO-synthase substrates, i.e. oxygen and/or L-arginine.
M Me et th ho od ds s: : Chloralose-anesthetized pigs (n = 18) were prepared for jejunal intraluminal perfusion with 150 mM NaCl or 3 mM L-arginine solution and then subjected to cardiac tamponade. Jejunal mucosal NO formation was measured with a tonometric technique. Mesenteric blood flow was measured as portal blood flow and mucosal perfusion was measured by laserdoppler flowmetry. Regional oxygen consumption was calculated from blood samples.
R Re es su ul lt ts s: : Cardiac tamponade reduced jejunal NO formation (-52%), mesenteric oxygen delivery (-75%), oxygen consumption (-39%) and mucosal laser doppler flow (-43%). Oxygenation of the jejunal intraluminal perfusate completely restored the intestinal NO levels within 30 min. Presence of L-arginine was without effect. C Co on nc cl lu us si io on n: : The study indicates that oxygen rather than L-arginine is the rate limiting factor for mucosal NO production during reduced splanchnic perfusion. O Ob bj je ec ct ti iv ve es s: : A Neonatal Tonometer (5 French) using saline capnometry has been developed. We compared these tonometers invitro, using 0.9% saline (NS) and phosphate buffered saline (PBS) as the CO 2 vehicle, along with a Tonocap (14 F) against a set of known PCO 2 's in a saline solution.

Åneman
M Me et th ho od d: : A sealed equilibration chamber containing 0.9% Saline was maintained at 37°C and the dissolved CO 2 was kept at constant pCO 2 's of 2.5, 5, 7.5, 10 kPa using a Paratrend 7 probe (Biomedical Sensors). Two Neonatal Gastric tonometers (Tonometrics) were positioned in the chamber along with a Tonocap monitor (Datex). NS was the CO 2 vehicle in the first tonometer, PBS (pH 6.0) in the second, and recirculating gas tonometry in the Tonocap. 20 consecutive measurements were taken, each after 60 min equilibration periods, from each of the tonometers at pCO 2 's of 2.5, 5, 7.5, 10 kPa and processed in the IL BGE blood gas analyser. Data was analysed by linear regression and Bland-Altman plots. C Co on nc cl lu us si io on n: : Recirculating gas tonometry is undoubtedly the best mode of tonometry. Whilst we await its development for neonates either NS or PBS may be used. We suggest that correction factors specific to each unit's blood gas analyser should be calculated before appropriate comparison can be made between the arterial pCO 2 and the Neonatal tonometer's pCO 2 . B Ba ac ck kg gr ro ou un nd d: : Gastric mucosal perfusion can be assessed tonometrically by measuring the gastric intra-mucosal pH (pHi) and its ability to predict outcome in the critically ill and following major surgery has been demonstrated by several previous studies. It has been suggested that the CO 2 gap (tonometer pCO 2 ) may provide a more sensitive measurement of mucosal hypoxia than pHi. Post-operative anastomotic leak and stricture following oesophageal resection and restoration of gastrointestinal continuity with a pro-peristaltic gastric tube has a multi-factorial aetiology. However, gastroplasty involves division of short gastric, left gastric and left gastroepiploic vessels and the consequent hypoperfusion and tissue hypoxia at the gastric end of the oesophago-gastric anastomosis is thought to be the most important causative factor. This study employed the new technique of automated gas tonometry to measure both gastric CO 2 gap and pHi following oesophagectomy to test the predictive ability of the technique for anastomotic complications.
M Me et th ho od d: : Gastric tonometers (Tonometric Division, Instrumentarium Division Helsinki, Finland) were placed in the gastric tube of 30 consecutive patients undergoing oesophageal resection and pro-peristaltic tubular gastroplasty based upon the right gastroepiploic and right gastric arteries. These were connected to a 'Tonocap' analyzer (Datex-Engstrom Division, Instrumentarium Corporation, Helsinki, Finland) which automatically samples gas from the tonometer balloon and measures the CO 2 concentration within it. In conjunction with simultaneously taken arterial blood samples the gastric CO 2 gap and pHi were calculated at 12 hourly intervals up to 48 h post-operatively. Those patients who survived were followed for 3 months and all post-operative complications recorded. Statistical comparison was made using the Mann-Whitney test for non-parametric data.
R Re es su ul lt ts s: : Eleven patients suffered an anastomotic leak or benign stricture post-operatively, whilst five others suffered a life threatening complication not related to the anastomosis, of whom two survived. Because of balloon failure or re-operation within 48 h of initial surgery data was not available for one patient from each of the complication and no complication groups. Mean (SD) CO 2 gap and pHi over the first 48 post-operative hours were 1.7 kPa (0.8) and 7.26 (0.06) in the no complication group and 3.5 kPa (1.4) and 7.18 (0.09) in the complication group, respectively. The difference in CO 2 gap between the two groups was more significant than in pHi (P < 0.005 and P < 0.05). A mean CO 2 gap of 2.5 kPa or above had a sensitivity of 82% and a specificity of 70% for predicting anastomotic complications. The CO 2 gap was a better predictor of outcome than the pHi (<7.22 for predicting complications), with areas under their respective ROC curves of 0.847 and 0.684. C Co on nc cl lu us si io on n: : Gastric tube CO 2 gap and pHi are easily measured post-operatively using recirculating gas tonometry. Mean CO 2 gap was higher and pHi lower over the first 48 h following surgery in those patients in whom an anastomotic complication subsequently developed than in those in whom it did not. The CO 2 gap proved to be a better predictor of complications than the pHi. These findings confirm the suggestion that the CO 2 gap may be a more useful clinical tool than the pHi and that measures to improve gastric tube CO 2 gap post-operatively might reduce the incidence of anastomotic failure.
mucosal perfusion, and that its measurement can predict for poor outcome in the critically ill. There is increasing evidence that not only in this group of patients but also following upper gastrointestinal surgery that the early introduction of enteral nutrition may reduce morbidity and mortality and increase enteric mucosal tissue perfusion. Theoretically this effect may be especially desirable following oesophagectomy and oesophagogastric anastomosis as gastric blood flow is compromised following gastroplasty. However, the measurement of gastric pHi using the tonometric method is thought to be confounded during infusion of enteral feed by the release of carbon dioxide from the feed itself following enzymic digestion. This study assessed the effect of a standard enteral feed upon both gastric and jejunal pHi measured using gas tonometry when delivered via a feeding jejunostomy.
M Me et th ho od d: : Nineteen patients undergoing oesophageal resection for carcinoma and reconstitution of gastrointestinal continuity using a gastric tube were studied. During surgery tonometer balloons (Tonometric division, Instrumentarium Division, Helsinki, Finland) were placed 5 cm distal to the anastomosis within the stomach and 10 cm from the duodeno-jejunal flexure within the jejunum. The jejunal tonometer was placed alongside a standard 8 F Foley feeding jejunostomy tube. The tonometers were connected to separate 'Tonocap' analysers (Datex, Helsinki, Finland).
Five days following surgery all the patients had left the intensive care unit and had returned to the surgical ward and were being fed (Fresubin Standard, Fresenius Ltd, Runcorn, Cheshire, UK) via the jejunostomy tube (mean rate 108 ml/h). The feed was stopped for a minimum of 6 h and then both jejunal and gastric pHi was measured using a simultaneously taken arterial blood gas sample. The feed was then recommenced and after 2 h the measurements were repeated. The results were analysed using the Mann-Whitney test for non-parametric data.
R Re es su ul lt ts s: : Prior to the commencement of feeding mean (SD) jejunal and gastric pHi were 7.44 (0.06) and 7.37 (0.08) respectively. Following 2 h of enteral nutrition jejunal and gastric pHi had fallen to 7.26 (0.09) and gastric pHi to 7.29 (0.12). These falls were both significant (P < 0.005 and P < 0.05, respectively). C Co on nc cl lu us si io on n: : Standard enteral nutrition delivered via a feeding jejunostomy appears to cause a fall in tonometrically measured jejunal pHi. That this may at least in part reflect a fall in mucosal blood flow rather than have been caused by the release of carbon dioxide from the feed is supported by the finding that gastric pHi also falls despite the fact that no feed was introduced into the stomach. That an enteric reflex may be responsible for this finding seems likely although its significance with regard to its effect upon anastomotic perfusion remains unknown. I In nt tr ro od du uc ct ti io on n: : The intramucosal pH (pHi) is a sensitive and early parameter of various shock states. Its prognostic and therapeutic value has been demonstrated. Tonometry relies on the measurement of intragastric pCO 2 via a nasogastric probe and the arterial bicarbonate. There are several shortcomings of the tonometric method (e.g. handling and measurement errors, and the need for a long equilibration time (30-90 min)). Therefore, we evaluated a modified pH-glass electrode for fast and continuous pCO 2 -monitoring in the stomach.
M Me et th ho od ds s: : Our in vitro measurements were performed using a special designed pH-metry glass electrode with an outer diameter of 4.5 mm (GK2801C;Radiometer) covered with a thin gas permeable teflon membrane. Thus, CO 2 may easily diffuse through the membrane and induces changes of the pH of an interspersed electrolyte solution. At first, the membrane and the membrane covered electrode were tested for chemical and mechanical stability in aggressive and acidic fluids. Secondly, the precision of two one-point calibrated electrodes to measure the pCO 2 in the range of 20 to 250 mmHg was tested. For each of five given pCO 2 -levels a set of five measurements was done. Thirdly, the response time of two electrodes to reach 90% of the maximum (t 90 ) was tested by exposing the electrodes rapidly to two different solutions with a pCO 2 of 28.9 and 85.9 mmHg, respectively. R Re es su ul lt ts s: : The teflon membrane has proved to be stable against 0.1N hydrochloric acid, gastric and biliary secretions adjusted to pH 1, and mechanical irritations. In acidic fluids a linear relationship between the measured pCO 2 and the defined pCO 2 for both electrodes was observed. The slope of the regression line was y=24.43x+7.64 (r = 0.99, n = 25) and 27.57x+4.64 (r = 0.99, n = 25) respectively. The deviation from the line of identity was only caused by the one point calibration. The reponse time t 90 of the electrode was 19.5 ± 1.38 s and 25.5 ± 2.42 s (± SD), respectively. C Co on nc cl lu us si io on n: : This teflon membrane covered modified pH-glass electrode offers a fast, real time, and continuous measurement of the pCO 2 in the acidic environment of the gastric lumen.
M Me et th ho od ds s: : After Ethical committee approval and patient assent, two tonometric catheters were inserted into patients who after an overnight fast, were given 30 ml per hour of water for 4 h and following a 1 h rest, 30 ml per hour of enteral feed for 4 h. All patients received iv ranitidine. The PrCO 2 was measured hourly using both techniques.
R Re es su ul lt ts s: : Eight neurosurgical intensive care patients were studied (mean age 52 years, SD 15.3 years). All patients were stable and had no significant changes in cardiovascular or blood gas parameters during the study.
Bland-Altman analysis showed the mean bias between TCS and Tonocap PrCO 2 and -1.85 kPa with a precision of ± 3.49 kPa.
There was no significant difference between the effects of feeding with the two techniques.
C Co on nc cl lu us si io on n: : Enteral feeding has no effect on PrCO 2 in neurosurgical patients. Saline tonometry under reads compared to the Tonocap similar to that of general intensive care patients. I In nt tr ro od du uc ct ti io on n: : Measurement of pHi or the mucosal-arterial PCO 2 gap is advocated to detect splanchnic ischaemia and covert shock. Nasogastric feeding may significantly affect these measurements. We used a previously described animal model [1] to evaluate the effect of enteral feeds on the luminal PCO 2 response to intermittent splanchnic ischaemia.
M Me et th ho od ds s: : Adult male Wistar rats (285-425 g) were anaesthetised with sodium pentobarbitone 60 mg/kg i.p. and ventilated with 100% oxygen and isoflurane via tracheostomy to a PaCO 2 of 30-40 torr. Distal aortic pressure was monitored continuously. A sensor (Paratrend 7, Diametrix Medical Inc., Bucks, UK) was inserted into the ileal lumen to record PCO 2 measurements every 2 s. Four rats received no feeds (controls) whilst in another four rats an ileal cannula was inserted and feed (Nutrison, Nutricia, Zoetermeer, Holland) infused at 3 ml/h. In each rat, five twominute episodes of aortic hypotension were induced to a mean pressure of 30 mmHg by intermittent elevation of a silk sling placed around the proximal aorta.
R Re es su ul lt ts s: : See Table. Feeds significantly elevated the mean baseline luminal PCO 2 , and delayed and blunted the PCO 2 increases (∆PCO 2 ) in response to transient ischaemia.
C Co on nc cl lu us si io on n: : Assuming no differences in PaCO 2 in both groups, the data suggest that enteral feeding increases the baseline mucosal-arterial PCO 2 gap and reduces baseline pHi. It may also impair the detection of splanchnic ischaemia by delaying and blunting the responses of these indices to reduced mucosal perfusion. I In nt tr ro od du uc ct ti io on n: : Real time assessment of gut luminal PCO 2 is possible with rapidly responsive tissue CO 2 sensors [1]. The impact of the presence of feeds in the gut on the rapidity of response of the sensor to a change in mucosal CO 2 tension has not been evaluated.

Reference
M Me et th ho od ds s: : The speed of onset of response and the 90% response time of a commonly used tissue gas sensor the Paratrend 7 (Diametrics Medical, UK) to a change in ambient CO 2 tension were compared in normal saline (control) and an enteral feed solution (Nutrison, Nutricia, Zoetermeer, Holland). Probe onset and 90% response times were determined for a step up and step down change in CO 2 tensions in saline and feed solutions by bubbling the following three pairs of gases A) 2% CO 2 and 10% CO 2 B) 10% CO 2 and 5% CO 2 and C) 5% CO 2 and 2% CO 2 through these solutions maintained at 37°C in a bubble tonometer. After calibration, the sensor was equilibrated in saline bubbled with the first gas of each pair. After equilibration the second gas of each pair was bubbled through the solution. This was repeated for a total of six equilibrations between each pair of gases. The experiment was then repeated with the feed solution.
R Re es su ul lt ts s: : See Table   C Co on nc cl lu us si io on n: : The presence of enteric feed significantly slows down the onset time and response time of the sensor to a change in ambient CO 2 tension. Altered viscosity and CO 2 binding by the feed are possible mechanisms for the altered response of the sensor.
The reduction in response time may impact on the ability of tissue CO 2 sensors to provide accurate real time data in clinical practice. Step up 39 ± 9 65 ± 10* 188 ± 25 307 ± 42* Step down 30 ± 6 52 ± 9* 191 ± 18 297 ± 20* Overall 34 ± 8 59 ± 11* 189 ± 21 302 ± 32* The data are presented as mean ± SD (*P < 0.001) I In nt tr ro od du uc ct ti io on n: : The aim of this prospective non intervention study is to evaluate if the analysis of some perfusional indexes, as gastric intramucosal pH (pHi, U) and plasma disappearance rate of indocyanine green (PDR dye, %/min), may be useful for prognostic evaluation in patients with MODS.

P175 Splanchnic and haemodynamic data as prognostic indexes in MODS patients
M Ma at te er ri ia al l a an nd d m me et th ho od ds s: : Eighty-four medical or surgical patients, with MODS ( mean age 51, SD 17; mean SAPS II (1 st day) 56, SD 9), were studied. After 6 h of ICU stay, a gastric tonometer, a 7.5 pulmonary artery catheter and a 4 F femoral artery catheter were inserted. The vascular catheters were connected to 'COLD System', an integrated monitoring system which uses the double indicator technique and studies hepatic perfusion, by analysis of PDR. All patients were in CMV and received ranitidine. The haemodynamic management was realized in order to optimize cardiac output (CO, l/min/m 2 BS) and systemic oxygen delivery (DO 2 , ml/min/m 2 BS). All data were recorded at the beginning of the study (T0) and after 6 (T1), 12 (T2), 24 (T3) and 36 (T4) hours. Statistical analysis of data was performed using Manova Test, considering the significant differences in the times of study between survivors (S) and non-survivors (NS) and analysing the variance of repeated measures. Levels of P < 0.05 were accepted. R Re es su ul lt ts s a an nd d c co on nc cl lu us si io on ns s: : 40 (47.6%) patients died. Some data are shown in the Table (as mean and (SD); S vs NS: *P < 0.0001; $ P < 0.005; T vs T0: § P < 0.05).
In this group of patients, a precocious splanchnic hypoperfusion seems to be the main prognostic factor. In NS group, gastric intramucosal acidosis is present in the early period of study and it is possible to notice a continuous worsening of liver perfusion. According to this point of view, perfusional parameters may give more prognostic informations than systemic data. I In nt tr ro od du uc ct ti io on n: : Gut mucosal ischemia can initiate a systemic inflammatory response sometimes leading to multiple organ failure. The adequacy of splanchnic perfusion during major abdominal surgery can be evaluated by an easy, non invasive, new method : gastric air tonometry. Air tonometry is an important technical advance which eliminates errors involved in saline tonometry.
The aim of our study was to investigate wether there was a relationship between a perioperative tonometric parameter and clinical outcome.
M Me et th ho od ds s: : 27 patients, ASA 1-3, admitted for major abdominal surgery (hepatectomy, pancreatoduodenectomy, colorectal resection) were prospectively studied between March 98 and October 98. After induction of anesthesia, intramucosal PCO 2 (PrCO 2 ) was measured by a gastric tonometer placed in the stomach and then connected to a TONOCAP ® (Tonometrics-Datex-Engstrom). PCO 2 gap was measured immediately after tracheal intubation and until discharge of the SICU at H24 postoperatively. Post operative complications were recorded during the entire hospital stay. Statistical analysis used FISCHER's Exact Test.
R Re es su ul lt ts s: : 19 out of 27 patients suffered complications (bleeding, SIRS, sepsis, MOF, pancreatitis, wound infection, hepatic failure, anastomotic leakage) leading to death for two of them. Fifteen out of these 19 patients had a PCO gap >15 mmHg during surgery. The FISCHER's Exact Test (P < 0.002) was conclusive for both group. According to these results, PCO 2 gap can predict complications with a sensibility of 78.5% and a specificity of 88%.
C Co on nc cl lu us si io on n: : During abdominal surgery, the assessment of splanchnic perfusion can be easily achieved with air tometry ; a PCO 2 gap >15 mmHg seems to be predictive of postoperative complications.

Complications No complication
Gap ≥15 15 1 Gap <15 4 7 D Di is sc cu us ss si io on n: : Decreasing tissue perfusion causes hypoxia and then acidosis that provokes a cellular damn, increasing of cellular permeability with loss of barrier function of gut mucosa. This induces the liberation of some substances, such as endotoxins, which start the inflammatory cascade of TNFa, Il-1, Il-6, Il-8. Moreover, another way to induce the formation of toxic substances, in the presence of ischemia followed by riperfusion, is the activation of purine metabolism with activation of xantine-oxydase (XO) and consequent production of the anion superoxydodismutasis, that, in the presence of iron (Fenton reaction) causes the formation of ossidryl ion, very dangerous for the organism. The hemodynamic response of these two cases (high CO, low SVR, pHa, pHi and increased blood lactate) and the increasing of Il-6 are not explanable only in terms of hypoxia and it could be supposed that these changes probably are due to a septic state, caused by the substances liberated from the hypoxic splanchnic tissue. This experimental model could be useful in the comprehension of physiopathology of hypoxia and perhaps of septic shock and, in some way in the experimentation of new drugs against the effects of hypoxia.

P178 Continuous assessment of colonic perfusion during abdominal aortic reconstruction using a modified Paratrend 7™
HI Rashid, N van Heerden*, PR Taylor  Under general anaesthetic, a modified Paratrend 7™ probe was inserted transanally to the rectosigmoid junction. Continuous intraoperative analysis was compared to intermittent intraoperative inferior mesenteric vein (IMV) sampling for pH, P CO2 and P O2 was observed on aortic cross clamping and declamping. The 95% limits of agreement with IMV pH, in patients with complete bowel preparation were 0.16 and -0.1 for the calculated intramucosal pH and 0.72 and -0.48 for the luminal pH. The estimated bias for the calculated pHi was -0.03. Results were directly affected by the condition of bowel preparation C Co on nc cl lu us si io on n: : The modified Paratrend 7™ effectively detects changes in colonic perfusion during abdominal aortic reconstruction. However, complete bowel preparation is essential and modifications may be required to increase its precision. A decrease of liver blood flow leads to a dysfunction of hepatocytes and Kupffer-cells with subsequent local and systemic liberation of proinflammatory mediators [1] that may maintain SIRS and may lead to MODS [2]. There is only limited knowledge about the hepatic micro-and macrocirculation during sepsis or endotoxemia. Therefore, aim of our study was to investigate alterations in hepatic portal (PBF) and sinusoidal blood flow (SBF) during endotoxemia.

P179 Effect of endotoxemia on hepatic portal and sinusoidal blood flow in rats
In male Wistar rats endotoxemia was induced by continuous infusion of 2 mg/kg/h lipopolysaccharides (LPS) from E. coli 026:B6 immediately after baseline measurements (LPS group; n = 10). The control group (n = 10) received an equivalent volume of Ringer's solution. MAP, HR, CO, PBF and SBF were measured at baseline, and 60 min, and 120 min after induction of endotoxemia. PBF was measured using a laser-doppler flow probe that was positioned around the portal vein. SBF was detected by in vivo videomicroscopy of the left liver lobe. Statistical analysis was performed using Mann-Whitney's U-test.
MAP and CO remained at baseline values in both groups. In the LPS-group HR significantly increased. During endotoxemia PBF and SBF significantly decreased (Table).
Our results demonstrate that during early endotoxemia hepatic macro-and microcirculatory perfusion is significantly decreased despite unchanged MAP and CO. This early reduction of hepatic perfusion might be caused by an increased hepatic vessel resistance as a consequence of liberation of vasoconstrictive mediators (e.g. endothelin) or/and by a decrease in intestinal perfusion.  Table. C Co on nc cl lu us si io on n: : MOD score higher than 2 is related to serious pancreatitis. It is important to preserve different organ functions together with nosocomial vigilance with the support of biological markers just to indicate specifif antibiotherapy if sepsis is present, and surgery in early phase as a last resort.  I In nt tr ro od du uc ct ti io on n: : To determine the outcome and prognostic factors of patients with cirrhosis of the liver requiring medical intensive care P Pa at ti ie en nt ts s a an nd d m me et th ho od ds s: : All patients with chronic liver disease and cirrhosis admitted to the medical ICU between 7/95 and 6/97 were enrolled in the study. Prospectively the reason for ICU admission, acute diagnoses, presence of co-morbid illness, stage of liver disease, number and length of organ failures, daily APACHE II and TISS classification and outcome were documented. Laboratory values were drawn retrospectively from the charts. Patients with multiple ICU treatments were reviewed only for the initial admission. Contingency tables were analysed using χ 2 test, continuous variables were compared using Mann-Whitney U test.
C Co on nc cl lu us si io on n: : Among critically ill patients with cirrhosis of the liver ICU mortality was 38%, in comparison, the mortality for all ICU admissions in this period of time was 23%. APACHE II score and variables describing single or multiple organ dysfunction and pulmonary infection are excellent predictors of mortality. B Ba ac ck kg gr ro ou un nd d: : The outcome of severe hepatic necrosis following acetaminophen overdose is unpredictable and may have up to a 90% mortality. IL-10 is an anti-inflammatory cytokine which plays a pivotal role in inflammation and potentially in multiorgan failure. Elevated plasma levels of IL-10 are found in patients with acute liver failure. Polymorphisms in the promoter region of the IL-10 gene have recently been described comprising three single base-pair substitutions at positions (-1082, -819, -592) resulting in three common three haplotypes GCC, ACC and ATA. The GCC/GCC genotype is associated with higher IL-10 production, and ATA haplotype with lower production. P Pa at ti ie en nt ts s a an nd d m me et th ho od ds s: : 96 patients with severe acetaminophen hepatotoxicity requiring intensive care were studied. IL-10 gene polymorphisms were determined by sequence-specific oligonucleotide probing using a standard PCR based technique. Haplotype frequencies were compared with those of 71 racially and geographically matched controls. R Re es su ul lt ts s: : See Table. C Co on nc cl lu us si io on n: : There is no significant association between outcome or incidence of multiorgan failure in patients with acetaminophen induced acute liver failure and these three common IL-10 gene promoter haplotypes.   Table. C Co on nc cl lu us si io on n: : These qualitatively summarized results indicate the potential role of an advanced monitoring for the differentiation of infectious complications. Elevated PCT, IL-6 and s-IL-2 R levels are found in severe bacterial infections, whereas very high levels of TNF alpha and s-IL-2 R seem to be more specific for fungal infections. These findings may be a useful guide for the initiation of a specific diagnostic work up, for the induction of an adequate treatment and/or for an appropriate modification of the immunosuppressive treatment. B Ba ac ck kg gr ro ou un nd d: : Acute fatty liver of pregnancy (AFLP) and the syndrome of haemolysis, elevated liver enzyme levels, and low platelet count (HELLP) are rare but major disorders of the third trimester of pregnancy and are maybe related to pre-eclampsia. Mortality of 9-24% has been reported and complications include pulmonary oedema, adult respiratory distress syndrome, abruptio placentae, disseminated intravascular coagulation, ruptured liver haematomas, and acute renal failure increasing mortality to 50-90%. Multi-organ failure may result requiring full intensive care support. Perinatal mortality is equally high, ranging from 79 to 367 per 1000 live births, and neonatal complications correlate with the severity of maternal disease. Most presentations of AFLP and HELLP require monitoring and supportive care, however, early recognition of rarely associated complications and their appropriate treatment is of paramount importance to the survival of the mother and child.
Of the five, patient one did not survive long enough for transplantation, however, the others successfully received liver transplants (7.4 ± 6 days post caesarean section). Unfortunately patient two developed hepatic artery thrombosis and was re-transplanted, but died soon after.
The other three patients remain alive and well. Patients have been investigated for pro-thrombotic disorders, evidence of which is not present.
C Co on nc cl lu us si io on n: : We describe potentially fatal complications in five patients initially presenting with mild AFLP and or HELLP associated with pre-eclampsia with a mortality of greater than 90%. These rare complications include hepatic rupture, hepatic infarction and necrosis and veno-occlusive disease. Clinical suspicion must be high if there is evidence of hypotension, altered conscious state, metabolic acidosis, hyperlactataemia and deranged liver function. The early recognition of the changing clinical parameters of disease, multidisciplinary support, and specialist intensive care is required for the survival of this rare group of patients and their children. Patients with fulminant hepatic failure have a higher mortality rate after orthotopic liver transplantation than patients with chronic liver disease. Due to the shortage of cadaveric livers for transplantation, the concept of perfusion through a liver outside the body has recently been reintroduced in the clinical setting.
We describe a venovenous perfusion circuit with two Biomed pumps and one oxygenator connected to the patient's venous system via two hemodialysis catheters. The circuit provided adequate flow during ex vivo pig-liver perfusion in a critically ill patient with a stage 5 coma. The procedure lasted 4.5 h and was terminated when the oxygen extraction and bile production decreased, and the total bilirubin level went back up. During the period of ex vivo perfusion the patient moved all four extremities spontaneously within 30 min of perfusion, serum total bilirubin, and the serum ammonia level decreased by 50% and 60% respectively. The patient eventually developed sepsis and the therapy was discontinued.
C Co on nc cl lu us si io on n: : For patients with acute hepatic failure and encephalopathy associated with cerebral edema in whom cadaveric liver transplantation is not an immediate option, extracorporeal ex vivo pig-liver perfusion is a reasonable alternative in the critical care setting.
I In nt tr ro od du uc ct ti io on n: : H. pylori is known for its causative role in gastric and duodenal ulcer disease. However, it is unknown whether H. pylori plays a role in the formation of stress ulceration in critically ill patients. Therefore we studied the presence of H. pylori infection in critically ill patients on admission to the intensive care and the relation to gastric and duodenal mucosal injury.  All trauma patients admitted to ICU and ventilated for more than 2 days were included. All patients were submitted to the same diagnostic and therapeutic procedures. None of them received systematic anti-acid therapies except in the case of documented history of GIB. GIB patients (confirmed by endoscopy) were compared to all others patients (GIB-). ventilation: 13.5 ± 11 days) were included. The incidence of GIB is 2.3% (n = 28). Table 1 summarizes the only significant risk factors after univariate analysis, Table 2 the final model after stepwise logistic regression.
All patients were successfully treated (medical treatment: 22, surgical treatment: 6) and the occurrence of GIB resulted in no additional mortality.
C Co on nc cl lu us si io on n: : Unavailability of the gastro-intestinal tract, acute renal failure and spinal cord injury are important risk factors for GIB. These results emphasize the importance of early enteral feeding which certainly represents the best prevention from the occurrence of GIB.   Our aim is to describe the evaluation of bleeding peptic ulcer, the indications for surgical treatment and the type of operative procedure or method that it has to be performed. From these patients 78 have cured and 6 died because of the high severity of the bleeding in association with their old age (age >80 years) and their general health status (2 of them had coronary disease and another one was diabetic with respiratory deficiency).
C Co on nc cl lu us si io on ns s: : Bleeding as complication of peptic ulcer remains up today very serious factor that increases the morbidity and the mortality. The aim is to avoid operating on all patients who would recover on medical treatment, but to operate on all patients who if treated medically would bleed again to a dangerous extent. Furthermore, if surgical treatment is undertaken, it should be performed at the optimal time and the safest operative procedure should be used, by a highly skilled surgeon. The age and general condition of the patient are important factors to consider. The amount of hemorrhage and the rate of hemorrhage are of prognostic significance. Many potentially pathogenic bacteria adhere to enterocytes via a mannose-specific adhesin. This mechanism has also been found in Lactobacillus plantarum 299v (L p 299v) both in the jejunum in the rectum. It is when bacteria are adhered to the mucosa that they interact with the enterocytes, both in negative and positive fashions.

P192 Probiotics in critical illness
Given in fermented oatmeal soup to healthy subjects, L p 299v was detected even 11 days after termination of intake.
In patients with persisting Clostridium difficile infections this fermented oatmeal soup containing L p 299v has also been effective in normalising gutflora and function.
O Ob bj je ec ct ti iv ve es s: : The prime objective was to study if L p 299v could survive and colonise on the mucosa in the intestine of patients treated in an ICU. Stool consistency and frequency were among other parameters studied.
M Me et th ho od d a an nd d m ma at te er ri ia al ls s: : In a randomised prospective trial 8 patients received 200 ml daily for 3 days and then 100 ml of an oatmeal soup containing 10 9 cfu/ml of L P 299v through out their stay in the ICU. Enteral nutrition as well as the oatmeal soup was started within 24 h after admission to the ICU. Control-patients (7) were treated in the same fashion except for the fermented oatmeal soup. The rectal mucosa was biopsied after admission and then twice a week. Biopsies were analysed blindly for bacterial content and species.
R Re es su ul lt ts s: : Four of the control patients were colonised with L p 299v on admission, but at the second biopsy they were all negative. Of eight treated patients none had positive cultures for L p 299v on admission but from the second sample and through their ICU-stay three of them had the bacteria adhered to the mucosa confirmed by cultures from homogenised biopsies.
Bacterial analyses revealed a reduction of sulphite-reducing clostridia in the treatment group. In treated patients lactobacilli increased while they remained at the original level in controls.
Diarrhoea was less frequent in treated patients.
D Di is sc cu us ss si io on n: : The initially positive biopsies in four of the control patients were probably due to that these patients had ingested L p 299v through the commercial L p 299v-containing 'Proviva', which is sold in almost all grocery-shops in southern Sweden. The use of antibiotics leads to a level of L p 299v below the limit of detection in those that were colonised from the beginning.
In the treatment group the L p 299v adhered to the mucosa in 3/8 patients although they as well were treated with antibiotics. It seems that repeated administration is essential if the bacteria should remain in sufficient numbers adhered to the mucosa. Our study shows that antibiotic treated patients in an ICU environment can benefit from probiotics. Less diarrhoea means less impact on the gutflora. B Ba ac ck kg gr ro ou un nd d Early enteral nutrition is an accepted gold standard in the treatment of critically ill patients. The major limiting factor is depressed gastric motility. However, while small bowel function usually remains intact, the placement of postpyloric feeding tubes increases the number of patients absorbing a sufficient volume of enteral nutrition early in their ICU course. To eliminate the need of invasive and expensive interventions, many bedside techniques have been proposed. Recently one effective way has been described in critically ill children [1]. We modified this method and used it in 27 adult patients.
M Me et th ho od ds s: : Thirty-one postpyloric feeding tubes were placed blindly in 27 consecutive ventilated postsurgical ICU patients using a bedside protocol. The feeding tube was considered to be postpyloric when following the insufflation of 20 ml of air an amount less then 5 ml could be reaspirated. The explanation is the immediat collapse of the narrow small intestine lumen, when air is reaspirated. The tube position was confirmed by abdominal radiography . I In nt tr ro od du uc ct ti io on n: : Peritoneal ventilation was studied few years ago [1,2] to be a successful auxiliary extrapulmonary method for improving oxygenation and CO 2 elimination in laboratory animals with experimental ARDS. Bowel ischemia during hemorrhagic shock is known to cause, after initial fluid resuscitation, late hazardous remote effects with multiple organ system failure and high mortality rate [3].
H Hy yp po ot th he es si is s: : In severe volume controlled hemorrhagic shock, peritoneal ventilation with oxygen would: 1) improve local oxygenation of the abdominal viscera, preventing later multiple organ failure, and 2) increase survival rate.   patients. This study was undertaken to test the hypothesis that the incidence of fungal infections is lower in critically-ill patients under mechanical ventilation receiving enteral rather than parenteral nutrition.
M Me et th ho od ds s: : By using a prospectively-built database, we analyzed retrospectively the charts of 110 critically-ill, intubated patients hospitalized in surgical and medical ICUs and receiving selective digestive decontamination (SDD). SDD is the prophylactic use of topical, nonabsorbable antibiotics to reduce the incidence of respiratory tract infections in critically-ill patients. It is known that this therapy significantly reduces the incidence, but not the mortality rate of pneumonia in ICU patients. In this study the SDD for all patients comprised of a PNV solution (polymyxin B, neomycin, vancomycin) at a dosage of 15 ml administered six times daily.
Seventy-nine patients received enteral nutrition and 31 patients parenteral nutrition.
Those patients without contraindications, and expected to be intubated for more than 72 h, received enteral nutrition which was started within 24 h after intubation. Patients with contraindications for enteral nutrition received parenteral nutrition which was discontinued when the criteria for enteral nutrition were met. We compared the incidence of fungal infections in both subgroups of patients, i.e., enteral versus parenteral nutrition.
R Re es su ul lt ts s: : The two subgroups were similar with regard to their APACHE II score, in age, sex distribution and comorbidities at the time of study entry. The rate of fungal infection was seen to be higher in the parenteral nutrition group, 5 out of 29, as compared to 7 out of 71 in the enteral nutrition group. However, this difference was not considered to be statistically significant.
C Co on nc cl lu us si io on n: : No significant difference is observed between enteral vs. parenteral nutrition in the incidence of fungal infections in critically-ill patients receiving SDD. Hyperglycemia is encountered during nutritional support in diabetics and in patients with stress-related glucose intolerance. Aim of this study is to determine the incidence of central venous catheter-related sepsis (CRS) and its relationship with the serum glucose levels in diabetic patients treated with Total Parenteral Nutrition (TPN R Re es su ul lt ts s: : There was no difference to the lenght of TPN therapy between diabetics and non diabetics. In 20 diabetics the serum glucose levels remained <200 mg/dl, and in 7 were constantly high (>200 mg/dl) in all measurements during TPN administration. Eleven out of the 96 non diabetics (11.4%) and 3 out of the 20 'euglycemic' diabetics (15%) presented CRS, but this difference was not significant (P = 0.8); however, CRS was presented in 5 out of the 7 diabetics whom serum glucose levels were >200 mg/dl during TPN therapy (P = 0.01).

P197 Hyperglycemia predispose to catheter-related sepsis in diabetic patients receiving Total Parenteral Nutrition
C Co on nc cl lu us si io on n: : The results of out study suggest that CRS is serious risk in diabetics receiving TPN if good control of glycemia is not maintained; in adverse, the incidence of CRS doesn't seem to be significantly increased in well-controlled diabetics.  Traditional methods of assessing respiratory muscle strength in the critically ill rely on some degree of co-operation from the patient, and are of limited use. We performed bilateral magnetic stimulation of the phrenic nerves, using two 43 mm magnetic coils placed anteriorly on the neck [1]. The transdiaphragmatic pressure change (TwPdi) was recorded using oesophageal and gastric balloon catheters in the conventional manner. Twitch endotracheal tube pressure (TwPett) was also recorded, which reflects twitch oesophageal pressure (TwPoes). The pressure readings, Poes, Pgas and Pett were displayed and recorded on a computer together with the calculated value for Pdi (Pgas-Poes).

P198 Dysregulation of glucose metabolism in enterally fed patients with acute pancreatitis
Twenty critically ill patients were studied (12 male, eight female), with a mean age of 59 years. Average length of ICU stay prior to the study was 27 days. The mean TwPdi was 9. Diaphragm contractility can be assessed in the sedated ICU patient, by magnetic stimulation of the phrenic nerves. This technique is non-volitional and is reasonably well tolerated. Our data shows that diaphragm conctractility in the critically ill patient is considerably less than in the laboratory based control subject [1]. Also, we report a good correlation between TwPoes and TwPett, leading to the possibility of further simplification of the technique.
A Ac ck kn no ow wl le ed dg ge em me en nt t: : Funded by the Wellcome Trust  Median (95% CI) Tw Q in the patients was 3.5 (2.6-5) kg compared with 9.5 (7.8-11.7) kg in controls (P < 0.01, Mann Whitney U Test) and weakness was not correlated with length of ICU stay.
The data demonstrate that profound quadriceps weakness can occur in critically ill patients. This weakness may influence mobilisation and rehabilitation. It is likely that other skeletal muscles are similarly affected, including the muscles of respiration. If so, this would in part, determine weaning outcome.
A Ac ck kn no ow wl le ed dg ge em me en nt t: : R Re es su ul lt ts s: : AII/ANP group: treatment with AII decreased VU, V Na U,V K U and GFR and increased RVR. ANP restored renal function to control levels (V Na U, V K U, GFR) or above (VU, RVR). ALD/ANP group: treatment with ALD induced an increase of V K U. Subsequent treatment with ANP further increased V K U and slightly decreased RVR. No effects on GFR, VU or V Na U were observed.
C Co on nc cl lu us si io on n: : Our findings suggest that blunted ANP effects during increased RAAS-activity are mainly determined by ALD. Interestingly, this is not only due to a blunted increase of sodium excretion (tubular mechanism) but also due to a blunted increase of glomerular filtration rate (altered glomerular vascular reactivity). Crit Care 1999, 3 3 ( (s su up pp pl l 1 1) ):P207 B Ba ac ck kg gr ro ou un nd d: : Treatment with urodilatin (URO; ANP-95-126), a kidney derived natriuretic peptide, may be beneficial in patients with incipient acute renal failure after cardiac surgery [1]. The findings about mechanisms regulating endogenous production and renal excretion of URO are controversial. Recent evidence suggests that urinary excretion of urodilatin (V URO U) is increased in patients after uncomplicated cardiac surgery and positively correlated with blood pressure [2]. Hourly urine output is one of the foremost indices which are monitored in patients who have undergone cardiac surgery. Traditionally, a urine output of less than 0.5 ml/kg/h triggers intervention [1]. However, renal failure may supervene in patients who have seemingly adequate renal function according to conventional monitoring. In this pilot study we analysed urine output and alterations in serum creatinine in 31 consecutive patients in the first 12 h following coronary artery bypass and valvular surgery. We found that an 'adequate' urine output alone is an unreliable predictor of subsequent renal impairment (indicated by a rise in 24 h post-operative serum creatinine by more than 50%). Six patients demonstrated such a rise in serum creatinine despite their producing urine outputs of greater than or equal to an average of 0.5 ml/kg/h. Of these, five subsequently required renal replacement therapy. This finding may have implications for the monitoring of cardiac patients who return to low-dependency patient care areas within 12 h of surgery.  Urine Output over first 12 hours (ml kg -1 ) The authors monitored the effects of the different diuretics upon the natrium and water homeostasis and acid-base balance. The aim of the study was to clarify the exact mechanisms of their action and possible ways of monitoring of their homeostatic effects. The effects of furosemide (18 patients), hydrochlorothiazide (eight patients), spironolactone (14 patients), acetazolamide (10 patients), amiloride (four patients) and manitol (eight patients) were monitored in critically ill patients using computer pro-gramme utilizing 17 routinely monitored input values and calculating creatinine clearance, tubular resorbtion , excretion fractions of sodium, potassium, water and osmotically active substances, clearance of the osmotically active substances, clearance of solute free water, electrolyte clearance, electrolyte free water clearance, urine outputs of sodium, potassium and urea, urea concentration index and serum and urine anion gaps. The development of parameters typical for each diuretic was evaluated using Student's t-test comparing the values before and during the treatment with the agent.

P209 Homeostatic indications for the administration of diuretics
M Ma ai in n r re es su ul lt ts s: : The natriuresis caused by furosemide is less important than the disturbing of the kidney concentrating ability. It is indi-cated in hyponatremia. Any of the evaluated parameters except serum potassium levels were not typical for the treatment with spironolactone. Hydrochlorothiazide reduces adverse effects of furosemide upon the kidney concentrating ability. It is useful in hypernatremia especially in the secondary nephrogenic diabetes insipidus and it is indicated in the secondary renal tubular acidosis. Amiloride was proved as the ideal therapy of chloride resistant metabolic alkalosis and hypokalemia. In comparison to acetazolamide it is potassium sparing drug and it seems to be less natriuretic. The indication for the use of acetazolamide is metabolic acidosis with the need for quick correction. Hyponatremia and hypoosmolality were not proven as the homeostatic indications for manitol. M Me et th ho od ds s: : Thirty-six children (median age 5.9 months, range 0.06-182 months) underwent corrective cardiac surgery for congenital heart disease (CHD). The patients were divided into two groups, Group A (n = 12) received 1 mg/kg of intravenous Lasix at the end of the surgical operation, Group B did not receive Lasix and acted as the control group. Urine samples were collected over the 1 st (t = 0) and 16 th (t = 16) postoperative hour and sent with paired blood samples for electrolyte measurements. Sodium excretion (NaEx) and urine volume (ml/kg/h) was compared between the two groups at t = 0 and t = 16 using the Mann-Whitney test. NaEx was calculated by multiplying urine volume by urine Na concentration and expressed as mmoles/kg/h. R Re es su ul lt ts s: : There were no significant differences in age, weight, preoperative renal function, CPB times or underlying heart disease (cyanotic vs acynotic) between the two groups.

P210 The effect of intraoperative Lasix on sodium excretion following cardiac surgery
NaEx and urine volume were significantly greater at t = 0 in the group that received Lasix (P = 0.013 and P = 0.001 respectively). These differences were no longer present at the 16 th postoperative hour (P = 0.67 and P = 0.38 respectively). In both groups sodium excretion correlated with urine volume (r = 0.98).
C Co on nc cl lu us si io on n: : Although intraoperative Lasix transiently increases sodium excretion and therefore urine volumes in the immediate post operative period it does not appear to offer any advantage by the 16 th postoperative hour, a time when renal water and sodium conservation is maximal.  D Di is sc cu us ss si io on n: : Lactate buffered CVVHDF leads to the removal of large amounts of endogenous bicarbonate per day (600-1.000 mmol). Its impact on the acid-base balance in septic shock is considerable. The approach with bicarbonate replacement flluid for the treatment of acute renal failure in septic shock seems to be advantageous to normalize an impaired acid-base balance.

Bicarbonate-buffered (n = 7)
Lactate-buffered (n = 6) P  S Su ub bj je ec ct ts s a an nd d m me et th ho od ds s: : The subjects were 10 patients with subarachnoid hemorrhage, who were admitted to our ICU, and received clipping operation within 48 h after the disease onset. Patients who had heart or renal diseases were excluded from this study. CNP levels in the plasma and CSF were measured at 6.00 on days 1, 3, and 7 of hospital admission by radioimmunoassay (RIA). As a control, CNP levels in CSF were measured in patients who received spinal anesthesia for orthopedic surgery. Differences between the measured levels on Day 1 and that on Day 3 or Day 7 were analyzed with Student's t test, and values less than 0.05 were considered statistically significant.
R Re es su ul lt ts s: : Plasma CNP levels in the subject and control patients were within normal range, and there were no significant group differences. Mean CNP levels ± SD in the CSF was 13.1 ± 2.4 pg/ml in the controls and 15.5 ± 2.8 pg/ml on Day 1 in the subjects and there were also no significant group differences. However, CNP levels in the CSF of our subjects was significantly different between Day 1 (15.5 ± 2.8) and Day 7 (10.6 ± 3.6) (P < 0.05).
D Di is sc cu us ss si io on n: : CNP levels is known to be highest in the brain, and that is thought to regulate the local cerebral blood flow, because some studies demonstrated that CNP induced relaxation of cerebral arterioles through cGMP in rat brain. Our findings show that CNP in the CSF acts as an inhibitor of vasospasm on Day 1, and 3 because CNP levels in the CSF decreased significantly on Day 7. C Co on nc cl lu us si io on n: : Any specific role of CNP was not indicated from our findings, but we presume that CNP in the CSF could function as a vasodilator when vasospasm occurs in the brain. Infective complications often occur after craniocerebral wounds caused by explosive fragments. These dangerous sequelae are, in our experience, the chief cause of delayed death after injury. Initial contamination. The presence of retained bone and metal fragments acting as a nidus for micro growth,and disturbances in cerebrospinal fluid (CSF) circulation, especially when the ventricular system is involved, are challenging problems in the management of missile wounds of the brain. The analysis covers 53 penetrating craniocerebral wounds, treated in ZhuHai and ZhoungShan in the period from 1988 to 1996. In 35 cases the head injuries were produced by explosive fragments and in the remaining 18 cases by low-velocity bullets. We have analysed the significance of these factors in cases undergoing operation within 24 h, the incidence of infection was 12.6%. Rising to 29.3%when delay in execess of 72 h after injury was unavoidable. We formed the opinion that the risk of infection was not significantly increased by failure to remove small inaccessible bone chips. The most formidable complication was CSF leakage which often resulted in infections of the central nervous system. This implies that successfully addressing the risk of infection is, potentially, the most powerful method of improving outcome from penetrating injuries to injuries to brain.   R Re es su ul lt ts s: : All patients presenting usually with clinical signs warranting early laparotomy. There were six full-thickness, and two partial thickness gastric injuries located in the anterior wall in eight cases. All injuries could be managed with simple surgical techniques without resections. Two patients exsanguinated on the operating table from associated injuries. All but one of the survivors had postoperative complications with a mean hospital length of stay of 18.4 ± 7.6 (range 10-30) days. C Co on nc cl lu us si io on n: : Blunt gastric injury is usually diagnosed at laparotomy for associated injuries but occasionally may be suspected from specific clinical findings. In most cases the injury is on the anterior wall. Simple repair is usually sufficient and the prognosis depends on the severity of the associated injuries.

P220 Serum sodium is inversely proportional to intracranial pressure in acute liver failure
P222 Protein S100 as a marker for cerebral outcome after cardiopulmonary resuscitation I In nt tr ro od du uc ct ti io on n: : S100 is a CNS-specific protein, derived from the cytosol of glial cells, that can be detected in peripheral blood after structural brain damage. We prospectively examined the prognostic value of S100-levels after cardiopulmonary resuscitaton (CPR  Table. D Di is sc cu us ss si io on n: : The use of tracheostomies and vasoconstrictors has increased during this period of study. Tracheostomy has no effect on mortality whereas vasoconstrictor usage appears to result in an increase in mortality. Morbidity and mortality among the population of Greece were 33238 and 2139 respectively within the above period.
C Co on nc cl lu us si io on n: : Nevertheless newspapers, radio and TV pay more attention to the narcotic and other causes victims than that of the car accidents which is the main reason of deaths among the young.
We can conclude that the suitable prehospitalized care of the injured victims and the rapid assessment and resuscitation at the Trauma Centers are the cornerstones of the current treatment and improve the outcome of the injured significantly. Having in mind that most of these patients are in the age group between 18 to 40 years, sports activity after survived polytrauma is a decisive factor of their quality of life.
M Ma at te er ri ia al ls s a an nd d m me et th ho od d: : In a prospective trial we evaluated the sports activity and sports performance of a group of 50 polytrauma patients (ISS >15) after a minimum time of 6 months following their discharge from hospital. The characteristics of the group were as follows (mean values): age 28.5 years, follow-up time 18 months, sex: male 40, female 10 patients. The ISS was 50. Cause of injury was in 96% a MVA, in 4% a fall from great height. The APACHE II on the first day was 17. The patients' time on respirator was 7 days. The stay on ICU was 11, and the stay in hospital 26 days. The sports activity and performance were evaluated according to a standardized score in all patients who practiced sports before the trauma. In addition, a performance test with spiroergometry and serum lactate samples could be performed in seven cases.
R Re es su ul lt ts s: : During the 6 months after discharge from hospital four patients died. The sports status of these patients could not be evaluated. Forty-six patients (92%) were available for further evaluation. The pre-and post-traumatic status of their sports is listed in Table 1. Sports performance and participation levels in differ-ence activities were evaluated in the 36 patients who practised sport before the trauma. The results are listed in Table 2.
C Co on nc cl lu us si io on n: : According to our results a decrease in activity and performance levels is obvious in the post trauma patient group. More than 70% of the patients practising sports before the trauma had to reduce their activity level. 23.86% had to quit their former activities. However, more than 50% of the patients were able to practice sports after their trauma. I In nt tr ro od du uc ct ti io on n: : During surgery, anaesthetists take extra care to prevent awareness of any patient who is having any kind of operation done, mainly under general anaesthesia either by using inhalational or intravenous medications, but patients in the ICU, mainly those on ventilatory support, with intubation and sedation, pass through a lot of psychological stress and frustration, which most of the times is not documented in the genera intensive care, and has never been done in our unit.
A Ai im m o of f t th hi is s s st tu ud dy y: : In this study at our general ICU, we tried to have a proper assessment of this problem in order to avoid it in the future, and to get a proper consensus regarding its existence and solution.
M Me et th ho od ds s: : Seventy patients between the ages 20-60 years, were interviewed 1 day after discharge from the ICU, about their memory of events during their stay. Patients with head injury, CNS infection or those who were disoriented at the time of interview were excluded from the study. The remaining 55 patients were oriented to place and time.
Intravenous opiates (morphine, pethidine) were used for analgesia as required, while sedation was achieved using midazolam and morphine infusions in appropriate doses as decided by the attending doctors and nurses.
Questions asked were generally about patients' memory of events and about their distressing experiences regarding pain, anxiety, dreams, fear, noises, causes of discomfort and others which will be displayed in the results section.
The same questions were repeated 5 days later.
R Re es su ul lt ts s: : The sample of patients were representative of our regular ICU admissions in their age group, APACHE II score and duration of stay.
The most distressing and commonest experiences recalled were: anxiety (68%), discomfort from endo-tube (60%), fear (54%), pain (52%), discomfort from N/G tube (48%), difficulty in communicating (33%), dreams and hallucinations (31%), discomfort from physiotherapy (24%), noise (15%), insomnia (13%), thirst (10%), some of these like anxiety, fear, dreams, hallucinations and insomnia had continued since discharge in 6% of patients. None of the studied experiences correlated with age, sex, or with the APACHE II score. On interviewing the patients 5 days later, there were no significant changes in their responses. C Co on nc cl lu us si io on n: : Our sedation and analgesia in the ICU is not enough to prevent unpleasant experiences, mainly those related to patient awareness.
More work is still needed, i.e. using sedation scores to improve our sedation and analgesia in the ICU. I In nt tr ro od du uc ct ti io on n: : A multi-centre study examining the safety and efficacy of the novel sedative agent dexmedetomidine, a highly selective alpha-2 agonist, possessing analgesic and sympatholytic properties.

P232 Sedation of patients in intensive care units by midazolam (MDZ): clinical and biological evaluation
M Me et th ho od ds s: : One-hundred and nineteen post-operative patients who required sedation and ventilation for at least 6 h on the ITU were enrolled. Ninety-eight completed the randomised, placebocontrolled, double-blind study (81 cardiac and 17 general surgical) in four centres in the UK, but all patient data was used in the safety analysis. Within 1 h of return from theatre, the study drug was started with a loading dose of 1 µg/kg for 10 min, followed by a maintenance infusion of 0.2-0.7 µg/kg/h to maintain a Ramsay sedation score of ≥3 and was continued for 6 h after extubation (maximum duration 24 h). Rescue sedation and analgesia was provided with midazolam and morphine respectively. Heart rate, systolic, diastolic pressures and central venous pressures were recorded at 10 min intervals for the first 30 min and then hourly.
R Re es su ul lt ts s: : Patient demographics were comparable as were Ramsay sedation scores between the two groups. The average dexmedeto-midine infusion rate was 0.35 µg/kg/h whilst intubated and 0.15 µg/kg/h after extubation (range 0-0.7 µg/kg/h). Data was collated for the initial 6 h of the infusion and for the period pre-and post-extubation ± 4 h, hence, allowing for the variation in the duration of intubation in the data analysis. Once adequately sedated the patients receiving dexmedetomidine achieved greater cardiovascular stability as compared to the placebo group, with a significantly lower and less variable heart rate (P = 0.0001), this was clearly demonstrated in the period around extubation when mean heart rate in the dexmedetomidine group was 75 (SEM ± 2.0), versus 92 (± 2.9) in the placebo group. Diastolic blood pressure showed a similar trend with a reduction of 5 mmHg in the dexmedetomidine group, but no sustained significant differences in systolic arterial pressure or central venous pressures. Of the 66 patients who received dexmedetomidine, 16 had transient episodes of hypotension (MAP <60 or >30% reduction from pre-infusion BP) and/or bradycardia (HR <50), mainly during the loading dose, of which three patients required temporary interruption of the infusion and three others required termination of the infusion.
S Su um mm ma ar ry y: : Dexmedetomidine may improve cardiovascular stability. Ultrasound examination of the abdomen showed ascites, pelvic fluid and enlarged ovaries (in our patients >11 cm in diameter) in all patients and chest X-ray revealed hydrothorax in five patients (71%). Ovarian hyperstimulation syndrome clinical feature is due to exaggerated ovarian response characterized by marked elevation of serum oestradiol levels and the presence of a large number of follicles (>20).

P234 Ovarian hyperstimulation syndrome (OHSS) at a maternity hospital
M Ma an na ag ge em me en nt t a an nd d o ou ut tc co om me e: : All patients had bed rest, fluid input-output control, adequate fluid intake, high protein oral intake, human albumin solutions iv and LMWH sc (nadroparin D Di is sc cu us ss si io on n: : This study has shown that the locally developed guidelines can be used to safely initiate warfarin in cardiothoracic patients immediately following cardiac surgery. In the future we intend to undertake an analysis to produce maintenance dose guidelines which are specific to cardiothoracic patients. A combination of these two guidelines should optimise the dosing of warfarin in cardiothoracic patients and contribute to an overall improvement in their care. I In nt tr ro od du uc ct ti io on n: : To study the incidence and prognosis of thrombocytopenia in an adult critically ill population, 329 patients consecutively admitted during a 5-month period to the medical intensive care unit (ICU) of a university hospital (212 patients) and a medical-surgical ICU of a regional hospital (117 patients), were prospectively surveyed. The primary outcome measure was ICU mortality.

P236 A prospective study of thrombocytopenia and prognosis in intensive care
R Re es su ul lt ts s: : One hundred and thirty-six patients (41.3%) had at least one platelet count < 150×10 9 /l. These patients displayed a higher APACHE (Acute Physiology and Chronic Health Evaluation) II, SAPS II (new Simplified Acute Physiology Score) and MODS (Multiple Organ Dysfunction Score) at admission, longer ICU stay (8 versus 5 days median (interquartile range)) and a higher mortality rate (crude odds ratio, OR = 5.0, 95% confidence interval, CI 2.7-9.1) than those who never developed thrombocytopenia (P < 0.0005 for all comparisons). Bleeding incidence rose from 4.1% in non-thrombocytopenic patients to 21.4% in patients with minimal platelet counts between 101 and 149×10 9 /l (P = 0.0002), and to 51.9% in patients with minimal platelet counts <100×10 9 /l (P < 0.0001). 19.5% of the study population died in the ICU following the index admission. Eighteen of 193 patients (9.3%) who never became thrombocytopenic died, versus 31 of 89 patients who were thrombocytopenic at admission (OR = 5.2, 95% CI 2.7-9.8, P < 0.0001) and versus 15 of 47 patients (31.9%) who developed thrombocytopenia later on during ICU stay (OR = 4.6, 95% CI 2.1-10.0, P = 0.0002). In addition we found that a drop in platelet count to ≤50% of admission was associated with higher death rates (OR = 6.0, 95% CI 3.0-12.0, P < 0.0001). In a linear regression analysis, adjusting for admission APACHE II, SAPS II and MODS, admission thrombocytosis and the occurrence of bleeding, nadir thrombocytosis remained significantly related to ICU mortality. C Co on nc cl lu us si io on n: : Thrombocytopenia is a simple and readily available risk marker for ICU mortality, independent of and complementary to established severity of disease indices. Both a low nadir thrombocytosis and a significant fall of platelet count predict a poor vital outcome in adult ICU patients. B Ba ac ck kg gr ro ou un nd d: : In septic patients disseminated intravascular coagulation is a severe complication whereby an altered platelet function appears contributory. Clinical outcome depends on an early diagnosis and sufficient therapy. In the present study the association of platelet function to inflammatory markers indicating disease severity was investigated.

P237 Platelet function and inflammatory markers in septic patients
M Me et th ho od ds s: : Inflammatory markers C-reactive protein, procalcitonin, interleukin-6 and interleukin-10 were measured using standard methods in 18 patients fulfilling clinical, inflammatory and hemodynamic criteria of sepsis. Platelet activation marker P-selectin was flow cytometrically analysed ex vivo and after stimulation using 5 µmol/l ADP and 10 µmol/l TRAP-6. R Re es su ul lt ts s: : Flow cytometrically measured platelet function was tightly associated with inflammatory markers. Pre-activation of platelets in the circulation was significantly correlated to plasma levels of procalcitonin (P < 0.023), whereas in vitro induced reagibility after ADP-and TRAP-6 stimulation correlated well with the plasma concentration of the C-reactive protein (P < 0.001; P < 0.012). Furthermore, a close relation of IL-6, but not of IL-10, plasma levels to TRAP-6 stimulated P-selectin expression was observed (P < 0.033).
C Co on nc cl lu us si io on n: : Platelet function was demonstrated to be tightly associated with the inflammation process in septic patients. Whether this finding may be a useful marker for disease severity and the development of a disseminated intravascular coagulation should be clarified in prospective studies. O Ob bj je ec ct ti iv ve e: : Analyses from patients with haematological malignancies admitted in a medico-surgery ICU of an oncology hospital. P Pa at ti ie en nt ts s a an nd d m me et th ho od ds s: : Retrospective observational study on patients with haematological malignancies admitted in ICU from October/96 to October/98, coming from Paediatric Department (PD), Onco-Haematological Unit (OHU) and Bone Marrow Transplantation Unit (BMTU). We analysed the patient data, namely the underlying malignancy, the reason for admission, the type and number of organ dysfunction (including neutropenia and requirement of mechanical ventilation), the time in ICU, acute physiology, age, chronic health evaluation (APACHE II) and sepsis-related organ failure assessment (SOFA). R Re es su ul lt ts s: : Between October/96 and October/98, 46 onco-haematological patients were admitted in the UCI (56 inpatients) with ages from 9 months to 70 years old, 23 female/23 male: 6 came from PD (13%), 29 from OHU (63%) and 11 from BMTU (24%).

P238 Retrospective study of patients with haematological malignancies admitted in an intensive care unit
Underlying haematological malignancy: Non Hodgkin Lymphoma (34%), Acute Myeloid Leukaemia (21%), Chronic Myeloid Leukaemia (15%), Hodgkin Disease (15%), Acute Lymphoid leukaemia (11%), Multiple myeloma (4%). Six of the 46 patients were excluded because of the short time in ICU (≤12 h). Six patients were readmitted. The mean time of stay was 8.2 days. The reasons for ICU admission were: acute respiratory failure (54%), multi-organ dysfunction (MOD; 14%), post-surgery (14%), septic shock (8%), tumour lysis syndrome (6%), hypovolemic shock (2%) and neurological dysfunction (2%). The ICU mortality was 52.5%, being 76% of them neutropenic patients with MOD and requiring invasive ventilation. 89% of the patients coming from BMTU died. C Co on nc cl lu us si io on n: : The main risk factors to dead in an ICU are the number of organ dysfunction at admission, the requirement of invasive ventilation, BMT, APACHE II ≥20 and SOFA ≥15. O Ob bj je ec ct ti iv ve es s: : Characterise the bone marrow haematological changes in severe sepsis/septic shock patients and to evaluate the prognos-tic value of the marrow cell differential count (myeloid, lymphoreticular, erythroid series).
M Ma at te er ri ia al ls s a an nd d m me et th ho od ds s: : Prospective study of 29 patients with the diagnosis of severe sepsis/septic shock of different etiologies. Age, SAPS II in the first 24 h, organ dysfunctions according to SOFA, organ failure according to Knaus and the final outcome were considered in the present study.
The bone marrow of each patient was studied and a differential count considering the myeloid, lymphoreticular and erythroid series was made. The patients were separated according to final outcome (dead and alive) and the bone marrow differential counts were compared between the two groups applying t Student test.
R Re es su ul lt ts s: : See Table. C Co on nc cl lu us si io on n: : In the present study significant statistical correlation was found between lymphoreticular count and mortality. We can conclude that bone marrow evaluation has had a prognostic value in this patient group.
Dead (n = 17) Alive (n = 12) P  Aortic reconstructive surgery is associated with post-ischemic reperfusion and oxidative stress. It is expected that oxidative stress should be self-limiting during healing process in the postoperative period. Lipid hydroperoxides (LHP) are one of oxidative stress markers therefore we evaluated changes in LHP level in the course of uncomplicated healing in patients who underwent abdominal aortic reconstruction. Ten male patients, aged 56-74 years (mean 65.5 ± 6.01) with abdominal aortic aneurysm or aortoiliac occlusive disease were submitted to aortic grafting operation. LHP concentration was measured in blood samples collected via central line prior to (P), at the end of (E) and 1 h, 24 h, 48 h and 72 h after surgery. The results are presented as mean ± SEM. *Nonparametric one-way ANOVA-Kruskal-Wallis test LHP concentration was significantly increased at the end of surgery and started to decrease just after 1 h later reaching the initial level within 48 h. The obtained results indicate limitation of the oxidative stress in the course of uncomplicated healing. The results also suggest that LHP level can be used for monitoring of oxidative stress activity in humans. I In nt tr ro od du uc ct ti io on n: : Too little is still known about the incidence rate of thromboembolic complications in polytrauma patients after ICU treatment, as only a small amount of data is available on this topic. The majority of the studies published to date that have been performed to assess the incidence rate of thrombosis in multiply injured patients only refer to the clinical symptoms of a venous thrombosis. A systematic screening-examination for the assessment of the incidence rate of thromboembolic complications in the above-mentioned patient collective has not yet gained acceptance as a routine method in clinical practice.

P241 Deep leg veign thrombosis in multiply injured
M Ma at te er ri ia al l a an nd d m me et th ho od d: : Between January 1996 and December 1997, 50 polytrauma patients were included in a prospective clinical study. Including criteria were: an initial ISS-score >16, a stay on the ICU of at least 72 h and a time on the respirator of at least 72 h. All patients were examined for a deep veign thrombosis by using a standardized protocol and by means of a colour-coded duplex (ccd) sonography. In cases in which the clinical or/and sonographic examination yielded results of a suspected veign thrombosis, a phlebography was performed. In cases of a suspected pulmonary embolism a pulmonary angiography was performed. The colourcoded duplex sonography was used before the patients were mobilized or transfered to an other ward (generally after 15 days).
R Re es su ul lt ts s: : If not indicated otherwise numbers are given as median.
The age of the 38 male and 12 female patients was 38.6 years. The severity of trauma was characterized by an ISS-score of 39.5 points. Eight patients died of a multiorgan-failure during their stay on the ICU. The autopsy findings reveal that no patient died of the of a veign thrombosis or a pulmonary embolism. Of the remaining 42 patients, 8 patients (19%) showed deep leg veign thrombosis in the ccd. In three of these patients (7%) also a pulmonary embolism occurred. C Co on nc cl lu us si io on n: : Having in mind the results of our study the incidence rate of thromboembolic complications in polytrauma patients seems to be much higher as expected in comparison to the published results of other authors. 13/24 patients with OS were given thrombolysis according to BAPE regimen (rTPA 0.6 mg/kg over 15 min); 11/24 patients with OS were not given thombolysis because of absolute contraindications. Thrombolytic therapy decision-making rested on clinical data, on echocardiography in 38% of case and on echocardiography and lung scan in 61% of cases. Intra-hospital overall death-rate was 37.5% (9/24 patients); all 13 patients given thrombolysis were alive at discharge, whereas, 9/11 (81.8%) patients not given thrombolysis died in the hospital. C Co on nc cl lu us si io on n: : We found OS in 10.2 % of PE cases; 13 patients given thrombolysis all were alive and showed stable hemodynamic parameters at discharge, whereas 9/11 patients not thrombolysis given died during hospital stay. This outlines the need of an expeditious clinical and instrumental diagnosis as a tool of decisionmaking, especially about thrombolytic therapy. Moreover, we found a 100% sensibility of D-dimer, hypoxemia as detected by ABG analysis, echocardiography and perfusion radionuclide lung scan. I In nt tr ro od du uc ct ti io on n: : Aortic valve prosthesis determine a transvalvular gradient (∆p) with changes in aortic flow that can affect left ventricular geometry and function, implanted prosthesis and aortic route. In order to improve prognosis a new prosthesis has been proposed , the so called 'stentless valve' (SV), whose main characteristic is the absence of the supporting ring. Aim of our study was to evaluate if such valve could influence early postoperative course in ICU.

P242
M Me et th ho od ds s: : Forty patients, age 67 ± 11 and EF 56 ± 14, undergoing aortic valve replacement were enrolled. Nineteen patients, 12 with aortic stenosis (AoS) and seven with insufficiency (AoI) underwent SV implantation (group S); 21 patients, 13 with AoS and eight with AoI, received mechanical valve (group M). Anaesthesia with remifentanil and propofol, moderate hypothermia (30°C) and anterograde blood cardioplegia were used. In all cases mechanical ventilation (MV) and intubation time, need for inotropic support and blood loss were registered during ICU stay. R Re es su ul lt ts s: : No differences were found in duration of MV (125 ± 30 min in group S versus 136 ± 12 min in group M, P > 0.05) and intubation (3 h in group S versus 3.4 h in group M, P > 0.05) in patients with aortic stenosis. In patients with AoI MV and intubation time was shorter in group S (respectively 140 ± 25 min versus 155 ± 18 min in group M, P < 0.05; 3.9 h versus 4.5 h in group M, P < 0.05). No differences were observed in blood loss between two groups, nor in dopamine dosage (5.4 µg/kg/min versus 6.6 µg/kg/min, P < 0.05).
D Di is sc cu us ss si io on n: : Our results show no differences in early postoperative outcome in patients with AoS when treated with mechanical or SV. Vice versa SV seems to improve ICU course of patients with AoI, with regard to duration of MV and intubation. Other studies with echocardiography are necessary to clear if these differences can be due to a lower aortic transvalvular gradient of SV. Lung resections are correlated to high mortality (4-6%) and morbidity (20-40%) that can increase in high risk patients.
Objectives of this study is to analyze preoperative risk factors, in a group of high risk patients undergoing thoracotomy for lung cancer and to assess the relationship with postoperative complications.
M Me et th ho od ds s: : From January 1996 to December 1997 43 patients, undergone pulmonary resection for lung cancer, were considered at high risk and enrolled in this study according to one or more of the following including criteria: age older than 70 years; previous cardiovascular disease; poor pulmonary function (FEV1 <65% predicted values, PaO 2 <65 mmHg or PaCO 2 >42 mmHg); chronic systemic disease. Patients clinical data are reported in the Table. Anesthetic technique was the same for all patients. All the patients were monitored with EKG, pulse oxymetry (Nellcor N200), invasive arterial pressure, pulmonary artery catheter when necessary and in-end expiratory gas analysis. Preoperatively an epidural catheter was inserted in T6-T11 space. Anesthesia was maintained with isoflurane 0.5% vecuronium and fentanyl combined with epidural analgesia (bupivacaine 0.5% and fentanyl). A continuous infusion through the epidural catheter of morphine 20 mg in 250 ml normal saline 0.9% at 5 ml/h was used for postoperative pain relief. Surgical procedures included: 33 lobectomy, 4 bilobectomy, 3 sleeve resections, 2 pneumonectomy. Relationship among different preoperative risk factors and postoperative complications were performed with χ 2 test and corrected with Fisher's exact test.
R Re es su ul lt ts s: : Mean age was 69 yrs (range 50-83 years). All patients were extubated in the operative room at the end of surgical procedures.
C Co on nc cl lu us si io on n: : In our experience lung resections in high risk patients have low mortality and morbidity. Therefore, age over 70 years alone has no longer to be considered a limiting factor in patients undergoing surgery for lung cancer. High risk patients need a very careful preoperative evaluation of cardiovascular and pulmonary function in order to avoid perioperative complications and to reduce the morbidity. An appropiate surgical and anesthetic technique, and postoperative pain relief improves outcome in high risk patients. C Co on nc cl lu us si io on n: : We conclude that this SRD patients should be admitted to the ICU on the same basis as other patients. In this population neither the diagnostic of the underlying disease, nor the use of immunosuppresive therapy did influence the short and long outcome. Long-term survival depended only on the age and the need of mechanical ventilation. Calibration using the Hosmer-Lemeshow goodness of fit test, showed a χ 2 16.15, df 8, significance 0.04 for PRISM and χ 2 17.05, df 8, significance 0.03 for PIM. Using a cut off at P = 0.5, sensitivity and specificity for PRISM was 98.3% and 33.3%, and 99.2% and 26.7% for PIM. C Co on nc cl lu us si io on n: : Neither PRISM, nor the new scoring system PIM are well calibrated for predicting individual mortality. However, despite the small numbers, the area under the ROC plot for PIM compares favourably with the original work by Shann et al. [1] (0.87 vs 0.83). Therefore we would concur with their conclusion that PIM is accurate enough to describe the risk of mortality in groups of children, and has the added advantage of needing less data collection than PRISM.  A number of near drowned children needed admission to a PICU due to the severity of their condition. The aim of this review is to illustrate the epidemiology, the clinical features, the management and the outcome of near drowned children admitted to our PICU. For this purpose, we reviewed the charts of near drowned children admitted to our unit during the last 11 years. The study population consisted of 11 children (7 boys and 4 girls) aged 2.5-12.5 years (mean age 7.2 years, SD4,1, SE1.3). At the same period a five-fold number of near drowned children were hospitalized in pediatric wards. The submersion site, among our patients, was sea in 6 cases, a swimming pool in 3, a pond in 2 cases. Ten children were transferred from district hospitals where they had initially received advanced life support. 6/10 children were transferred intubated. 4/10 children were in cardiac arrest after the accident, 3/4 had been given basic life support at the accident site and sub-sequently 2 of them were intubated on their arrival at the nearest district hospital. 1/4 was intubated in the nearest hospital where he was transferred with brain death without having received appropriate basic life support for about 30 min. 2/3 children who initially were apnoeic and comatose, required intubation. The remaining 4 children had respiratory distress and irregular respiration but only one needed intubation. 8/11 patients have clinical and roentgenographic features of pulmonary oedema. The intubated children remained on mechanical ventilation from 12-36 h. Convulsions occurred in 3 children. In 1/11 patients there were signs of high intracranial pressure with good response to mannitol administration. 10/11 patients survived and discharged from hospital after 3-6 days of hospitalization overall. The patient who was admitted with brain death, never recovered. All survived children had no neurological sequalae on their follow up 2-5 years later. Our results emphasize that even the most severe cases of near drowning have a favorable outcome, provided that the victims are given basic life support at the accident site.  Table. C Co on nc cl lu us si io on n: : The timing of development of severe sepsis and multiple organ failure appears to be an important factor for outcome with a significantly higher mortality among those admitted with sepsis. The admission APACHE II score and the score on the day of development of sepsis were lower among those who developed sepsis. This may be attributable to these patients already being in the ICU environment and thus receiving closer monitoring and more timely intervention. This finding may be of importance in the design of future trials to evaluate new treatment modalities.

P254 Characteristics of patients with sepsis and multiple organ failure in the UK
All severe sepsis and multiple organ failure patients  O Ob bj je ec ct ti iv ve es s: : The SOFA (sepsis-related organ failure assessment) score describes quantitatively the degree of organ dysfunction. Although primarily not designed to predict outcome any assessment of morbidity must be related to mortality to some degree. We therefore investigated whether an increasing SOFA score is associated with a higher hospital mortality in patients (pts) of a medical intensive care unit (ICU Hospital mortality was 14.5%. SOFA score for NS was significantly higher than for S (5.9 ± 3.7 vs. 1.9 ± 2.3, P < 0.05). The AUROC was 0.82 ± 0.04 for the SOFA score and 0.77 ± 0.04 for SAPS II. C Co on nc cl lu us si io on n: : SOFA score discriminates well between S and NS 24 h after admission. Respiration, liver and coagulation showed an increasing mortality rate with a higher SOFA score for each organ.
Although the SOFA score was primarily designed for use in septic patients it may be also applied for pts of a medical intensive care unit. I In nt tr ro od du uc ct ti io on n: : Severity scoring models can provide accurate outcome prediction but their performance is very influenced by variations in patient case-mix. Therefore, none of the usual scoring systems (APACHE II, SAPS II and MPM 24) fitted to this ICU: they had good discriminatory power but poor calibration. Logistic regression analysis of their variables was performed to identify the most predictive association to ICU mortality.
M Me et th ho od ds s: : Data of 823 consecutive patients (pts) admitted to the ICU were prospectively collected. Pts who stayed less than 24 h at the ICU or were burn or had less than 16 years old were excluded.
For pts with several admissions, only the first ICU admission was considered. The remaining 709 pts were divided in two groups: 418 (59%) pts constituted the development set and 291 (41%) pts became the validation set. After calculating the scoring indices, their variables and respective weights were separately analysed. Variables with P value <0.05 at univariate analysis were included as independent variables at logistic regression and vital status at ICU discharge was considered as dependent variable. hospital mortality 33.7%. APACHE II was 16.7 ± 8.4 and SAPS II was 33.5 ± 16.5. Through statistical modeling, an hibrid model was generated, with variables and points from the three indices. With this model, the prediction obtained was: development set with discrimination ROC = 0.89 and calibration goodness-of-fit C = 1.68 and validation set with ROC = 0.84 and goodness-of-fit C = 7.72. C Co on nc cl lu us si io on n: : Hemodynamic instability, infection, impaired renal function, respiratory failure and coma were the best predictors of death. Early identification of patients at major risk may allow treatment with more resources and interventions, in order to improve survival. Furthermore, this study shows that suitable statistical management may be useful to customize and enhance the prognostic accuracy of the currently available scoring systems. The purpose of this study was to calculate nurse/patient radio by using TISS-28, and to assess time allocation to nursing activities in the intensive care unit. In this study the TISS scores of 416 patients were calculated in the intensive care unit 10 weeks long using the TISS-28 form. In order to determine the duration of the nursing care activities due to nursing care categories the work sampling method was used. A sampling matrix for 10 weeks was created and the nursing care activities were observed 7 days a week for two day shifts (08.00-16.00). The data collection instruments were, the 'TISS-28' and 'Work sampling form for intensive care unit nursing activities'. The TISS-28 point for ICU was 40.41 for day shift. One TISS-28 point equals 11.88 min of the 480 min in each shift. Related literature shows that nursing care activity for one day makes 40-50 TISS score. The percentage of nursing time spent on nursing activities in the ICU was calculated by using work sampling. Results indicated that 44.25 % of nurses time was spent in activities in TISS-28; 12.87% in activities not in TISS-28; 25.8% in indirect patient care, 6.21% in organnizational activites, 10.64% in personnel activities and 0.15% in other activities. It is shown that category one represents TISS-28 and that the increase in TISS score results in the increase in nursing care activity duration. These result show that the TISS-28 can be useful to determine the patient/nurse ratio in intensive care units.
centers continue to perform autopsy as a means of quality control. From 1995 to 1996, we performed an autopsy study in a medical intensive care unit of a university hospital: 93% of the 140 deceased patients in our medical ICU underwent an autopsy, 100 consecutive patient files were studied.
The clinical diagnosis were made by internists, specialized in intensive medicine; the diagnosis on autopsy were made by a pathologist. According to the criteria of Goldman [1], the clinical and autopsy findings were categorized into major and minor diagnoses. A missed diagnosis on clinical grounds was classified as a class I error (if detected before death, this would probably have caused a therapeutic change with possible altered outcome) or as a class II error (if known before death, this diagnosis would not have led to a change in therapy).
In 16% of the patients, a class I missed diagnosis was detected (cardiac tamponade, myocardial infarction, fungal pneumonia); in 9%, a class II missed diagnosis was detected (most frequently tumors). Sometimes the diagnosis was missed due to a combination of severe, acute problems (e.g. development of cardiac tam-ponade after insertion of a venous catheter during hemorraghic shock), or due to a lack of sensitive and specific investigational methods (fungal pneumonia is frequently suspected in immunocompromised patients, but is often difficult to confirm), or due to logistic transportation problems in the hemodynamically unstable patient (e.g. retroperitoneal hemorrhage is not always detectable on bedside echography; for diagnosis, CAT-scan is needed). C Co on nc cl lu us si io on n: : Even in the era of increasing diagnostic possibilities, due to improved medical technologies in the ICU, postmortem examination still remains useful in detecting unexpected diagnoses, missed in the premortem clinical evaluation. Our observations suggested the need for constant alertness and an aggressive investigational planning in patients with unexplained shock or pulmonary infiltrates. I In nt tr ro od du uc ct ti io on n: : We prospectively analysed changes in the quality of life (QOL) in patients before and 6 months after admission to a medical intensive care unit (ICU).
P Pa at ti ie en nt ts s a an nd d m me et th ho od ds s: : All patients admitted to the ICU were eligible for inclusion. Patients <18 years and those who died or were discharged within 24 h of admission were excluded, QOL measures were collected during interview during the first 24 h of ICU stay and 6 months after admission using a questionnaire especially designed for ICU patients developed by the Spanish Group for Epidemiological Analysis of Critical Care Patients [1]. Baseline QOL referred to the 2 months prior to admission and were compared with measures at 6 months using Wilcoxon matched-pairs test, P < 0.05 was considered statistically significant.
R Re es su ul lt ts s: : During the first 12 months of the study period 326 patients met the study criteria; mean age was 58 ± 17 (± SD) years, median 60, range 19-95 years, 55% were male. Mean ICU length of stay was 10.4 ± 15.1 days, range 2-127 days. Mean APACHE II score was 23 ± 10, range 0-51. Mean TISS score after 24 h was 33 ± 14, range 0-69. Mortality rates were: ICU 24%, hospital 6%, 9% within the following 6 months after hospital discharge. Up to now 147 patients completed the questionnaire after 6 months, six patients (1.8%) were lost to follow-up. Relative to baseline a significant worsening was noted in the subscale of normal daily activities (P = 0.013). No significant changes were seen in total QOL score (P = 0.25) an the subscales of physiologic basic activities (P = 0.06) and emotional state (P = 0.09). No correlation existed between APACHE II scores and QOL (r = 033). C Co on nc cl lu us si io on n: : Six months after ICU treatment patients had a significant decrease in the level of their daily activities. Basic physiologic activities and emotional status are not significantly altered. 90% of the long-term survivors were living at home and all previously occupied patients were able to return to their previous profession. Death should be managed as vigorously as life saving. Historically intensive therapy is withdrawn in the intensive care unit, but we would like to present four cases where intensive care treatment was withdrawn at home. The staff of the Intensive Care Unit at Middlemore Hospital have taken four patients home, on ventilatory and inotropic support, and withdrawn care when the patient was settled in their home, surrounded by family. This is felt to aid in the grieving process, and in many cases is culturally desirable. The cases thus far are subarachnoid haemorrhage, massive intracerbral bleed and intractable septic shock. From our experience we recommend that certain selection criteria are observed. There is a need for the patient and family to live locally, support is required from the local general practitioner and district nursing service, and a clear explanation of the whole process must be understood by all family members prior to leaving the intensive care unit. A palliation plan must be commenced prior to leaving the intensive care unit.
We see this as a practical option in selected intensive care patients. concordance of what was felt should be done and what was done in clinical scenarios. Respondents considered that they provided high information to patients, including in the event of iatrogenic complication. In terms of end-of-life decisions, 35% of respondents wrote that they would involve the family in discussions (not an option available for selection): this appeared to be a pragmatic approach to dealing with relatives. Withdrawal of treatment was considered to be different to withholding treatment by 43% of respondents. 34% of respondents would change a do-not-resuscitate order that had been previously instituted. 15% of respondents considered that an Ethics Consultant would assist in their practice, with 95% supporting the inclusion of ethics teaching during medical training. The following data were collected for all the admitted ICU patients during a 2-month study period: age, sex, SAPS II, main diagnostic, previous chronic disease. In patients for whom WH or WD were indicated, additional data were recorded. The reasons to withheld or withdraw treatments and the type of WH or WD life support treatments were recorded. R Re es su ul lt ts s: : Treatments were withheld or withdrawn in 807 out 7309 (11%). WH and WD were indicated in 336 patients (4.6%) and 471 patients (6.4%) respectively. ICU patients undergoing WH or WD were older and had higher SAPS 2 than the remaining patients.

P267 Results of an ethical questionnaire distributed to members of the Australian and New Zealand Intensive Care Society
Decisions of withhold or withdraw were more frequent in patients with previous chronic diseases or cardiac arrest before admission in ICU. Futility and the poor expected quality of life were the most frequently cited reason for WH or WD. Decision to not ventilate the patient was the most frequently reported withheld treatment (n = 214; 15%). Vasopressors were either not started or limited in their dosage in 196 patients (14%). The most frequently withdrawn life support treatment were vasopressors (19%). Extrarenal epuration was discontinued in 67 patients (7%). Lowering FiO 2 to 21% was indicated in 155 patients (14.5%), discontinued ventilation was ordered in 101 patients (9.4%) and extubation was performed only in 34 patients (3.1%). Withdrawal of hydration was rarely performed (n = 16 ; 1.5%). 1176 out of the 7309 (16.1%) included patients died. 628 out 1176 died (53%) after support was withheld or withdrawn. Most of the time WH or WD was decided by the medical team. A unique M.D. was involved in the decision in 37 (12%) of cases. Paramedic (nurses) opinions was taken in account for the decision in 482 (59.7%). Family was involved in the process in less than 50%. C Co om mm me en nt ts s: : The reality and the frequency of WH and WD life support treatments have been demonstrated in this large study involving an important number of French ICU.