An appraisal of respiratory system compliance in mechanically ventilated covid-19 patients

Background Heterogeneous respiratory system static compliance (CRS) values and levels of hypoxemia in patients with novel coronavirus disease (COVID-19) requiring mechanical ventilation have been reported in previous small-case series or studies conducted at a national level. Methods We designed a retrospective observational cohort study with rapid data gathering from the international COVID-19 Critical Care Consortium study to comprehensively describe CRS—calculated as: tidal volume/[airway plateau pressure-positive end-expiratory pressure (PEEP)]—and its association with ventilatory management and outcomes of COVID-19 patients on mechanical ventilation (MV), admitted to intensive care units (ICU) worldwide. Results We studied 745 patients from 22 countries, who required admission to the ICU and MV from January 14 to December 31, 2020, and presented at least one value of CRS within the first seven days of MV. Median (IQR) age was 62 (52–71), patients were predominantly males (68%) and from Europe/North and South America (88%). CRS, within 48 h from endotracheal intubation, was available in 649 patients and was neither associated with the duration from onset of symptoms to commencement of MV (p = 0.417) nor with PaO2/FiO2 (p = 0.100). Females presented lower CRS than males (95% CI of CRS difference between females-males: − 11.8 to − 7.4 mL/cmH2O p < 0.001), and although females presented higher body mass index (BMI), association of BMI with CRS was marginal (p = 0.139). Ventilatory management varied across CRS range, resulting in a significant association between CRS and driving pressure (estimated decrease − 0.31 cmH2O/L per mL/cmH20 of CRS, 95% CI − 0.48 to − 0.14, p < 0.001). Overall, 28-day ICU mortality, accounting for the competing risk of being discharged within the period, was 35.6% (SE 1.7). Cox proportional hazard analysis demonstrated that CRS (+ 10 mL/cm H2O) was only associated with being discharge from the ICU within 28 days (HR 1.14, 95% CI 1.02–1.28, p = 0.018). Conclusions This multicentre report provides a comprehensive account of CRS in COVID-19 patients on MV. CRS measured within 48 h from commencement of MV has marginal predictive value for 28-day mortality, but was associated with being discharged from ICU within the same period. Trial documentation: Available at https://www.covid-critical.com/study. Trial registration: ACTRN12620000421932.


Background
Millions of people have been infected by SARS-CoV-2 worldwide, and many of those have been hospitalized for respiratory complications associated with coronavirus disease-2019 . Many of those COVID-19 hospitalised patients have received mechanical ventilation (MV), due to the development of acute hypoxemic respiratory failure and acute respiratory distress syndrome (ARDS) [1][2][3][4]. To date, several landmark studies [5][6][7][8] have improved our understanding of COVID-19 pulmonary pathophysiology, but pulmonary derangement in COVID-19 and appropriate ventilatory management remains incompletely characterized.
Earlier reports on the pulmonary pathophysiology of COVID-19 patients reported conflicting results and extreme heterogeneity in levels of pulmonary shunting, static respiratory system compliance (C RS ), [9][10][11][12] and substantial heterogeneity in lung recruitability [13,14]. Adding further to the controversy over C RS in COVID-19 patients, Grasselli and collaborators [7] have compared findings from an Italian repository of COVID-19 ARDS with previous ARDS cases of different etiologies. They found statistically significant higher C RS in patients with COVID-19 ARDS. In addition, they found that patients who presented with lower C RS and higher D-dimer values had the greatest mortality risk. In line with these figures, in a small-case series, Chiumello and collaborators found that COVID-19 patients presented higher C RS levels in comparison with patients with ARDS from other etiologies and matched levels of hypoxemia [12]. Regrettably, those previous reports did not provide any information on how C RS progressed beyond a punctual assessment during the period of MV. In contrast, in another landmark study by Ferrando et al. [6], C RS figures from a Spanish database were very similar to previously published cohorts of ARDS patients. The authors also found that intensive care unit (ICU) discharge and mortality were not influenced by the initial levels of C RS .
In a pandemic caused by a novel virus, access to international data is vital, because it may help account for differences in populations, access to medical care, equipment and critical variations in clinical managements among countries. Thus, analysis of international repositories improves the overall understanding of a novel disease and helps establishing best practices to enhance outcome. One example of how single-center or single-country studies can influence medical care early in a pandemic, before being contradicted by subsequent international findings is the issue of C RS . Indeed, as this parameter can be markedly impacted by fine variations in ventilatory management, extrapolations from monocenter or single-country studies may be challenging. In early January 2020, the COVID-19 Critical Care Consortium incorporating the ExtraCorporeal Membrane Oxygenation for 2019 novel Coronavirus Acute Respiratory Disease (COVID-19-CCC/ECMOCARD) group was founded to investigate patients presenting to ICUs worldwide.
Here, we present a comprehensive appraisal of C RS in mechanically ventilated COVID-19 patients enrolled into the COVID-19-CCC/ECMOCARD international study, in order to understand the dynamics of C RS during the first week of mechanical ventilation and its potential impact on patient outcomes.

Study design and oversight
The COVID-19-CCC/ECMOCARD is an international, multicentre, cohort observational study ongoing in 351 hospitals across 53 countries. The full study protocol is available elsewhere [15]. To summarize, participating hospitals obtained local ethics committee approval and a waiver of informed consent was granted in all cases. ISARIC/SPRINT-SARI data collection began at admission to hospital, while data collection for the COVID-19-CCC observational study commenced at admission to the ICU. De-identified patient data were collected retrospectively and stored via the REDCap electronic data capture tool, hosted at the University of Oxford, United Kingdom or Monash University, Melbourne, Australia.

Study population
We reviewed data of all patients admitted to the ICU at a COVID-19-CCC collaborating site, from January 14 through September 30, 2020, with a clinically suspected or laboratory confirmed diagnosis of SARS-CoV-2 infection, through naso-pharyngeal swab for realtime PCR SARS-CoV-2 detection. Of note, suspicion of SARS-CoV-2 infection was based on symptoms and onset of infection and was confirmed by the clinician when COVID-19 infection was the most likely cause of the symptoms experienced. Patients excluded were those under the age of 15 years or admitted to an ICU for other reasons. We focused our analysis on patients on controlled MV and with a computed C RS value within 48 h of MV commencement.

Definitions and pulmonary mechanics computations
C RS was calculated as: tidal volume (mL)/[(airway plateau pressure-PEEP (cmH 2 O))]. Of note, we provided to data collectors a detailed data dictionary, with instructions on how to collect airway plateau pressure values, via an inspiratory pause of approximately 3 s. We computed C RS using the first measured tidal volume, airway plateau pressure and PEEP values, within 48 h of MV commencement. In the sub-population of patients on controlled MV, without ECMO support, we analysed key pulmonary variables, such as tidal volume, positive end expiratory pressure (PEEP), static driving pressure, inspiratory fraction of oxygen (FiO 2 ), and gas exchange, recorded during routine clinical practice and only. Tidal volume was reported in mL/kg of predicted body weight (PBW) [16].

Data collection
After enrolment, data on demographics, comorbidities, clinical symptoms and laboratory results were collected by clinical and research staff of the participating ICUs in an electronic case report form [15]. Details of respiratory and hemodynamic support, physiological variables, and laboratory results were collected daily. Of note, the worst daily values were preferentially recorded. The duration of MV and ICU stay, and hospital mortality were recorded. Analysis of daily data was restricted to the first seven days from commencement of MV.

Statistical analyses
Descriptive statistics summarised demographics, clinical signs on ICU admission, ICU management and clinical outcomes for the overall study cohort and subjects with baseline compliance measured within the first 48 h of controlled MV. Statistics were reported as medians (interquartile range) for continuous variables and numbers (percentage) for categorical variables. Linear regression was applied to summarise associations between baseline compliance with body mass index (BMI) (including interaction between BMI and sex), days from symptom onset to MV commencement and PaO 2 /FiO 2 , adjusted for BMI. Linear mixed modelling was used to investigate trends in compliance over time and associations with key respiratory parameters during the first 7 days of controlled MV. Models assumed a linear effect for days and a random intercept per subject to account for repeated measures. Consistent with exploratory analyses, BMI was included as a fixed effect to adjust for potential confounding in the clinical characteristics and management of patients with different BMI. Hypothesis testing was applied to all fixed effects, assuming a 5% level of statistical significance. Results were summarised graphically with uncertainty in estimated trends represented by 95% prediction intervals. Expected patient outcomes including length of ICU stay, duration of MV and risk of ICU mortality versus discharge were examined using multi-state modelling [17]. Compared with exploratory analyses of clinical outcomes, the multistate model accounted for ICU discharge and death as competing events and allowed data from all patients to be included, regardless of study follow-up time. The model comprised of four states, to describe patients prior to commencement of MV (non MV), on mechanical ventilation (MV), ICU discharged (Discharge) and mortality (Death). States were presented as percentage and standard error (SE) in the text. Patients extubated before death or discharge were assumed to transition between MV an non-MV states. State transitions were modelled by Cox proportional hazards, with patients censored at last known follow-up, up to 28 days from ICU admission. Follow-up analysis considered Cox proportional hazard regression to examine associations between baseline compliance and competing risks of ICU mortality and discharge, following commencement of MV. Baseline compliance was included as a linear effect, with age, sex, BMI and comorbidities (hypertension, chronic cardiac disease, chronic kidney disease) as additional covariates and adjusted for recruiting centre. A shared frailty term (Gamma distributed) was included to account for residual variation between study sites. Analyses were conducted using R version 3.6.2 or higher (The R Foundation).

Results
We studied 745 patients from 22 countries, who required admission to the ICU and MV from January 14 to December 31, 2020, and presented at least one value of C RS within the first seven days of MV. Among those, 597 (80%) had laboratory-confirmed diagnosis of SARS-CoV2 infection, while in 148 (20%), infection was clinically suspected. Enrolment rate, since January 2020, is reported in Fig. 1. C RS , within 48 h from endotracheal intubation, was available in 649 patients (Fig. 2). No association between C RS and days from onset of symptoms to commencement of MV was found (Fig. 3). Median C RS (IQR), within the first 48 h of mechanical ventilation, was 34.1 mL/cmH 2 O (26.4-44.0) and PaO 2 /FiO 2 113.0 mmHg (84.0-161.3), without any linear association between these parameters. In particular, 16%, 46% and 38% of the patients presented  with mild, moderate or severe hypoxemia, respectively (Fig. 4a). Female sex was associated with a significantly lower C RS than in males (95% CI of difference between genders: − 11.8 to − 7.4 mL/cmH 2 O p < 0.001) (Fig. 4b).
Females also presented higher body mass index (BMI) (95% CI of difference between males and females: − 1.9 to − 5.5, p < 0.001), but as shown in Fig. 5, C RS and BMI were not linearly associated. Our model estimated that C RS was 37.57 cmH 2 O/mL (95% CI 36.5-38.6) upon commencement of MV (Fig. 6), with further worsening in the first seven days of MV (estimated decrease − 0.31 cmH 2 O/mL per day, 95% CI − 0.48 to − 0.14, p < 0.001). In addition, as detailed in Fig. 7, PaCO 2 , tidal volume, PEEP, driving pressure and FiO 2 significantly varied across the range of C RS , and a significant association was found between inspiratory plateau pressure and C RS changes (Fig. 8).
Baseline characteristics upon ICU admission, applied interventions and outcomes, are summarized in Table 1.

Discussion
This large observational report from intensive care units throughout the world found that initial static respiratory system compliance was only associated with hazard of being discharged from the ICU within 28 days. The duration from onset of symptoms to commencement of MV did not influence C RS , and interestingly lower C RS was found in female patients. In the evaluated population, neuromuscular blocking agents and prone position were commonly applied and ventilatory management Fig. 4 a Linear regression analysis of arterial partial pressure of oxygen (PaO 2 /FiO 2 ) and respiratory system compliance (C RS ), based on the first measurement obtained within 48 h from commencement of mechanical ventilation, with an interaction of gender and adjusted for body mass index (BMI). No statistically significant association was found between PaO 2 /FiO 2 and C RS (estimate 0.49, 95% CI − 0.09-1.07 p = 0.100). Typical acute respiratory distress syndrome stratification groups [35] (severe, moderate and mild based on levels of hypoxemia) are highlighted in dark, medium and light grey, respectively. b Static respiratory system compliance (C RS ) distribution by sex, based on the first measurement obtained within 48 h from commencement of mechanical ventilation. Dashed black lines depict median values for females and males In comparison with previous reports on ARDS patients without COVID-19 [18], we similarly found that the majority of patients exhibited moderate hypoxemia, even when presented higher C RS . We also noted a larger range of C RS in line with previous studies [7,8], but in contrast with values from a larger COVID-19 ARDS series from Spain [6]. Considering that we focused our analysis on static compliance of the respiratory system, without partitioning into the pulmonary and chest wall components [19,20], it is interesting that C RS was not associated with BMI, suggesting that patients with higher BMI potentially presented also with higher lung compliance. Irrespective, we found lower C RS in female patients, who also presented higher BMIs. To the best of our knowledge, no studies have systematically investigated the effects of gender/BMI on COVID-19 severity; thus, whether obesity might be a crucial risk factor for ICU admission and mechanical ventilation, specifically in female patients, and its effects on lung compliance should be further explored. We also found that throughout the range of C RS values, plateau pressure was within what is typically presumed as lung protective ranges [21], but this resulted in potentially harmful driving pressures, specifically for patients with the lowest C RS values. As many of these patients were obese, this raises the question of whether these modest pressures might have increased the risk of pulmonary derecruitment, or in patients with normal BMI, the resulting driving pressure might have been related to pulmonary overdistention. These factors could have contributed to sustained hypoxemia and impaired lung function throughout the study period. In such circumstances, it is questionable whether MV guided by oesophageal pressure monitoring may have some benefits [22], but more research is needed to corroborate such reasoning.
Phenotypic subsets of COVID-19-associated ARDS have been proposed [9,13,[23][24][25]. Recent study has also explored whether C RS -related phenotype patterns existed among patients with ARDS before the COVID-19 pandemic [26]. Various investigators [7,27], who did not find significant C RS variability among COVID-19 patients requiring MV, questioned the overall clinical value of C RS in the COVID-19 population. In a very small case series, Gattinoni et al [9] found an initial C RS of 50 mL/ cmH 2 O, but high levels of shunt fraction that could have explained the resulting severe hypoxemia. In subsequent study, Chiumello and collaborators found higher C RS in patient with COVID-19 ARDS and ARDS caused by  other injuries, while matching for similar levels of PaO 2 / FiO 2 [12]. Interestingly, these findings were in line with computed tomography studies results, corroborating higher proportion of normally aerated tissue in COVID-19 ARDS. In similar reports, heterogeneous pathophysiology among patients with different levels of pulmonary compliance has been implied [10,25]. As corroborated by landmark post-mortem studies [28] and clinical studies [7,29], SARS-CoV-2 heterogeneously affects pulmonary ventilation and perfusion. Hence, it could be argued that the use of C RS as key pathophysiological parameter to predict clinical evolution might be over simplistic and in-depth characterization of pulmonary pathophysiology should be recommended for COVID-19 patients, specifically when obese. Interestingly, our report is the first that specifically focused on the dynamics of C RS , rather than only baseline C RS . We found that C RS was not related to the duration from the onset of symptoms to commencement of MV, emphasising the need for inclusive data on mechanisms of lung injury in not ventilated COVID-19 patients [30]. The median C RS value found in our population was 34.1 mL/cmH 2 O, similar to findings by Ferrando et al. [6], not dissimilar to findings by Bellani et al. on patients with non-COVID-19 ARDS [31], but lower than figures recently reported by Grasselli [7] and Grieco [32] in COVID-19 patients. In addition, we found a further decrease in C RS during the first week of MV. This could have been related to the specific ventilatory management in our reported population, but such discrepancy further highlights the need of a comprehensive appraisal of pulmonary and chest wall mechanics in COVID-19 patients [20]. One of the most striking results was the continued use of high PEEP over the first seven days of MV, even in patients with high compliance. This seems counterintuitive, given that current recommendations in ARDS suggest decreasing PEEP, especially in the face of high compliance. As hypoxemia persisted even with high PEEP and high compliance, our results add to the hypothesis that maintaining high PEEP may worsen gas exchange from lung overdistension, resulting in increased dead space and intrapulmonary shunting. Other authors have speculated that using high levels of PEEP in COVID-19 patients with low recruitability may be detrimental, and that lowering PEEP may improve gas exchange and limit ventilator-induced lung injury [33]. Our results in this large cohort of patients from multiple global areas support this theory. Finally, we found that patients required two weeks of MV, and 28-day mortality in the overall population was 35.6%, with hospital mortality up to 40%. These figures are in line with mortality rates reported by Grasselli [7] in the subgroups characterized by low D-dimer, and mortality in severe-moderate COVID-19 ARDS, as corroborated by Ferrando [6]. Nevertheless, we found that C RS was only associated with the discharge from ICU within 28 days. Thus, the marginal clinical Percentages are calculated for non-missing data C RS , static compliance of respiratory system; CRP, c-reactive protein; MV, mechanical ventilation; ICU, intensive care unit; IQR, interquartile range; ECMO, extracorporeal membrane oxygenation a Administration of neuromuscular blockade drugs administered during the first day of invasive mechanical ventilation was not included in the analysis  25:199 value of C RS as a predictor of mortality in COVID-19 patients calls for urgent identification of valuable markers that could inclusively describe pulmonary derangement and guide personalized treatment.

Strengths and limitations
Collaborations between international data collection efforts have the ability to answer many questions related to COVID 19 and to pave the way for future novel diseases to achieve rapid and global data access to help guide best practice. The international COVID-19 Critical Care Consortium study [15], in collaboration with the ISARIC/SPRINT-SARI networks [34], provides inferences not limited by ventilatory management specific to small patient cohort or single-country studies. In addition, in comparison with previous studies, we provided more granular data to inclusively appraise the dynamics of C RS in COVID-19 patients on MV and to study its association with laboratory, and clinical features. A few limitations of our observational study should also be emphasized. First, we centred our analysis on COVID-19 patients, without comparisons against previous repositories of patients with ARDS from different aetiologies. Yet, we provided a wide-ranging discussion of the characteristics of our population in the context of previous analyses in ARDS patients. Second, inferences on pulmonary perfusion disorders in our population can only be speculative, since D-dimer was only available in a small subset of patients (Table 1). Third, as reported by the enrolment rate ( Fig. 1 Supplemental Digital Content), patients were mostly enrolled in the early phase of the pandemic, hence extrapolations from our findings should take into account potential biases related to overwhelmed critical care services. Fourthly, it is important to emphasise that we centred our analysis on C RS , but due to the complex respiratory pathophysiology in COVID-19 patients and the high percentage of patients with increased BMI, the use of oesophageal pressure monitoring to fully describe lung and chest wall compliances is advisable and should be prioritised in future investigations. Fifth, the majority of patients were admitted in centers located in North America, Europe and South America. Although these findings are in line with the global distribution of COVID-19 cases, extrapolations of our findings in other regions should be applied cautiously.

Conclusions
Our comprehensive appraisal of COVID-19 patients on MV from a large international observational study implies that expected C RS within 48 h from commencement of MV is not influenced by the duration from onset of symptoms to commencement of MV, but after intubation, a further decrease in C RS might be expected during the first week of ventilation. In addition, baseline C RS is associated with the chance of being discharged from the ICU within 28 days, but it is not a predictive marker of 28-day mortality. Based on potential inferences from our findings, future studies that could provide an in-depth characterization of lungs and chest wall compliance in COVID-19 patients will be critical to guide best practice in ventilatory management.