Patients’ selection
Patients with COVID-19-associated ARDS (COVID-19 cohort). Adult patients admitted from March 3 to April 27, 2020, to two French tertiary care teaching medical ICUs (University hospitals of Angers and Strasbourg, France) and intubated for COVID-19-associated ARDS, were prospectively included within 24 h after ARDS diagnosis for longitudinal physiology assessment. ARDS was defined according to the Berlin definition criteria [3]. SARS-Cov-2 infection was confirmed by real-time reverse transcriptase-polymerase chain reaction (RT-PCR) assay of nasal swabs or lower respiratory tract samples (bronchoalveolar lavage or endotracheal aspirate). Exclusion criteria were age lower than 18 years and use of extracorporeal membrane oxygenation (ECMO) within 24 h after ARDS diagnosis. Some of these patients have been included in previously published studies [9, 17].
Patients with pulmonary non-COVID-19 ARDS (non-COVID-19 cohort). Patients with non-COVID-19-associated ARDS came from the Express study, a large randomized control trial performed from September 2002 to December 2005, and were eligible as control patients [19]. In brief, patients with ARDS or acute lung injury using the American-European Consensus Conference on ARDS criteria [20] were enrolled in the study within 48 h after ARDS diagnosis. Patients were then randomly assigned to two different positive end-expiratory pressure (PEEP) titration strategies: PEEP was set to a level of 5 to 9 cmH2O in the minimal distension strategy or to a level set to reach a plateau pressure of 28 to 30 cmH2O in the increased recruitment strategy.
Ventilation and sedation strategies
Both in the two centers and in the Express trial, recommendations for initial management included a deep sedation and the use of neuromuscular blockers for 24 to 48 h. Patients were ventilated in volume-controlled mode with a tidal volume of 6 ml kg−1 of predicted body weight (PBW) and a respiratory rate up to 35 min−1, adjusted according to arterial pH (objective between 7.30 and 7.45). The fraction of inspired oxygen (FiO2) was set for an arterial oxygen saturation between 88 and 98%.
In the COVID-19 cohort, PEEP setting was left to the discretion of attending physician according to gas exchange and hemodynamic tolerance with an upper limit of plateau pressure of 28 cmH2O, similar to Express.
All patients were switched to pressure-support ventilation when oxygenation improved and PEEP level was decreased to 5–8 cmH2O.
The COVID-19 patients were ventilated using a heated humidifier or a heat and moisture exchanger (HME, Humid-Vent Compact, Teleflex, Athlone, Ireland, dead space = 35 ml or Clear Therm 3 Filter, Intersurgical, Wokingham, UK, dead space = 59 ml). All the patients with non-COVID-19-associated ARDS were ventilated using a heated humidifier.
Data collection
Day 0 was defined as the first calendar day after the onset of ARDS in the COVID-19 cohort or as the day of inclusion in Express trial in the non-COVID-19 cohort (mean time from the onset of ARDS to inclusion = 26.1 ± 23.1 h in Express [19]).
Baseline characteristics (including age, body metrics, simplified acute physiologic score II (SAPS II) [21], partial pressure of arterial oxygen (PaO2), FiO2, partial pressure of arterial carbon dioxide (PaCO2), set tidal volume (Vt), measured respiratory rate, measured minute ventilation, set PEEP and plateau pressure) were collected on day 0 in the two cohorts.
The type of humidification device, HME or heated humidifier was also recorded in the COVID-19 cohort.
The following parameters were recorded at days 1, 3 and 7 in the two cohorts (values measured from 6 to 12 am): PaO2, FiO2, PaCO2, set Vt, measured respiratory rate, measured minute ventilation, set PEEP and plateau pressure (measured by performing an inspiratory hold of 0.2 to 0.3 s). The use of prone positioning and inhaled nitric oxide before day 28 was also recorded.
The diagnosis of thromboembolic event (including deep venous thrombosis on Doppler Ultra Sound or acute pulmonary embolism on CT pulmonary angiography) before day 28 was recorded in the COVID-19 cohort.
Mortality was assessed at day 28 in the two cohorts.
Calculated parameters
PBW was calculated using the following formula: PBW (in kg) = 50 + (0.91 × [height in cm − 152.4]) in men and PBW = 45.5 + (0.91 × [height in cm − 152.4]) in women [22].
The alveolar-arterial oxygen gradient (A-a O2 gradient) was estimated as follows:
A-a O2 gradient = ([(PB-PH2O) × FiO2) − (PaCO2 (mmHg)/RQ)] − PaO2 (mmHg)) where PB is the barometric pressure, PH2O the partial pressure of water and RQ the respiratory quotient. PB, PH2O and RQ were considered as equal to 760 mmHg, 47 mmHg and 0.8, respectively.
Estimated CRS was computed as tidal volume divided by the difference between plateau pressure and set PEEP.
VR was computed as minute ventilation (ml/min) × PaCO2 (mmHg)]/(PBW (kg) × 100 × 37.5) [23].
PaO2/FiO2, A-a O2 gradient, CRS and VR were calculated at days 1, 3 and 7.
Statistical analysis
To select well-balanced subsets of patients from the COVID-19-associated ARDS cohort and non-COVID-19-associated ARDS cohort, the following covariates were identified to build a propensity-score: age, SAPS II score, PaO2/FiO2 ratio and PEEP level on day 0 [24]. The closest controls (from the non-COVID-19-associated ARDS cohort) for COVID-19 cases were identified with the smallest average absolute distance across all the matched pairs using the “optimal” method package MatchIt [24, 25]. Only controls with pulmonary non-COVID-19-associated ARDS were kept in the final analysis sample (see details in Additional file 1).
Results are presented as median [interquartile range] or number (%). Baseline characteristics and ventilatory parameters at days 1, 3 and 7 were compared between the two groups using Mann–Whitney test for quantitative variables and Chi-square test for categorical variables. A Wilcoxon signed rank test was used to compare variables between day 1 and day 7. Correlations between ventilatory parameters were assessed using Spearman test. To identify variables associated with CRS and VR at day 1 and day 7 successively, multiple linear regression models were built separately for CRS and VR-dependent variables, including COVID-19 diagnosis, set PEEP, PaO2/FiO2 ratio as independent variables and additionally humidification device for VR. These independent variables were predefined based on a physiological reasoning. The mortality at day 28 was compared between the patients with VR at day 1 lower or higher than 2 in the patients with COVID-19 and in those with pulmonary non-COVID-19-associated ARDS.
All tests were performed with a type I error set at 0.05. The statistical analysis was performed using R version 3.6.2 (R Core Team (2019), R: a language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.) and Prism (GraphPad Software v5.0b, La Jolla, CA, USA).