Respiratory Physiology of COVID-19-Induced Respiratory Failure Compared to ARDS of Other Etiologies

Domenico Luca Grieco; Filippo Bongiovanni; Lu Chen; Luca S. Menga; Salvatore Lucio Cutuli; Gabriele Pintaudi; Simone Carelli; Teresa Michi; Flava Torrini; Gianmarco Lombardi; Gian Marco Anzellotti; Gennaro De Pascale; Andrea Urbani; Maria Grazia Bocci; Eloisa S. Tanzarella; Giuseppe Bello; Antonio M. Dell'Anna; Salvatore M. Maggiore; Laurent Brochard; Massimo Antonelli


Crit Care. 2020;24(529) 

In This Article


This prospective study was conducted in the dedicated COVID-19 ICU of a tertiary care university hospital in Italy during March 2020. Approval was obtained by local institutional review board, and informed consent was obtained according to committee recommendations.


We screened all consecutive adult patients admitted to ICU between March 16 and 27, 2020, who were intubated due to acute hypoxemic respiratory failure with confirmed molecular diagnosis of COVID-19 (positive real-time polymerase chain reaction for viral RNA performed on an upper or lower respiratory tract specimen). Patients fulfilling criteria for moderate and severe ARDS according to the Berlin definition (i.e., PaO2/FiO2 ratio ≤ 200, measured at PEEP = 5 cmH2O)[24,25] were enrolled within 24 h from endotracheal intubation.

Exclusion criteria were as follows: (1) age < 18 years, (2) undrained pneumothorax, and (3) hemodynamic instability, defined as > 30% increase in vasopressor requirement during the previous 6 h or norepinephrine > 0.5 μg/kg/min.

Procedures and Measurements

For each patient, demographics, comorbidities, and 28-day clinical outcome were recorded.

All measurements were conducted in the supine semi-recumbent position within 24 h from endotracheal intubation, before any session of prone positioning. All patients were sedated and paralyzed with cisatracurium continuous infusion at a standard dose of 35 mg/h.[26] Mechanical ventilation was applied in the volume-controlled mode with a heat and moisture exchanger, with the following settings: tidal volume 6 ml/kg of predicted body weight (PBW), inspiratory flow 60 l/min, inspiratory pause 0.3 s, respiratory rate titrated to obtain pH > 7.30 and < 35 breaths per minute, and FiO2 titrated to achieve SpO2 between 90 and 96%.

Two PEEP levels were tested in a sequential order: 15 (or the highest PEEP to obtain plateau pressure ≤ 28 cmH2O) and 5 cmH2O. After 30 min of ventilation with PEEP = 15 cmH2O, arterial blood gasses and hemodynamics were recorded. Inspiratory (1.5 s) and expiratory (4 s) holds were performed, and the following parameters collected:

  • Respiratory mechanics: Peak airway pressure, plateau pressure, and total PEEP were measured. Driving pressure, respiratory system compliance, and its PBW-indexed value were computed. Ventilatory ratio, which is an estimate of dead space fraction, was calculated (tidal volume × respiratory rate × PaCO2)/(PBW × 100 × 37.5).[27]

  • Recruitability: A single-breath derecruitment maneuver was performed by decreasing PEEP by 10 cmH2O;[28,29] exhaled tidal volume after PEEP lowering was recorded, and recruitment-to-inflation ratio was computed[29]—patients with recruitment-to-inflation ratio ≥ 0.5 were considered having high recruitability.

Afterwards, ventilation was resumed with previous settings and PEEP = 5 cmH2O. After 30 min, blood gasses, hemodynamics, and respiratory mechanics were re-assessed as described. Lastly, low-flow (5 l/min) inflation was performed after prolonged exhalation to assess airway closure, and airway opening pressure was recorded if present.[30–33]

Comparison With non-COVID-19 ARDS

COVID-19 patients were compared to a cohort of subjects with moderate-to-severe ARDS from other etiologies who underwent exactly the same procedures in a previous study.[29] Two investigators (DLG and LC) were directly involved in patients' enrolment in both studies. This ensures reproducibility of the measurements and consistency in ventilator settings and circuit setup. COVID-19 patients were matched in 1:1 ratio to patients from the non-COVID-19 ARDS cohort. Matching was based on PaO2/FiO2 (± 20 mmHg), FiO2 (± 0.2), PEEP (± 3 cmH2O), and tidal volume (± 1.5 ml/kg of predicted body weight). For matching, priority was given to PaO2/FiO2 (100% adherence to the criterion), followed by FiO2 (93% adherence to the criterion, for "2 matches" criterion increased to ± 0.4), tidal volume (93% adherence to the criterion, for "2 matches" criterion increased to ± 2.5 ml/kg), and PEEP (90% adherence to the criterion, for "3 matches" criterion increased to ± 5 cmH2O). Individual data of matched subjects are provided in Supplementary Table 1.


The aims of the study were to describe respiratory mechanics, potential for lung recruitment, and response to PEEP in COVID-19-induced ARDS patients and to compare these features to those of patients affected by ARDS of other causes.

Sample Size and Statistical Analysis

At the time of study design, systematic data on respiratory mechanics in COVID-19 patients were lacking. Hence, a convenience sample of 30 consecutive patients was chosen to provide a timely report. Categorical data are reported as number of events (%), and continuous data are displayed as medians [interquartile range]. Comparisons of continuous variables at the two PEEP levels were performed with the T test for paired samples: mean differences [95% confidence intervals, CI95%] are displayed for most significant results. Categorical variables were compared with the McNemar test. Inter-individual variability was calculated as the ratio of standard deviation to mean of the measurements.

Comparisons of continuous variables between COVID-19 and ARDS cohort were performed with the T test for independent samples: mean differences are displayed for significant results. Categorical variables were compared with the chi-square or Fisher exact test, as appropriate.

Correlations were assessed with Pearson's correlation: r and p are provided for each comparison. Results with two-tailed p ≤ 0.05 were considered statistically significant. Statistical analysis was performed with SPSS 20.0 (IBM Corporation, Armonk, NY, USA). Manuscript figures were prepared with GraphPad Prism (La Jolla, CA, USA).