I) Basic patient characteristics
Basic patient characteristics are displayed in Table I. As severe COVID-19 affects predominantly men, the COVID-19 group has an excess of male patients (20 of 25 (80%) vs. 16 of 25 (64%). Apart from a non-significant age difference (61,7 years [53,1 – 68,1 years] for COVID-19 vs. 55,7 years [45,5 – 65,4 years] for Influenza, p = 0.08), both groups were well comparable regarding height, weight, body-mass index (BMI) and oxygenation impairment at the beginning of invasive ventilation.
Most frequent comorbidities were adiposity, arterial hypertension, Diabetes mellitus and coronary heart disease.
Both groups showed severe hypoxemia at the time of intubation. Mean arterial oxygen partial pressure before intubation was slightly higher in the influenza group (COVID19: paO2 84.4 mmHg [76.8 – 92.1 mmHg] vs. 93.3 mmHg [71.8 – 114.8 mmHg] for Influenza, p = 0.31). Prior to intubation, patients received either oxygen supplementation via high-flow nasal cannula or non-invasive ventilation. Mean inspiratory oxygen fraction during non-invasive oxygenation was slightly lower in the COVID-19 group (0.6 [0.49 – 0.70], n = 14) compared to control (0.75 [0.53 – 0.84], n = 9].
13 of 25 COVID-19 patients (42%) and 16 of 25 Influenza (64%) patients received veno-venous extracorporeal membrane oxygenation for respiratory failure few days after intubation.
II) Horizontal comparison of ventilation parameters
On the first day of invasive ventilation, there were no significant differences in P/F ratio (127.9 mmHg [112.8 – 161.3 mmHg] for COVID-19 vs. 135.4 mmHg [100.3 – 180.4] mmHg, p = 0.56), positive end-exspiratory pressure (12.0 mbar [9.6 – 13.5 mbar] for COVID-19 vs. 13.0 mbar [9.0 – 15.5 mbar] for Influenza, p = 0.37), plateau pressure (23.5 mbar [20.8 – 29.0 mbar] for COVID-19 vs. 25.0 mbar [24.0 – 28.5 mbar] for Influenza, p = 0.56) or driving pressure (13.5 mbar [10.8 – 16.0 mbar] for COVID-19 vs. 12.0 mbar [11.0 – 14.5 mbar] for Influenza, p = 0.64) between both groups.
Similar ventilation pressures produced higher tidal volumes in relation to predicted body weight in COVID-19 patients (7.69 ml/kgBW [7.12 – 8.12 ml/kgBW]) compared to control (5.12 ml/kgBW [3.14 – 8.40 ml/kgBW], p = 0.06), indicating less impaired ventilatory system compliance in the COVID-19 group. COVID-19 patients required significantly higher ventilation frequencies (20.0 min-1[15.5 – 21.3 min-1] vs. 14.0 min-1 [12.5 – 17.5 min-1], p = 0.011) and minute ventilation (10.7 l/min [7.2 – 12.2 l/min] vs. 6.0 l/min [2.5 – 10.1 l/min], p = 0.013) for sufficient CO2 elimination. Indeed, calculated static compliance of the respiratory system was higher in COVID-19 patients than Influenza patients (40.7 ml/mbar [31.8 – 46.7 ml/mbar] vs. 31.4 ml/mbar [13.7 – 42.8 ml/mbar], p = 0.198) throughout the observation period. We calculated ventilatory ratio (VR) for both groups to assess ventilation efficacy. VR was significantly higher in the COVID-19 cohort (1.57 [1.31 – 1.84] vs. 0.91 [0.44 – 1.38], p = 0.006). For a comprehensive overview of ventilation parameters, s. Tables II, III and Figure I.
III) Longitudinal comparison of ventilation parameters
Immediately after intubation, most patients in both groups received pressure-controlled ventilation (PCV) during deep sedation or relaxation. The proportion of patients with strictly controlled ventilation on the first day was 79% in the COVID-19 group (11/14) and 100% in the Influenza group (9/9). All patients for whom ventilation data from the first day of mechanical ventilation were missing could not be included into the analysis.
As ventilation progressed, an increasing number of patients could be weaned to augmented spontaneous breathing (ASB) with continuous positive airway pressure (CPAP). On day 8, 41% of COVID-19 patients and 38% of Influenza patients breathed spontaneously with pressure support. On day 16, the rate of patients still on controlled ventilation had dropped to 9.5% in the COVID-19 group (2/21) but stayed relatively unchanged at 36% (4/11) in the control group.
Except for few Influenza patients developing prolonged disease, we observed that COVID-19 patients were ventilated much longer than most Influenza patients. Odds Ratio in the overall cohort for breathing free from ventilator 15 days after intubation was 3.5 for Influenza compared to COVID-19.
To assess how ventilatory system compliance developed over time, we performed longitudinal pair-wise t-test comparisons of the initial compliance from the day of intubation to day 8 and day 16 of invasive ventilation.
We detected a continuous decrease of compliance in the COVID-19 cohort, with a significant reduction after 2 weeks of ventilation (47.99 ml/mbar ± 32.80 ml/mbar vs. 24.13 ml/mbar ± 10.70 ml/mbar, mean difference after 15 days being -23.87 ml/mbar ± 32.94 ml/mbar, p = 0.037) (Figure II). In contrast, ventilatory system compliance improved in the Influenza cohort after 7 days (28.32 ml/mbar ± 21.98 ml/mbar vs. 34.85 ml/mbar ± 29.59 ml/mbar, p = 0.10). Since the course of ARDS tended to be much shorter for Influenza patients, we cannot provide a comparison in this cohort over 15 days, as only 2 patients would have qualified for pair-wise comparison between day 1 and 16.
IV) Subgroup analysis – ECMO vs. Non-ECMO
Gradual decline in compliance was particularly present in the subgroup of COVID-19 patients receiving extracorporeal support (mean loss of compliance after 7 days 19.5 ml/mbar ± 42.6 ml/mbar, p = 0.31, mean loss after 15 days: -36.19 ml/mbar ± 41.88 ml/mbar, p = 0.09). COVID-19 patients not on ECMO also deteriorated in compliance, however, the loss of compliance was not as strong (-9.08 ml/mbar ± 4.43 ml/mbar after 15 days, p = 0.010).
After ECMO was initiated in most patients, we checked whether ventilation strategies changed following ECMO initiation. Indeed, for the COVID-19 group, patients on ECMO received significantly higher PEEP on day 8 and 16 compared to patients without ECMO (Day 8: 12.5 mbar ± 2.5 mbar vs. 10.3 mbar ± 2.0 mbar, p = 0.025, Day 16: 12.3 mbar ± 2.3 vs. 7.8 mbar ± 1.8 mbar, p < 0.001). Attempting to limit VILI, patients on ECMO were ventilated with lower driving pressures. This difference was small for day 8, as not all patients had been cannulated up to that day but was significant for day 16 (Day 8: 13.8 mbar ± 4.6 vs. 14.2 mbar ± 2.6 mbar, p = 0.83, Day 16: 11.3 mbar ± 2.3 mbar vs. 16.6 mbar ± 4.5 mbar, p = 0.002).
In contrast, there were no differences between PEEP and driving pressures between patients with and without ECMO for the Influenza group (not shown).
Since patients on ECMO tended to receive lower driving pressures but also seemed to deteriorate in ventilatory system compliance over time, we checked for correlations between a reduction in driving pressure and decreases in compliance over 15 days. We found that decreased driving pressures as a strategy of protective ventilation on ECMO was positively correlated to a loss of compliance after 15 days in bivariate correlation analysis (Pearson’s R = 0.80, p = 0.058) in the COVID-19 group. In contrast, COVID-19 patients not receiving ECMO showed an inverse correlation between change in driving pressure and compliance on day 16 (Pearson’s R = -0.92, p = 0.025). Compared to the COVID-19 group, this correlation was weak for Influenza patients on ECMO between the first and eighth day of ventilation (Pearson-coefficient -0.10, p = 0.88).
We determined ‘death from any cause’ as an endpoint to assess survival rates after 28 and 60 days from the beginning of invasive ventilation.
28-day mortality was 16% for the COVID-19 group compared to 36% in the control group. As Kaplan-Meier-plots demonstrate (s. Figure III A to C), overall survival was slightly better for the COVID-19 group compared to control (40% vs. 48% for Influenza, p = 0.31). Requiring veno-venous ECMO was a negative predictive factor for survival in both groups (50% mortality for COVID-19 on ECMO vs. 27% without ECMO, p = 0.30 / 56% vs 34% mortality for Influenza A/B with and without ECMO, p = 0.31).