At the final date of follow-up, 111 COVID-19 positive ICU patients met study inclusion criteria which included requiring mechanical ventilation and meeting the Berlin criteria for ARDS. The median age for the overall population was 64 years (IQR 17) and the majority were male (76, 68.5%). Most patients were Caucasian (50, 45.1%), followed by Hispanic (39, 35.1%), and Black (15, 13.5%) (Table 1A). Of the total population, 40 patients (36%) were classified as phenotype H (Crs < 30 ml/cm H2O) and 71 (64%) were classified as phenotype L (Crs ≥ 30 ml/cm H2O).
The most common comorbidities among all COVID-19 ICU patients were hypertension (61.3%), diabetes (44.1%), and obesity with a median body mass index 30.8 kg/m2 (IQR 10.3). Baseline characteristics, listed in Table 1A, were similar among phenotypes H and L. Admission vital signs and laboratory values were also similar between the two groups (Table 1B). While not statistically significant, phenotype H had a lower heart rate compared to phenotype L (94.5 beats per minute [IQR25] vs 101 [29], p=0.059) and phenotype H trended toward a lower mean arterial pressure (89 mmHg [26] vs 95 [21], p=0.057).
Both phenotype H and L patients were critically ill. The majority of patients had shock requiring vasopressors during ICU admission (88.9%). All patients had ARDS; 83 (74.8%) had severe ARDS based on Berlin criteria. The mean PaO2/FiO2 ratio for the first 72 hours of mechanical ventilation was significantly lower in phenotype H (115 [87] vs 165 [87], p=0.016). Likewise, the PaO2/FiO2 ratio nadir for the entire mechanical ventilation course was lower in phenotype H than L (63 [32] vs 75 [59], p=0.026).
Complications were similar between both phenotype H and L groups (Table 2). Fifty-one (46.4%) patients had concomitant bacterial pneumonia and 26 (23.4%) had bacteremia. Nearly half (52, 47.3%) had acute kidney injury and 32 (28.8%) required renal replacement therapy. There was no difference among infectious, cardiac, or other complications between the two groups. While not statistically significant, fewer phenotype H patients underwent tracheostomy procedures than phenotype L (2 [5%] vs 12 [16.9%], p=0.070).
Outcomes and Mortality as Related to Respiratory System Compliance
The overall mortality was 57.7% for COVID-19 ARDS patients, with median mechanical ventilator days 10 [IQR11], ICU length of stay 12 days [12], and hospital length of stay 11 days [IQR 15] (Table 3A). There was no difference in ventilator days, ICU length of stay, hospital length of stay, or mortality when comparing phenotypes H and L. Twenty-seven (67.5%) phenotype H patients died as compared to 37 (52.1%) phenotype L patients (p=0.115). While there was no difference in mortality when stratifying respiratory system compliance based on phenotypes H and L, there was a trend toward increasing mortality with decreasing respiratory system compliance when dividing the patients into ordinal groups (Table 3B, Figure 1). There was 40% mortality among patients with Crs > 50 ml/cm H2O vs 55.1% mortality with Crs 30-39.9 ml/cm H2O and 80% mortality with Crs <20 ml/cm H2O.
Multivariable Analysis
In multivariable analysis models one and two, respiratory system compliance was defined as phenotype H and L. Using these models Crs was not associated with mortality and neither was PaO2/FiO2 ratio nadir; whereas abnormal pH was associated with a 2-fold increased risk for mortality (Table 4A). When evaluating Crs ordinally instead of as binary phenotype, a 2-fold increase in mortality was noted in Crs < 20 vs > 50 ml/cm H2O (RR 2.00, 95% CI 0.94-4.27); however, this association did not hold upon adjusting for pH nadir and PaO2/FiO2 ratio nadir (Table 4B). Ordinally, each additional group level (ex. Crs 30-39.9 ml/cm H2O to 40-49.9 ml/cm H2O) was marginally associated with a 14% lower risk of death (RR=0.86, 95% CI 0.72, 1.01, p=0.065), this attenuates (RR=0.94, 95% CI 0.80, 1.11) when adjusting for pH nadir and PaO2/FiO2 ratio nadir.