From March 5 to May 15, 2020, a total of 156 critically ill COVID-19 patients have been treated at Niguarda Hospital. Among these, 93 patients were admitted to the two study units: 52 to the standard ICU and 41 to the OR ICU. Four patients were excluded from the analysis (three patients did not need invasive mechanical ventilation; one patient was transferred within 24 hours). Eighty-nine patients were included in the study. Most of the patients in both groups (AM group and ICU-VENT group) were admitted to the ICUs during the month of March: 16 (94%) for the AM group and 58 (81%) for the ICU-VENT group (p=0.28).
For 17 patients (19%), an Anesthesia Machine was used (AM group), while for 72 (81%), an ICU ventilator was available (ICU-VENT group). Baseline characteristics are reported in Table 1. The only difference of note between groups with respect to staffing was that patients in the OR ICU had a slightly higher nurse-to-patient ratio. (See “ICU Staffing” in Supplemental Digital Content 1).
Patients in the AM group showed similar COVID-19 severity at admission compared to the ICU-VENT group (Table 2).
The degree of respiratory impairment and mechanical ventilation requirement were similar between the two cohorts. Although not statistically significant, the AM group showed slight lower levels of Mean Arterial Pressure at ICU admission (71 [67-76] mmHg vs. 78 [70-86] mmHg; AM vs. ICU-VENT group, respectively; p=0.050). No significant differences in mean arterial pressure were observed after day one between the ICU-VENT and AM groups (See Figure 1 in Supplemental Digital Content 2).
The intensity of treatment and the use of rescue therapies were comparable between the two groups (Table 2 and 3).
Mechanical ventilation settings were similar in the AM and ICU-VENT group. All patients received prolonged mechanical ventilation, with no difference between the AM and ICU-VENT groups (12 [4-28] days vs. 14 [10-27] days, AM vs. ICU-VENT group, respectively; p=0.364). Forty-four patients received at least one cycle of prone positioning during their ICU stay, with pronation rates similar between groups (64.7% vs. 46.5%, AM vs. ICU-VENT group, respectively; p=0.280). Inhaled nitric oxide was administered once or more during ICU stay to a total of 10 patients at an average concentration of 40 ppm for 48 hours (17.7% vs. 9.9%, AM vs. ICU-VENT group, respectively; p=0.399). Six patients, all in the ICU-VENT group (8.3%), underwent VV-ECMO support (p=0.591).
Patients in the AM group were more frequently treated, for one day or more, with inhalation anesthetics (0% vs. 82%, p < 0.001). In contrast, continuous intravenous (IV) sedation was used more often in the ICU-VENT group (84.7% vs. 17.7 %, p < 0.001).
Critical COVID-19 patients in the AM group died more frequently compared to those in the ICU-VENT group (Table 3 and Figure 1).
The overall 60-day mortality was 43.8% (39/89 patients). Comparing study groups, patients in the AM group experienced a remarkably reduced 60-day survival (deaths: 12/17 vs. 27/72; mortality rate: 70.6% vs. 37.5%, AM group vs. ICU-VENT group, respectively; p= 0.016). Figure 1 reports the Kaplan-Meier Survival curve at 60-days (log-rank p=0.007). Both the ICU and hospital length-of-stay did not differ between the two groups.
Comparing the two ICUs we report a 60-day mortality of 51.2% in the OR ICU and of 37.5% in the conventional-ICU (p=0,207). Moreover, the 60-day mortality among patients receiving ICU ventilator care (regardless of ICU setting) was identical (37,5%).
Table 1 of Supplemental Digital Content 2 details the causes of death among the two groups.
Care of patients that involved the use of Anesthesia Machines was independently associated with an increased risk of death, adjusting for potential confounding factors in a multivariable regression model (Table 4).
In a univariate analysis (see table 2 in Supplemental Digital Content 2), 60-day mortality was significantly higher in patients who at baseline had the following characteristics: were cared for with Anesthesia Machines, were older, had a higher body mass index, showed higher lactate, higher driving pressure, lower pH, lower hemoglobin, higher bilirubin, higher creatinine level, or those with a history of hypertension, diabetes, COPD or hypercholesterolemia. A higher mean arterial pressure was associated with a limited, although significant, protective effect (HR 0.96 per mmHg, 95% CI 0.93-0.99, p=0.008). Of note, we performed an analysis of blood pressure trends over the course of several ICU days (days 1, 2, and 7) among patients receiving volatile anesthetic and found no evidence suggesting lower blood pressures in this subgroup (see Figure 1 in Supplemental Digital Content 2).
After adjustment for confounders, care that involved the use of an AM was associated with a significantly increased risk of death at 60 days (HR 4.05, 95% CI 1.75-9.33, p= 0.001). Other variables associated with increased risk of death at 60 days after adjustment for confounders included: older age (HR 1.08 per year, 95% CI 1.02-1.13, p= 0.004), creatinine (HR 7.20 per mg/dl, 95% CI 2.57-20.21, p < 0.001), bilirubin levels (HR 1.45 per mg/dl, 95% CI 1.11-1.90, p= 0.007) at ICU admission, and a history of diabetes (HR 4.02, 95% CI 1.63-9.91, p = 0.003; Table 4).
In a sensitivity model in which model covariates were based on inclusion frequency in a bootstrapped sample, the final model included the covariates for creatinine, hypertension and bilirubin in addition to the use of anesthesia machines. In this adjusted model the association between the use of anesthesia machines and 60-day mortality remained robust (HR 3.46, 95% CI 1.57-7.63, p =0.002; Supplementary Digital Content 2 Tables 4 and 5).
The use of Anesthesia Machines for prolonged periods might be associated with the risk of technical failure or airway occlusion (Table 5)
During the study period, two cases of sudden Anesthesia Machine failure were observed that required emergent replacement of the workstation. No technical issues were experienced in the ICU-VENT group. Several episodes of mucus plugging of the endotracheal tube occurred in the AM group. In most instances, the obstruction resolved with vigorous suction, or fiberoptic bronchoscopy. Emergency tube exchange was needed in 3/17 cases (18%), compared to 1/72 (1%) in the ICU-VENT group (p= 0.021). One patient in the AM group died due to sudden complete airway obstruction following the accumulation of secretions at the level of the carina, which could not be effectively and timely relieved.