Tidal volume and mortality during extracorporeal membrane oxygenation for acute respiratory distress syndrome: a multicenter observational cohort study

Abstract

Background: Approximately half of the patients with acute respiratory distress syndrome (ARDS) receiving extracorporeal membrane oxygenation (ECMO) remain ECMO-dependent beyond 14 days after ECMO initiation. The identification of factors associated with mortality during an ECMO run may update prognostic assessment and focus clinical interventions.

Methods: In this observational study, data from 1137 patients with COVID-19 ARDS receiving ECMO support in 29 German centers between January 1st 2020 and July 31st 2021 were analyzed. Multivariable stepwise logistic regression analyses were performed to build survival prediction models with day-by-day data during the first 14 days of an ECMO run. The primary endpoint was all-cause mortality in the intensive care unit.

Results: Mortality in this cohort was high (75%). Patients who remained ECMO-dependent on day 14 of their ECMO run showed comparable mortality to all patients receiving ECMO support on day 1. Yet, factors associated with mortality changed during the first 14 days of ECMO support. On day 1 of ECMO support, only patient age and lactate remained in the final mortality prediction model. On day 14 of an ECMO run, tidal volume was independently associated with mortality (adjusted Odds Ratio 0.693 (95%CI 0.564-0.851), p < 0.001 for 1 mL/kg increase in tidal volume per predicted body weight). The adjusted mortality for patients with a tidal volume below 2 mL/kg on day 14 of their ECMO run was above 80% (lower limit of the 95%CI interval). Higher tidal volume was mainly based on higher respiratory system compliance. Yet, the benefit of higher compliance was not observed in some patients who were still ventilated with very low driving pressures despite remaining ECMO-dependent on day 14 of ECMO support.

Conclusions: Mortality predictors change during the course of an ECMO run. In a cohort with high mortality, on day 14 of ECMO support for ARDS, tidal volume may be an independent predictor of mortality. Further analyses on ventilation strategies in patients who remain ECMO-dependent are needed.

Trial registration number: DRKS00022964, retrospectively registered.

Keywords: ARDS; ECMO; Mortality prediction.

Fig. 1

Study Flow Chart. Definition of abbreviations: ECMO = extracorporeal membrane oxygenation; ARDS = acute respiratory distress syndrome

Fig. 2

Adjusted ICU Mortality based on Tidal Volume on Day 14 of an ECMO run for Patients still receiving ECMO support. Adjusted ICU Mortality based on tidal volume per predicted body weight on day 14 of an ECMO run for patients under controlled ventilation still receiving ECMO support. ICU Mortality was adjusted for all relevant covariates based on stepwise regression analysis as shown in Table 2 (ECMO blood flow on day 14 and lactate on day 14)

Fig. 3

Tidal Volume, Compliance and Driving Pressure on Day 14 of ECMO support. Using K-means cluster analysis (k = 2) we partitioned our data set of all patients under controlled ventilation still receiving ECMO support on day 14 of their ECMO run into clusters based on driving pressure (ΔP) (Clustering A), tidal volume (Vt/PBW) (Clustering B) and respiratory system compliance (Clustering C) and calculated the adjusted odds ratio for mortality. The upper scatter/error-bar diagrams illustrate the median values observed in each cluster. The error bars represent interquartil ranges. At the bottom, the respective adjusted odds ratios for ICU mortality are shown, calculated for each cluster after multivariable adjustment for all relevant covariates based on stepwise regression analysis as shown in Table 2 (ECMO blood flow on day 14 and lactate on day 14). Error bars represent 95% confidence intervals. Definition of abbreviations: Vt = tidal volume; PBW = predicted body weight; C_rs = respiratory system compliance; ΔP = driving pressure; aOR = adjusted odds ratio