Association of Standardized Liberation Trials and Duration of Venovenous Extracorporeal Membrane Oxygenation in Patients with Acute Respiratory Failure.

Abstract

Rationale: There is a paucity of evidence around strategies to liberate patients from venovenous (VV) extracorporeal membrane oxygenation (ECMO) for acute respiratory failure. Objectives: The primary aim of this study was to determine if adopting standardized liberation trials (SLTs) for VV ECMO is associated with the duration of ECMO. The secondary aim was to identify factors associated with unsafe liberation and the effects of unsafe liberation on mortality to intensive care unit (ICU) discharge. Methods: This was a single-center retrospective cohort study of patients on VV ECMO for severe respiratory failure comparing endpoints between intervention (SLT) and control (no SLT) periods. Results: A total of 262 patients were included in the study, 13% (35 of 262) received SLTs, and 150 patients were decannulated from ECMO. Implementing SLTs was strongly associated with the duration of VV ECMO to first successful liberation trial (hazard ratio [HR], 1.88 [95% confidence interval (CI), 1.16 to 3.06]; P = 0.01) and decannulation (HR, 1.92 [95% CI, 1.0 to 3.06]; P = 0.01) without increasing the frequency of unsafe liberation (21% [5 of 23] with SLTs vs. 19% [24 of 127] without SLTs; odds ratio [OR], 1.19 [95% CI, -0.4 to 3.5]; P = 0.7). Unsafe liberation was strongly associated with ICU mortality (HR, 4.15 [95% CI, 1.24 to 13.9]; P = 0.02). Factors associated with unsafe liberation were respiratory rate (OR, 1.49 per 5 breaths/min increase [95% CI, 1.07 to 2.08]; P = 0.02) and ratio of partial pressure of arterial oxygen to fraction of inspired oxygen (OR, 0.73 per 30 mm Hg increase [95% CI, 0.57 to 0.93]; P = 0.01) immediately before decannulation. Conclusions: Incorporating SLTs was significantly associated with the duration of VV ECMO, without increasing the frequency of unsafe liberation. Unsafe liberation was associated with increased ICU mortality.