Pre-cannulation ventilation duration and the trajectory of venovenous extracorporeal membrane oxygenation in COVID-19-related acute respiratory distress syndrome: a nationwide competing-risks cohort study

Abstract Purpose Pre-cannulation ventilation duration is often considered in selecting candidates for venovenous extracorporeal membrane oxygenation, but its relationship with the subsequent support trajectory is uncertain. We examined associations with first liberation, death before liberation, and time on support. Methods We analyzed a nationwide Japanese registry of patients receiving venovenous extracorporeal membrane oxygenation for COVID-19-related acute respiratory distress syndrome. First liberation and death before liberation were treated as competing transitions over 90 days. Cause-specific Cox models were stratified by epidemic period and adjusted for age, sex, and body mass index. Expected time on support was calculated from state-occupancy probabilities. Results Among 1,420 patients, the adjusted hazard ratio for first liberation relative to 0–1 days was 0.87 (95% confidence interval 0.72–1.05) for 2–3 days, 0.65 (0.54–0.78) for 4–7 days, 0.41 (0.32–0.52) for 8–14 days, and 0.58 (0.34–0.98) for at least 15 days. The association was strongest during the first 14 days. Death before liberation, which does not represent overall mortality because follow-up after liberation was short, showed no monotonic gradient below 15 days. Expected time on support was 26.9 days after more than 7 days of ventilation and 18.4 days after 7 days or less. Conclusions Among selected recipients, longer pre-cannulation ventilation, particularly 8–14 days, was associated with slower early liberation and longer support. It may inform the anticipated course and resource requirements after cannulation, but does not define a candidacy threshold.