Cumulative elevation of aPTT predicts time to major bleeding and death in ECMO: A joint longitudinal-survival model

IntroductionAnticoagulation management during extracorporeal membrane oxygenation (ECMO) remains an important challenge, with bleeding rates reaching 30%. Activated partial thromboplastin time (aPTT) is the most used biomarker for heparin titration, however static values may not reflect real-time bleeding risk. We hypothesized that cumulative aPTT exposure offers superior predictive value for bleeding and mortality compared to traditional threshold-based approaches.MethodsIn this retrospective cohort study, we analyzed 109 adult ECMO patients at a single Canadian center (2006-2021). Serial aPTT values were recorded approximately every 6 h. The primary outcome was time to first major bleeding event (BARC ≥3), with death treated as a competing risk. We applied a joint longitudinal-survival modeling framework to evaluate associations between dynamic aPTT trajectories, including slope, variability, and cumulative exposure,and adverse outcomes. Models were adjusted for baseline covariates and compared using deviance information criterion (DIC) and time-dependent AUC.ResultsFifty-one patients (46.8%) experienced major bleeding. The best-performing model incorporated the cumulative exposure to elevated log-transformed aPTT, which significantly predicted both bleeding (HR 2.39, 95% CI: 1.17-4.88, <i>p</i> = 0.0084) and death (HR 7.88, 95% CI: 2.59-23.94, <i>p</i> < 0.0001). This model outperformed static or trend-based approaches (AUC 0.77 at 48 h). There results were robust to sensitivity analysis. Hematocrit and ECMO configuration were also significant covariates.ConclusionCumulative aPTT burden is a strong and independent predictor of bleeding and mortality in ECMO patients. These findings support a shift toward trajectory-based anticoagulation monitoring to enable safer, personalized management in this high-risk population.