IntroductionBleeding and thrombosis are major causes of morbidity and mortality in pediatric extracorporeal membrane oxygenation (ECMO), yet their evolving, time-dependent risks remain poorly defined. This study evaluated clinical predictors of bleeding and thrombotic events using time-to-event analysis.MethodsA retrospective cohort study was conducted using the Pediatric ECMO Outcomes Registry (PEDECOR) from October 2011 to September 2024. Patients aged ≤18 years undergoing their first ECMO run were included. Two time-varying Cox proportional hazards models assessed associations between prior complications and time to first bleed or thrombosis, adjusting for clinical and ECMO-related variables.ResultsAmong 1444 patients, median age was 0.4 years (IQR: 0.0-4.4), and 53.6% were male. Bleeding occurred in 65% and thrombosis in 28%. Time to first bleeding events had a median of 2 days (IQR: 0-4) while time to first thrombosis had a median of 3 days (IQR: 2-6). Prior thrombosis was associated with lower early bleeding risk (HR: 0.39, 95% CI: 0.25-0.62, p < 0.001), which reversed over time (interaction HR: 1.44, 95% CI: 1.08-1.92, p = 0.013; crossover at 12.4 days). Similarly, prior bleeding was associated with reduced early thrombosis risk (HR: 0.49, 95% CI: 0.34-0.71, p < 0.001), but reversed over time (interaction HR: 1.63, 95% CI: 1.26-2.11, p < 0.001; crossover at 4.3 days). Central cannulation, cardiac surgery, and non-pulmonary indications were associated with bleeding; sepsis was associated with thrombosis. Anticoagulant type and sex were not significant.ConclusionsBleeding and thrombosis are common and time-dependent in pediatric ECMO. These findings support dynamic anticoagulation strategies tailored to evolving risk.
Keywords: anticoagulation; bleeding; extracorporeal membrane oxygenation; thrombosis; time-to-event analysis.