Extracorporeal membrane oxygenation (ECMO) provides circulatory and respiratory support in pediatric patients with life-threatening organ failure but can alter coagulation.

To assess the relationship between a plasma thrombin generation (TG) assay and conventional coagulation assays (activated partial thromboplastin time, antifactor Xa [anti-Xa], and D-dimer) in pediatric ECMO.

Platelet-poor plasma was collected every 4 hours during the first 48 hours and then daily after ECMO initiation. TG was measured by calibrated automated thrombography.

Fifteen ECMO (65% venoarterial and 35% venovenous) patients (median age, 3.3 months; IQR, 0.3-17.5) were included in this prospective, observational pilot cohort study. Median ECMO duration was 7 days (IQR, 5.5-20.5). Seven patients experienced bleeding events, and 5 developed circuit- or cannula-associated thrombosis; of these, 2 experienced both. TG parameters showed substantial interindividual variability. After an initial increase in the median thrombin peak, TG potential declined over the first 48 hours. Neither activated partial thromboplastin time nor D-dimer correlated significantly with TG parameters. Anti-Xa activity correlated significantly with TG peak (r = −0.84; P = .001) and endogenous thrombin potential (r = −0.46; P = .01). At anti-Xa levels ≥ 0.4 IU/mL, TG approached zero, indicating little additional suppression of procoagulant activity at higher anti-Xa levels.

Our data show substantial interindividual and temporal variability in TG parameters. Only TG peak and endogenous thrombin potential correlated with anti-Xa activity. The TG assay may capture plasma coagulation features not reflected by routine assays.