The Impact and Statistical Analysis of a Multifaceted Anticoagulation Strategy in Children Supported on ECMO: Performance and Pitfalls

Publication Date


Journal Title

J Intensive Care Med


BACKGROUND: We sought to determine whether the introduction of a new anticoagulation protocol improved the frequency with which target anticoagulation parameters were met in children supported with extracorporeal membrane oxygenation (ECMO). Additionally, we sought to correlate the results of various tests of anticoagulation with the heparin infusion dose (HID) for patients on ECMO and to evaluate the utility of these anticoagulation monitoring tests for the titration of the HID. METHODS: A retrospective chart review of 18 patients who received ECMO at an academic tertiary care children's hospital. Nine patients who were managed using a new anticoagulation protocol were matched by age and diagnosis with 9 patients managed with the old protocol. We collected data relating to patient demographics, type of extracorporeal support, disease process, and incidence of bleeding or thrombosis. Anticoagulation parameters collected include the activated clotting time (ACT), activated partial thromboplastin time (aPTT), prothrombin time/international normalized ratio, anti-factor Xa level, and antithrombin 3 level along with the HID at each time point. Patient groups were compared using a Generalized Linear Mixed Model, a mixed model analysis of variance, and correlational studies. MAIN RESULTS: The percentage of in-range ACT values was not different between the 2 protocols, whereas the percentage of in-range aPTT values was higher in the new anticoagulation protocol (ACT: 37.7% vs 39.5%; aPTT: 25.1% vs 39.8%). After accounting for repeated and variable measures within patients, the probability of obtaining an in-range ACT and aPTT did not differ significantly between the 2 protocols (ACT: P = .3488; aPTT: P = .16). The mean HID did not differ between the 2 groups (35.0 unit/kg/h vs 37.6 unit/kg/h, P = .56). Correlation coefficients demonstrated a significant inverse correlation between the ACT and the HID in both the groups (old: r = -.22, P < .0001; new: r = -.26, P < .0001). We observed a significant positive correlation between the aPTT and the HID in the historical group (r = .25, P < .0001), but no correlation between the aPTT and the HID in the current group (r = -.02, P = .71). The anti-factor Xa level showed a significantly positive correlation with the HID in the current group (r = .62, P < .0001). CONCLUSIONS: A multipronged monitoring regimen slightly increased the amount of time that anticoagulation parameters were within range. Correlations between the HID and the aPTT differed based on anticoagulation protocol, with a positive correlation in the older protocol and no correlation in the new protocol. This may highlight a problem in study design and analysis that requires further examination. Further trials are needed to assess the most useful markers with which anticoagulation protocols for ECMO can be created, adjusted, and evaluated.

Volume Number


Issue Number




Document Type


EPub Date



Faculty; Northwell Researcher


School of Medicine; Northwell Health

Primary Department

General Pediatrics





For the public and Northwell Health campuses