Association between platelet indices and next-day bleeding and clotting in children and adults on extracorporeal membrane oxygenation: A retrospective study

Abstract

Introduction

Bleeding and clotting phenotypes are common in patients supported with Extracorporeal Membrane Oxygenation (ECMO) and are associated with increased mortality. However, tools to distinguish bleeding from clotting phenotypes remain limited. Readily available platelet indices may help characterize these phenotypes.

Methods

We retrospectively studied patients admitted to neonatal, pediatric, and adult intensive care units (ICUs) at two centers from 2018 to 2022, who were on ECMO and required platelet transfusions. The association between pre-transfusion platelet indices (platelet count, mean platelet volume, platelet mass index) and bleeding or clotting on subsequent day was assessed with day-level univariable and multivariable logistic regression models.

Results

We enrolled 268 patients receiving ECMO support (75% veno-arterial), with a median age was 51 years (IQR 18.9–65.6). A total of 1395 platelet transfusion events were analyzed, of which 18% were followed by bleeding and 5% by thrombosis within 24 h. After multivariable adjustment, higher pre-transfusion platelet counts independently associated with subsequent bleeding (adjusted odds ratio [aOR] 1.001, p = 0.03) but not with subsequent thrombosis (aOR 0.99, p = 0.18). Lower pre-transfusion mean platelet volume (MPV) were independently associated with increased bleeding risk (aOR 0.87, p = 0.049), while MPV was not associated with thrombosis (aOR 1.03, p = 0.47). Platelet mass index (PMI) was not independently associated with bleeding (aOR 1.00, p = 0.80) or thrombosis (aOR 1.00, p = 0.32). All platelet indices demonstrated poor discriminatory performance for predicting bleeding or thrombosis, with area under the curve values ranging from 0.42 to 0.55.

Conclusions

Although platelet count and MPV were independently associated with subsequent bleeding, neither index reliably distinguished bleeding phenotypes. Improved tools are needed to predict hemostatic outcomes in patients supported with ECMO.