Variations in mortality in Medicare recipients on extracorporeal membrane oxygenation

Background: Extracorporeal membrane oxygenation (ECMO) is a rescue modality against severe cardiac and pulmonary compromise. We sought to assess variation in mortality and associated environmental and infrastructural predictors among Medicare beneficiaries on ECMO. Methods: We used Medicare claims data to evaluate hospitalizations between 2017 and 2019 during which beneficiaries required ECMO. The primary outcome of interest was mortality. We evaluated the influence on mortality of Medicare Case Mix Index (CMI), Medicare Wage Index, hospital size, ECMO cannulations, cardiology volume, region, and gender and modeled necessity and sufficiency relations involving ECMO volume, hospital size, cardiology volume, US region, and the mortality index through qualitative comparative analysis (QCA). Results: 5368 ECMO cases were performed at 306 hospitals. Compared to institutions with a mortality index equal to or below 2, those above this threshold had statistically significant higher number of beds, cardiology volumes, and lower survival percentages (p < 0.05). Moreover, we observed a smaller proportion of institutions with an ECMO volume < 20 (78.3% vs 63.4%), which had mortality index > 2. The QCA analysis indicated that low cardiology volume and central/east location are necessary but not sufficient conditions for a mortality index above 2. Conclusion: Trends in mortality are influenced by prevailing socioeconomic, utilization, infrastructural characteristics, and volume. As such, ECMO mortality may be more accurately predicted by models that account for more factors than clinical parameters alone.