National trends in volume-outcome relationships for extracorporeal membrane oxygenation

Background: The use of extracorporeal membrane oxygenation (ECMO) has emerged as a common therapy for severe cardiopulmonary dysfunction. We aimed to describe the relationship of institutional volume with patient outcomes and examine transfer status to tertiary ECMO centers. Materials and methods: Using the National Inpatient Sample, we identified adult patients who received ECMO from 2008 to 2014. Individual hospital volume was calculated as tertiles of total institutional discharges for each year independently. Results: Of the total 18,684 adult patients placed on ECMO, 2548 (13.6%), 5278 (28.2%), and 10,858 (58.1%) patients were admitted to low-, medium-, and high-volume centers, respectively. Unadjusted mortality at low-volume hospitals was less than that of medium (43.7% versus 50.3%, P = 0.03) and high-volume hospitals (43.7% versus 55.6%, P < 0.001). Length of stay and cost were reduced at low-volume hospitals compared to both mediumand large-volume institutions (all P < 0.001). In high-volume institutions, transferred patients had greater postpropensity-matched mortality (58.5% versus 53.7%, P = 0.05) and cost ($190,299 versus $168,970, P = 0.009) compared to direct admissions. On exclusion of transferred patients from propensity analysis, mortality remained greater in high-volume compared to low-volume centers (50.2% versus 42.8%, P = 0.04). Predictors of mortality included treatment at high-volume centers, respiratory failure, and cardiogenic shock (all P < 0.001). Conclusions: Our findings show increased in-hospital mortality in high-volume institutions and in patients transferred to tertiary centers. Whether this phenomenon represents selection bias or transfer from another facility deserves further investigation and will aid with the identification of surrogate markers for quality of high-risk interventions. (C) 2018 Elsevier Inc. All rights reserved.