Hospital costs associated with mechanical left ventricular mechanical unloading devices during VA ECMO for adult cardiogenic shock.

Importance: No works to-date have described the financial burden and behaviors of left ventricular mechanical unloading (LVMU) for patients on veno-arterial extracorporeal membrane oxygenation (VA ECMO). Given the uptrending use of VA ECMO, describing its associated cost is essential for its continued uptake. Objective: We describe the inpatient costs of patients who were managed with ECMO for cardiogenic shock (CS) with and without LVMU. Design Setting and Participants: We conducted a retrospective cohort study of adult (age ≥18 years) patients who received ECMO at some point during their hospital stay and were non-post operative patients (e.g. medical CS) using the IBM MarketScan database. Data were extracted from 1/1/2008-12/31/2021. Exposures: The exposure of interest was the additional use of LVMU (intra aortic balloon pump, peripherally inserted left ventricular assist device [pVAD], or "other") added to ECMO. Costs were calculated daily, and modeled according to the daily status of ECMO, LVMU, ECMO+LVMU, or no device. Main Outcomes and Measures: Patient characteristic, including age, sex, comorbidities quantified using the Charlson Comorbidity Index (CCI), etiology of heart failure (acute myocardial infarction [AMI] vs chronic heart failure [CHF]), hospital and intensive care unit (ICU) length of stay (LOS) and total inpatient costs were described using descriptive statics between groups. The outcomes of interest were total inpatient costs. Secondary outcomes included in-hospital mortality, and hospital and intensive care unit (ICU) length of stays (LOS). We stratified patients by receipt of LVMU, and used propensity score matching from patient level characteristics to balance the use of LVMU between groups. Cost outcomes were modeled using mixed effects linear regression clustered on matched groups and reporting incident rate ratios (IRR). LOS and mortality outcomes were modeled using Poisson (IRR) and logistic (adjusted odds ratio [aOR]) regression, respectively, conditional on matched groups. Results: Enrolled patients (n=1,596) were 56 years old (interquartile range [IQR] 47 to 62), had an ICU LOS of 9 (3 to 19) days, and a hospital length of stay of 18 (7 to 35) days, which were not different between groups. Patients who received LVMU had a higher CCI ( Conclusions and Relevance: CS patients managed with VA ECMO + LVMU had significantly increased cost, significantly longer LOS, and significantly decreased mortality compared to ECMO alone. Understanding the impact of LVMU on the cost of an ECMO course will aid appropriate resource allocation.