Abstract 11621: Impact of Data Reporting on Outcome Analyses of Intra-Aortic Balloon Pump versus Impella Devices in Bridging Adult Heart Transplant Patients

Abstract

Introduction: There is a paucity of evidence to guide temporary mechanic circulatory support (tMCS) as bridge to heart transplant (HTx). Question: Can matching techniques reveal device-related effects on peri-HTx outcomes between IABP (Intra-Aortic Balloon Pump) and Impella? Methods: Using the UNOS registry, this retrospective study selected adult patients listed as status 2 on IABP or Impella (V5.0 or V5.5) from 10/2018 to 1/2023. We assessed demographic, medical, hemodynamic, and biochemical data. Primary outcome was failure of status 2 listing strategy (composite of waitlist mortality, escalation of tMCS, and deactivation or delisting due to medical instability). Propensity scores were calculated using logistic regression and matching with a k-nearest neighbors classifier. Results: 3505 IABP and 648 impella patients were identified based on missingness: (1) No missing data (n = 991, 23%), (2) missing hemodynamic data (n = 1606, 39%), (3) missing both hemodynamic and biochemical data(n = 1664, 40%), and (4) the completedataset (n = 4153). Prior to matching,, failure on status 2 occurred in 12.1% of all patients, 11% of IABP patients, and 17.7% of impella patients in group 4 (p < 0.01). After matching, only in group 4 did impella patients have a significantly higher rate of failure on status 2 (12.6 vs 18%, p < 0.01). Average treatment effect for impella on failure of status 2 was nonsignificant for group 1(ATE 0.06, 95%CI -0.016-0.166) and significant for group 2(ATE 0.06; 95% 0.001-0.125), group 3 (ATE 0.05, 95%CI 0.012-0.094), and group 4 (ATE 0.067, 95%CI 0.037-0.102) cohorts, favoring IABP. There was an increase in request for status 2 by exception in group 4 for impella patients (9.8% vs 8.6%, p=0.04). Conclusion: IABP versus Impella had minimal to no impact on failure on status 2 for matched cohorts. Differences appeared when either biochemical or hemodynamic data were omitted. Complete data availability is crucial for developing trustworthy and actionable models.