Time-varying analyses of survival and outcomes in patients with HeartMate 3 left ventricular assist devices

Eur J Heart Fail. 2025 Jan 9. doi: 10.1002/ejhf.3577. Online ahead of print.

Abstract

Aims: As patients experience longer survival on HeartMate 3 left ventricular assist devices, there is a need to characterize long-term risks of adverse outcomes more precisely. This study characterized temporal variations in risks of mortality and adverse outcomes in patients with a HeartMate 3.

Methods and results: From October 2015 to January 2023, 431 HeartMate 3 devices were implanted at Cleveland Clinic. Survival was estimated to 5 years post-implant. Time-varying risks of death, neurological events, gastrointestinal bleeding, device-related infections, and other adverse events were characterized using multiphase hazard modelling. Survival on HeartMate 3 at 1 and 5 years was 88% and 58%, respectively. Risk of death peaked in the first postoperative month before declining rapidly to a constant, lower hazard. Cumulative number of neurological events/patient at 1 year and 5 years was 0.13 and 0.29, respectively; risk was highest within the first postoperative week, then rapidly declined by 1 month. Cumulative number of gastrointestinal bleeding events/patient at 1 year and 5 years was 0.32 and 0.78, respectively; risk was highest within 1 week postoperatively and gradually declined to a constant risk over the first year. Device-related infections developed in 136 patients. One- and 5-year freedom from device-related infection was 77% and 45%, respectively; risk was initially low before peaking at 6 months postoperatively and then gradually declining to a steady hazard.

Conclusion: Long-term survival on HeartMate 3 support was favourable in a large single-centre cohort. Strategies to reduce early postoperative risk of neurological events and late risks of gastrointestinal bleeding, infections and other adverse events are needed.

Keywords: Hazard function; Long‐term outcomes; Mechanical circulatory support; Single centre.