Secondary Logo

Journal Logo

Institutional members access full text with Ovid®

Risk Assessment in Patients with a Left Ventricular Assist Device Across INTERMACS Profiles Using Bayesian Analysis

Kanwar, Manreet K.*; Lohmueller, Lisa C.; Teuteberg, Jeffrey; Kormos, Robert L.§; Rogers, Joseph G.; Benza, Raymond L.*; Lindenfeld, Joann; McIlvennan, Colleen#; Bailey, Stephen H.*; Murali, Srinivas*; Antaki, James F.**

doi: 10.1097/MAT.0000000000000910
Original Article: PDF Only

Current risk stratification models to predict outcomes after a left ventricular assist device (LVAD) are limited in scope. We assessed the performance of Bayesian models to stratify post-LVAD mortality across various International Registry for Mechanically Assisted Circulatory Support (INTERMACS or IM) Profiles, device types, and implant strategies. We performed a retrospective analysis of 10,206 LVAD patients recorded in the IM registry from 2012 to 2016. Using derived Bayesian algorithms from 8,222 patients (derivation cohort), we applied the risk-prediction algorithms to the remaining 2,055 patients (validation cohort). Risk of mortality was assessed at 1, 3, and 12 months post implant according to disease severity (IM profiles), device type (axial versus centrifugal) and strategy (bridge to transplantation or destination therapy). Fifteen percentage (n = 308) were categorized as IM profile 1, 36% (n = 752) as profile 2, 33% (n = 672) as profile 3, and 15% (n = 311) as profile 4–7 in the validation cohort. The Bayesian algorithms showed good discrimination for both short-term (1 and 3 months) and long-term (1 year) mortality for patients with severe HF (Profiles 1–3), with the receiver operating characteristic area under the curve (AUC) between 0.63 and 0.74. The algorithms performed reasonably well in both axial and centrifugal devices (AUC, 0.68–0.74), as well as bridge to transplantation or destination therapy indication (AUC, 0.66–0.73). The performance of the Bayesian models at 1 year was superior to the existing risk models. Bayesian algorithms allow for risk stratification after LVAD implantation across different IM profiles, device types, and implant strategies.

From the *Cardiovascular Institute, Allegheny Health Network, Pittsburgh, Pennsylvania

Department of Bioengineering, Carnegie Mellon University, Pittsburgh, Pennsylvania

Cardiovascular Medicine, Stanford University Medical Center, Stanford, California

§Heart and Vascular Institute, University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania

Division of Cardiology, Duke University School of Medicine, Durham, North Carolina

Cardiovascular Medicine, Vanderbilt University Medical Center, Nashville, Tennessee

#Department of Cardiology, University of Colorado, Denver, Colorado

**School of Biomedical Engineering, Cornell University, New York.

Submitted for consideration April 2018; accepted for publication in revised form September 2018.

Disclosure: All authors have no other financial relationships relevant to the contents of this article to disclose.

Funding for this work was provided by the National Institute of Health Division of National Heart, Lung, and Blood Institute grant NIH R01 HL122639 CORA: A Personalized Cardiac Counselor for Optimal Therapy. Data for this study were provided by the International Registry for Mechanical Circulatory Support, funded by the National Heart, Lung and Blood Institute, National Institutes of Health and The Society of Thoracic Surgeons.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML and PDF versions of this article on the journal’s Web site (

Drs. Kanwar and Lohmueller contributed equally to this study.

Correspondence: Manreet Kanwar, Address: 320 E North Ave, Pittsburgh, PA 15212. Email:

Copyright © 2019 by the American Society for Artificial Internal Organs