Abstract 202: Association Between Implantable Ventricular Assist Device Receipt at Medicare Destination Therapy Centers and Mortality: An Instrumental Variable Analysis
Background: Comparing the effectiveness of contemporary implantable ventricular assist devices (VAD) with medical therapy using observational data is challenging because of treatment selection bias.
Objective: To use instrumental variable analysis to examine 1-year mortality of elderly Medicare beneficiaries with advanced heart failure receiving contemporary implantable VADs compared with conservative management.
Methods: We identified the first hospitalization for heart failure for at least 14 days between January 1, 2009, and December 31, 2011, among all Medicare beneficiaries 65 to 80 years of age, as well as receipt of an implantable VAD during that hospitalization. We estimated risk-adjusted differences in 1-year mortality for VAD versus medical therapy using a 2-stage least squares instrumental variable analysis method. The date that a hospital was certified by Medicare to implant VADs as permanent circulatory support (“destination therapy facility”) was used as an instrument for VAD receipt, since this variable directly influences the use of VAD therapy. Patients were censored at the date of heart transplantation or death, or at the end of the study period.
Results: Of 33,483 beneficiaries hospitalized, 827 (2.5%) underwent VAD implantation at 104 hospitals, 30 of which became destination therapy centers between 2009 and 2011. Among VAD recipients, 71 (8.5%) proceeded to heart transplantation. Unadjusted 1-year mortality rates were 26.5% for VAD versus 51.9% for medical therapy (risk difference,-25.4%; P < 0.001). The adjusted risk difference was -23.6% (95% CI -27.0% to -20.1%, P < 0.001). Using instrumental variable analysis, VAD receipt was associated with a significantly larger decrease in 1-year mortality (risk difference, -40.6%, 95% CI -52.2% to -29.0%; P < 0.001).
Conclusions: When confounding was controlled with instrumental variable analysis, VAD implantation among elderly Medicare beneficiaries, on the margin, was associated with a 41% reduction in 1-year mortality compared with medical therapy, consistent with randomized trial results. Use of a hospital’s “destination therapy” certification date as an instrumental variable may facilitate future comparative effectiveness analyses of VAD therapy.
Author Disclosures: J. Baras Shreibati: None. L.C. Baker: None. D. Banerjee: None. M.A. Hlatky: None.
This research has received full or partial funding support from the American Heart Association, Western States Affiliate (California, Nevada & Utah).
- © 2014 by American Heart Association, Inc.