Assembling and validating a heart failure-free cohort from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study

Abstract

Background: Studies examining incident heart failure (HF) have been limited to select populations. To examine incident HF with broader generalizability, there is need to assemble a HF-free cohort using a geographically-diverse sample. We aimed to develop and validate a simple medication-based strategy for assembling a HF-free cohort from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study. Methods: We examined REGARDS participants with ≥6 months of Medicare inpatient and outpatient claims data at the time of the baseline in-home study examination. To assemble a HF-free cohort, we identified and excluded participants taking HF-specific medications. To validate this approach, we evaluated event rates among this cohort and assessed diagnostic performance using Medicare claims-based definitions of HF as the referent standard. Results: Among 28,884 eligible participants, 3125 were excluded from the proposed HF-free cohort, leaving a total of 25,759 (89%) participants. Depending on the Medicare definition used as the referent, the negative predictive value of this approach ranged from 95.0-99.2%. Negative predictive value was stable across age, sex, and race strata. Conclusions: The approach to assemble a HF-free cohort in REGARDS can serve as the basis for future studies to examine incident HF in REGARDS and similar studies.