Abstract P399: Development of a Claims-Based Algorithm to Identify a Coronary Heart Disease Free Population in Medicare

Abstract

Introduction: There is increasing interest in using claims data to study risk for incident coronary heart disease (CHD). However, claims-based algorithms for identifying a population without a history of CHD are not well established. Objective: To develop a claims-based algorithm for identifying a population without previous myocardial infarction (MI) or coronary revascularization. Methods: The REasons for Geographic and Racial Differences in Stroke (REGARDS) study is a national cohort that included a baseline study visit in 2003-2007. Participants linked to Medicare claims were used to identify cohorts with continuous 6 month, 1 year, 2 year, or 3 year coverage prior to the REGARDS baseline (“look-back” periods). History of CHD at baseline was defined as a self-report of a previous MI or cardiac revascularization. We examined Medicare claims prior to baseline to identify evidence of CHD. Approaches used fixed-window look-back periods (e.g., claims within a 1 year window prior to baseline) and all available look-back (e.g. requiring a minimum of 1 year of Medicare coverage, but using all claims available prior to baseline). Among those without CHD claims in Medicare, we calculated the percent who self-reported CHD for the different look-back periods. Results: There were 8,104 participants with 6 months of look-back and no CHD claims. Requiring 1, 2 or 3 year look back periods reduced the number of participants by 1,001 (5%), 2,237 (16%) and 3,157 (26%), respectively. Using fixed-window look-back periods, the percent without CHD claims who self-reported CHD was lower with longer look-back periods (Table). Requiring 6 months of look-back and using all available claims resulted in a lower percent self-reporting CHD than any fixed-window look-back period. The percent self-reporting CHD was only modestly lower when requiring longer all available look-back periods. Conclusion: We found that a 6 month minimum look-back period is adequate for identifying a CHD-free population in Medicare when all available claims data are used.