Abstract P393: Completeness of Medicare Pharmacy Claims for Lipid-Lowering Medications: the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study.
BACKGROUND: Pharmacy data are increasingly being used to identify medication utilization in epidemiological studies and as a measure of quality of care. The objective of this analysis was to investigate the completeness of Medicare pharmacy claims for lipid-lowering medications (LLM) in a population-based cohort study.
METHODS: We analyzed data from 891 REasons for Geographic And Racial Differences in Stroke (REGARDS) study participants ≥ 65 years of age with Medicare-linked data. Our primary analyses were restricted to participants enrolled in 2006-2007 with Medicare Part D (pharmacy) coverage for 120 days prior to their REGARDS in-home examination at baseline. Secondary analyses were restricted to participants with 182 days of continuous Medicare Part D coverage prior to the in-home examination (n=852). During the baseline examination, bottles for prescription medications taken within the previous 2 weeks were inventoried and drug names were recorded. LLM included statins, ezetimibe, niacin, fibrates and bile acid sequestrants.
RESULTS: Among the 370 (41.5%) participants with a bottle for a LLM at their REGARDS study examination, 70 (18.9%) participants did not have a Medicare claim for a LLM within the preceding 120 days. Overall agreement between medication inventory and Medicare claims for LLM was 86.8% (95% confidence interval [CI]: 84.4%-88.9%) and agreement above that expected by chance (kappa statistic) was 0.72 (95% CI: 0.68-0.77). Agreement was similar across sub-groups defined by age, self-reported income and education, cognitive status, depressive symptoms, year of enrollment or dual (Medicaid-Medicare) eligible status. The Kappa statistic was lower for black males (0.49; 95% CI: 0.70-0.84) compared with black females (0.74; 95% CI: 0.65-0.82), white males (0.73; 95% CI: 0.65-0.82) and white females (0.77; 95% CI: 0.70-0.84). Among participants with a statin identified in their REGARDS medication inventory, 17.8% did not have a Medicare claim for a statin within 120 days prior to their examination. Overall agreement and the kappa statistic for the comparison between medication inventory and Medicare claims for statins were 87.9% (95% CI: 85.5%-90.0%) and 0.74 (95% CI: 0.69 - 0.79), respectively. Among participants with continuous Medicare Part D coverage for 182 days prior to their REGARDS baseline examination, 14.4% of those with a bottle for LLM had no Medicare claims.
CONCLUSIONS: A substantial percentage of participants with a LLM identified through the REGARDS medication inventory had no Medicare claims for LLM. Future studies should investigate reasons for the discordance between Medicare claims for LLM and medication inventory, particularly among black males.
Author Disclosures: L.D. Colantonio: None. S.T. Kent: H. Other; Significant; Salary support from Amgen. M.M. Safford: G. Consultant/Advisory Board; Modest; diaDexus. H. Other; Significant; Salary support from Amgen. E. Delzell: None. M.L. Kilgore: None. J.R. Curtis: None. P. Muntner: C. Other Research Support; Significant; Amgen.
- © 2015 by American Heart Association, Inc.