PMH62 - Comparative analysis of drug use in the ambulatory setting among the elderly medicare beneficiaries with alzheimer's disease using different public use files

Abstract

The Centers for Medicare and Medicaid Services recently increased public access to its Medicare data with the release of Basic Stand Alone (BSA) Medicare Claims Public Use Files. This study sought to compare the estimates obtained from new BSA claims and public survey data on the number of prescriptions per visit associated with Alzheimer's disease (AD) among elderly patients in the ambulatory setting. Visits of Medicare beneficiaries aged 65+ with AD (ICD-9-CM 290, 294, 331, or disease indicator variable) were identified in two available data years (2008 and 2010) of BSA claims and two national surveys (National Ambulatory Medical Care Survey [NAMCS] and outpatient component of National Hospital Ambulatory Medical Care Survey [NHAMCS]) of the same years. Descriptive statistics were compared between the claims and survey sources on patients’ demographics, co-morbidities, and drug use. Numbers of prescriptions per visit were compared between patients with and without AD through multivariate regressions (Generalized Linear Model with Poisson distribution), controlling for age, sex, and a series of co-morbidities. Ambulatory visits by Medicare-covered elderly AD patients were identified. As compared to the two surveys, the claims source involves more female beneficiaries (64.8% vs. 52.9%) and has higher proportion of patients with other chronic conditions in addition to AD (88.2% vs. 67.8%). The top 3 co-morbidities in the claims source are myocardial ischaemia, heart failure and diabetes, while the list is different in surveys as arthritis, diabetes and cancer. AD contributes an increased drug use of 2.3 prescriptions per visit in claims and 1.1 prescriptions in surveys, as predicted by multivariate regressions. This study is an initial exploration comparing the new Medicare claims source with national surveys on AD patients’ drug use. Further evaluations are needed to inform the use of this new public data source in health services research.