WITHDRAWN - Geographic Variation in Healthcare Spending for those with Functional Limitations & Chronic Conditions
Keywords: geographic,longitudinal,causal inference,regression,settings research,healthcare reform,chronic condition,functional limitation,medicare
Scope of the Problem: Research has found dramatic geographic variation in healthcare spending. Studies of this topic revealed that regions with higher spending sometimes have worse health outcomes, suggesting that policy opportunities exist to reduce national healthcare costs by targeting specific regions and populations for high-value care. While some of the regional variation in spending can be linked to differences in healthcare practices and economic status, there remains significant variation that has often been attributed to the population’s underlying health risks. Our recent analysis of Medical Expenditure Panel Survey (MEPS) 2006 for Assistant Secretary for Planning and Evaluation found that community residents aged over 65 with functional limitations and chronic conditions spent an average of $8,300 in Medicare, compared to $2,960 among those with chronic conditions only and $3,400 for those with functional limitations only. Comparative geographic analysis of this cost-driving population will examine the degree to which their functional limitation and chronic condition status is responsible for the regional differences in Medicare spending.
Objective: We investigate the co-occurrence of functional limitations and chronic conditions as a predicting factor for regional differences in spending among Medicare beneficiaries and geographic/environmental factors attributed to the differences in prevalence rates.
Method/Results: Five years of the Access to Care (ATC) module of the Medicare Current Beneficiary Survey (MCBS) data are pooled together into a single analytic file in which seven distinct geographic regions are defined. In addition, we use the Area Resource File (ARF) for specific environmental data such as physician supply by specialty, hospital beds, and managed care penetration. The first step involves tests of spending and prevalence rates of functional limitations and chronic conditions by region, and multinomial logit models to detect evidence of substantial regional differences in prevalence rates. The second step uses a generalized estimating equation method to measure the degree to which co-occurrence of functional limitation and chronic condition status contributes to high spending in different regions. Some preliminary geographic factors for prevalence rate variations are population density, poverty, access to care and insurance penetration. We also found great explanation power of prevalence rates to spending variation.