Do Medicare Advantage Enrollees Visit High-Cost Hospitals?

 Purpose and Background:The relationship between Medicare Advantage (MA) enrollment and the likelihood of visiting a high-cost hospital is not well understood. The lack of detailed hospital discharge summaries and unrepresentative of the traditional Medicare fee-for-service (FFS) beneficiaries makes it difficult to test on a large scale whether MA enrollees tend to visits high-cost hospitals and whether they receive better outcomes of hospital care than the mainstream FFS enrollees. The primary objective of this study is to examine the hospitals’ risk-adjusted costs and compare MA enrollees' and FFS beneficiaries' use of high-cost hospitals. The second objective of this study is to document the variation in racial and ethnic disparity in visiting high-cost hospitals within and between MA enrollees and FFS beneficiaries as policymakers mostly have focused on the location of care as an explanation for important disparities in many health outcomes.

Data and Methods: The Healthcare Cost and Utilization Project (HCUP) State Inpatient Databases (SID) were used in this analysis. We use hospital inpatient data for 2006-2010 from six states: California, Florida, Massachusetts, New York, Tennessee and Wisconsin.  The SID provide detailed diagnoses and procedures, total charges and patient demographics including gender, age, race and expected payment source. We obtained information about hospital characteristics using the American Hospital Association Annual Survey Database; and county level information from the Area Resource File. We initially calculated the total costs of all visits across all payor groups encountered during each year for each hospital. Next, we calculated the average costs for each primary diagnosis codes using all visits across all payor groups encountered during each year on a Core Based Statistical Area (CBSA) level. Then, we calculated the casemix adjusted total costs for each hospital using these CBSA based average cost values of primary diagnosis codes for each hospital and year separately. We then calculated the hospital cost index by dividing the actual total costs by casemix adjusted total costs. We finally calculated three categorical values to define the dependant variable for each state, which takes value 1 if the hospital cost-index is less than 0.95; value 2 if it is within 0.95 -1.05; and, value 3 if it is greater than 1.05. We used ordered logistic regression to examine an increasing preponderance of MA enrollees visits to high-cost hospitals relative to Medicare FFS beneficiaries. We also estimated the same model using a different specification of high-cost hospital definitions to ascertain any effects resulting from sample sizes.

Results: In our sample, we found lower prevalence of high-cost hospitals among MA enrollees than among FFS beneficiaries. Our risk adjusted results show that the odds ratios of visiting a high-cost hospital for MA enrollees range from 0.641 to 0.958 for all states. We also found sizable geographic variation in visiting high-cost hospitals among minority elderly population. Our results show that non-white elderly patients associated with lower likelihood of visiting high-cost hospitals in in California and New York, and higher likelihood of visiting high-cost hospitals in Florida, Massachusetts, Tennessee and Wisconsin.