Estimating the Impacts of Medicaid Managed Care in Rural Minnesota Final rept

摘要

Perhaps the single largest Medicaid policy initiative during the 1990s was shifting program beneficiaries into managed care. By 2000, more than 19 million Medicaid beneficiaries (56 percent) were enrolled in some type of managed care, up from about 1.5 million (less than 10 percent) in 1990 (CMS 2002 and Congressional Research Service 1993). The share of the Medicaid population enrolled in Medicaid managed care (MMC) is likely to continue to increase, especially in rural areas, as the Balanced Budget Act (BBA) of 1997 substantially expanded the authority of states to provide Medicaid services through managed care. States have turned to MMC to achieve a variety of objectives, including improving beneficiaries' access to care while controlling Medicaid spending (Rowland et al. 1995; Holahan et al. 1998; Davidson and Somers 1998). Given the magnitude of the shift to managed care across the Medicaid program, determining whether MMC achieved those goals is an important policy question. Answering this question, however, has proved to be difficult. Though several studies have examined the effects of MMC on beneficiaries, the results are mixed (Hurley and Zuckerman 2002). Owing to data constraints, most studies are limited to a single framework, which limits the ability to examine the sensitivity of the findings to alternative methods. Comparison groups are typically constructed on the basis of two factors: time and population characteristics. Because neither time- nor population-based comparison groups are selected through random assignment, both are susceptible to confounding factors that could affect the outcome of interest. Thus, the estimates of program impacts obtained using such methods may be incorrect. These limitations can be overcome to some extent by combining time- and population-based comparison groups in a difference-in-differences framework (e.g., Tai-Seale, Freund, and LaSasso 2001) and comparing changes over time for the treatment group to changes over the same time period for a matched comparison group. In this paper we assess the impact of Minnesota's 1115 managed care demonstration project- the Prepaid Medical Assistance Program (PMAP)- on access to and use of health care services using quasi-experimental alternative evaluation designs. Taking advantage of the gradual introduction of Medicaid managed care (MMC) across counties in rural Minnesota, we use pre-post, matched comparison group, and difference-in-differences methods to compare beneficiaries in counties that implemented PMAP to beneficiaries in counties that continued to operate traditional fee-for-service (FFS) Medicaid. The study uses two rounds of a survey of MMC and Medicaid FFS beneficiaries in rural Minnesota, conducted in 1998 and 2000.