The Effect of Report Cards on Medicare HMO Enrollment

In recent years, policy makers and health care leaders have become increasingly focused on improving the quality of the U.S. health care system. One promising mechanism for improving quality is medical report cards. If consumers are told that their health care provider scores well below average on quality measures such as the fraction of patients receiving routine screening tests or overall patient satisfaction, they may switch to another provider. Such behavior will benefit not only the patients who switch but also those who do not if low-quality providers feel compelled to make quality improvements. In order for medical report cards to improve quality, however, it must be the case that at least some consumers read them and switch providers on that basis. If most learning about the quality of health providers occurs through other mechanisms such as own experience or word-of-mouth, then the value of report cards will be minimal. In "Do Report Cards Tell Consumers Anything They Don't Already Know? The Case of Medicare HMOs" (NBER Working Paper 11420), Leemore Dafny and David Dranove examine whether the provision of quality information to forty million Medicare beneficiaries affected their enrollment decisions.

Medicare beneficiaries may choose to enroll either in traditional fee-for-service Medicare or in one of the Medicare HMOs available in their area. The Center for Medicare and Medicaid Services (CMS) collects quality data directly from HMOs as well as through a survey of Medicare beneficiaries. Since 1999, the CMS has mailed a pamphlet, Medicare & You, to all beneficiaries on an annual basis. In 1999 and 2000, the pamphlet included quality measures for all HMOs operating in the beneficiary's market area. These measures include the percent of women who had a mammogram in the past two years, the percent reporting that their physicians "always communicate well," and the percent rating their care as the "best possible" (a 10 out of 10).

The key measure in the authors' analysis is the share of Medicare beneficiaries in each county and year that are enrolled in each HMO and in traditional Medicare. Their sample has data for nearly 8,000 such county-plan-year observations during 1994-2002. For each observation, they have the quality measures as reported in Medicare & You as well as many characteristics of the plan and market such as the monthly premium, whether the plan has prescription drug coverage, and the average age and education level of the county population. Importantly, they also have several quality measures that were not included in Medicare & You, which they use to estimate consumer responses to quality that are generated by market-based learning rather than the report cards.

In their analysis, the authors estimate a model in which each Medicare enrollee selects the insurance plan that offers her the highest utility, choosing among traditional Medicare and all the Medicare HMOs operating in her county. The most convenient way to interpret the results is to simulate how enrollment in a typical market changes over time and in response to the publication of report cards. The authors consider a hypothetic market with three HMOs, each with the average national market share of 2.7 percentage points in 1994. One HMO is of high quality (one standard deviation above the average), while the other two HMOs are of average quality.

The results of this exercise are shown in Figure 1. With market learning only (represented by the solid lines), the market share of the high-quality plan more than doubles over an eight year period, while the market share of the average-quality HMOs and of traditional Medicare decline. The dashed lines show the effect of report cards, which boost the market share of the high-quality plan by an additional 2.2 percentage points. Thus, the effect of the report card is nearly as large as the total effect of market learning over eight years. The authors find that the report card effect is entirely due to beneficiaries' responses to consumer satisfaction scores; other quality measures such as the mammography rate do not affect enrollment.

The authors conclude that Medicare enrollees were switching into higher-quality plans independently of government-issued report cards during this period, but that the distribution of report cards greatly accelerated this trend. They find that the total market share of traditional Medicare fell, but not dramatically, as many of the switchers came from lower-quality HMOs. Finally, the authors suggest that in order to understand the full implications of report cards, it will be necessary to also examine supply-side responses, such as whether providers focus disproportionately on improving measures reported in the report cards and skimp on other quality measures.

The authors acknowledge funding from The Searle Fund for Policy Research.