Does Enrolling in Medicare HMOs Affect Mortality?

In 2002, 5 million Medicare beneficiaries, or 12% of the Medicare population, were enrolled in the Medicare+Choice (M+C) program. Under M+C, beneficiaries forgo the traditional fee-for-service (FFS) Medicare insurance program and enroll in a qualified HMO, which often provides benefits not covered by FFS Medicare, such as prescription drugs, eye care, or dental care. Medicare HMOs may become even more popular in the future, as the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 will increase payments to these plans, particularly those offering prescription drug coverage.

A critical question raised by the growth of the M+C program is whether beneficiaries experience better health outcomes in Medicare HMOs or in traditional FFS Medicare. In theory, beneficiaries in M+C may experience better outcomes because Medicare HMOs often provide greater coverage for preventative care such as diabetes screening and for prescription drugs. On the other hand, beneficiaries in M+C may receive fewer health care services - providers are reimbursed a fixed amount per patient per year and thus have lower profits when they treat patients more intensively - and this could result in worse health outcomes.

In Managed Care, Drug Benefits, and Mortality: An Analysis of the Elderly (NBER Working Paper 10204), Gautam Gowrisankaran and Robert Town explore how enrollment in Medicare HMOs affects one particular health outcome measure, mortality rates. Typically, this is a difficult relationship to quantify - beneficiaries who choose to enroll in Medicare HMOs may be systematically healthier than beneficiaries who opt for traditional Medicare, so that any observed relationship between HMO enrollment and health outcomes may not represent a causal effect of HMOs on health.

The authors offer a novel solution to this problem. They begin by noting that M+C payment rates are based on the average cost of treating FFS Medicare patients in that county three to eight years earlier, and suggest that insurers will be more likely to offer M+C plans and to provide drug coverage as part of those plans if payment rates are higher. The authors then predict the fraction of the population in a given county and year that will be enrolled in M+C plans with and without drug coverage based on the payment rate. In their analysis, the authors estimate the relationship between the predicted M+C enrollment rates and the mortality rate at the county level.

The sample for the study is counties with population over 100,000 in the years 1993-2000. The data comes for a variety of sources, including the National Vitality Statistics, Center for Medicare and Medicaid Services, and the Bureau of the Census.

The authors' principal finding is that mortality rates for beneficiaries in M+C plans with drug coverage are similar to those for beneficiaries in traditional FFS Medicare, while mortality rates for beneficiaries in M+C plans without drug coverage are substantially higher. The authors' estimates imply that a ten-percentage point shift in coverage from FFS Medicare to M+C without drug coverage would result in 51,000 additional deaths per year among the elderly.

These findings have important policy implications, though one must be cautious in drawing inferences from the results, as the plan types differ in many ways, making it difficult to attribute mortality differences to a single factor.

One implication is that prescription drug coverage may reduce elderly mortality. Roughly two-thirds of FFS Medicare beneficiaries currently have prescription drug coverage, typically through Medicaid or a supplemental Medigap policy, as do all beneficiaries in M+C plans with drug coverage. Thus it is plausible that increased access to prescription drugs explains the lower mortality rate for beneficiaries in these two plan types, though it could also result from other differences in covered benefits.

A second implication is that Medicare HMOs with drug benefits provide care (as measured by mortality outcomes) that is as good as that received by the typical beneficiary enrolled in traditional FFS Medicare. This could imply that the financial incentives for providers to offer fewer services to patients in Medicare HMOs either do not affect treatment decisions or that the reduction in services has no effect on mortality. On the other hand, it is also possible that there are negative mortality impacts of managed care but that they are offset by the positive impact of greater drug coverage.

Finally, it is worth noting that the authors look only at the effect of Medicare HMO enrollment on contemporaneous mortality; thus, their analysis does not measure other health effects, such as changes in level of functioning, or any mortality effects that might occur in future years. Nonetheless, the study's findings should be of great interest to those involved in Medicare policymaking and analysis.

This research was summarized by Courtney Coile.
NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us