Being enrolled in Medicaid really does help both patients and health care providers in important ways.
That’s according to results from a groundbreaking study out of Oregon released July 7 by the National Bureau of Economic Research. While the findings may sound obvious, the study is significant and unique in the quality of its evidence on the effects that being insured have on poor people.
The study has been generating a lot of buzz in Medicaid and health policy circles, and was even featured in a recent New York Times article. The only trouble is, if you’ve seen one Medicaid program, then you’ve seen one Medicaid program. Because of the wide variation in Medicaid programs among states, the study may be less helpful in gauging the effects of Medicaid enrollment on the health care of recipients nationally.
The results of the Oregon study, which are qualitatively about as good as we’re going to get, are extremely positive. Those on Medicaid were far more likely to receive health care than those who remained uninsured and were significantly less likely to have financial problems related to unpaid medical bills. Out-of-pocket health bills were lower, and Medicaid enrollees reported a sizable improvement in their overall happiness levels. While the authors are careful to note that these results reflect differences in a relatively small subset of the Oregon low-income population and probably shouldn’t be generalized to low-income populations elsewhere, this is still the best evidence around on the consequences of having health insurance.
The study is unique because it relied on an opportunistic experiment that produced a randomized control research design, which is virtually unprecedented in health policy research. Here’s how it came about: Starting in 2008, Oregon officials wanted to expand the state’s Medicaid program, but could afford to add only 10,000 new people to the program. So they used a lottery system to select new Medicaid recipients from about 90,000 eligible applicants. This provided two similar groups to compare. The fact that the lottery was random meant that the new Medicaid recipients and those who remained uninsured were more or less identical except for their insurance status. This meant that any differences between the two groups in health care usage, out-of-pocket spending on health care, and health status could be assumed to be the result of differences in insurance status rather than other factors. There have been many studies comparing more conventionally enrolled Medicaid recipients and the uninsured, but these studies have been unable to completely control for differences between the two groups — in health, income and other factors besides Medicaid enrollment, which might affect health care use and spending.
The group which produced the study, headed by economists from MIT and the Harvard School of Public Health, took full advantage of this design to investigate a wide range of questions. They collected administrative data on hospitalizations of the two groups and surveyed both groups to measure their use of different kinds of health care, their out-of-pocket spending on health, and their overall well-being. They also obtained data from credit agencies on unpaid medical bills sent to collection agencies to measure the effects of Medicaid enrollment on family finances.
The results, as noted, were extremely positive — as far as Oregon is concerned. We should be careful about generalizing from Oregon Medicaid to Medicaid in the rest of the country. States have a lot of discretion in how they design and manage their Medicaid programs, and there are enormous differences between states in overall Medicaid spending, coverage of and payment rates for different classes of services, eligibility and enrollment practices, and a host of other factors. These policies also affect the level and mix of services to which Medicaid clients have access, and it shouldn’t be assumed that the favorable results in Oregon would appear in every state if this study were to be replicated elsewhere. There are really 50 different Medicaid programs rather than just one, and what happens to Medicaid recipients in Oregon can’t be assumed to hold true in New York, Alabama or Wisconsin.
While documentation of the differences in state Medicaid programs is abundant, it is particularly instructive to consider the Oregon results in light of another study published last week, in the widely read journal Health Affairs. This study, by economists Todd Gilmer and Richard Kronick, documents and explains variations in Medicaid acute care spending for disabled patients across states. Gilmer and Kronick document significant disparities in spending levels across states — the most expensive state spent more than twice as much per recipient as the least expensive, and there were comparable disparities in hospitalization rates, use of outpatient services and prescription drugs. These differences persist even after controlling for case-mix, meaning that spending is driven more by differences in service volume and price than by differences in the patients being treated.
Gilmer and Kronick also present some results that suggest that the services to which this group has access differ considerably across states. While they are careful to note that the spending data they analyze are of limited use in measuring health care quality and patient outcomes, they also point out some suggestive relationships between relative spending on different services. States with more primary care physicians had lower rates of hospital admissions for so-called “ambulatory care sensitive” conditions such as diabetes and asthma, and states with high levels of spending on outpatient care and prescription drugs frequently had lower levels of hospital admissions and spending. These patterns suggest, though they do not prove, that Medicaid patients in some states had better access to primary care and other outpatient care, including prescription drugs, than Medicaid patients in other states, and spent less time in the hospital as a result.
Gilmer and Kronick’s findings, among others, strongly reinforce the notion that Medicaid is not the same everywhere, and that replication of the Oregon experiment in other states might produce different findings. State policies and administrative practices may have as much to do, if not more, with Medicaid patients’ outcomes than just the simple fact of being covered or not. What happens after people are covered can’t be assumed to be the same everywhere.
States also vary in the accessibility of care for uninsured patients, so that the difference between Medicaid patients and the uninsured may not be as large in some states as in Oregon. The New York City-based Health and Hospitals Corporation, for example, the largest public health care system in the country, provides care to some 450,000 uninsured patients every year. This level of service almost certainly does not exist elsewhere, although other public hospitals may provide considerable amounts of care to the uninsured.
How do we continue to sort out the potential costs and benefits of expanding health care coverage along the lines proposed by national health reform legislation? While the results from the Oregon study deserve wide attention because of the high quality of the research design and the sophistication of the analysis, the Oregon study is unlikely to be replicated elsewhere. It is broadly, and appropriately, considered unethical to design research studies that explicitly deny care to some people. Rather, the Oregon lottery was undertaken to solve a particular problem of allocating coverage among a large number of applicants, and the researchers were not involved in designing the lottery system used.
Given the ethical problems involved in using lotteries for research purposes, studies such as Gilmer and Kronick’s offer a more feasible way of describing and analyzing differences between states in determining the quality of Medicaid coverage across the fifty states. Medicaid has traditionally not been well researched, and serious studies which try to sort out the performance of state Medicaid programs in providing care to their clients are still few and far between. The Commonwealth Fund, for example, publishes a series of well-regarded “scorecards” of the performance of state health care systems, but has not been able to separate Medicaid patients from others.