Hitting the lottery

[ Winter 2012 ]

Oregon’s experiment with Medicaid gives an HSPH economist a rare chance to analyze effects of extended coverage.

Katherine Baicker, professor of health economics

In March 2008, a colleague burst into the office of HSPH health economics professor Katherine Baicker at the National Bureau of Economic Research (NBER) in Cambridge, Massachusetts. She breathlessly told Baicker about a story she’d just heard on the radio: The state of Oregon was expanding its Medicaid coverage, and it was choosing recipients by lottery.

Baicker’s brain lit up. Researchers have been studying the effects of Medicaid on recipients since 1965, when the subsidized health insurance program was created. But because so many differences between insured and uninsured groups—such as income and education—underlie health issues, it has been nearly impossible to untangle the effects of the insurance from those other factors.

Baicker immediately saw a golden research opportunity: to observe the effects of Medicaid in a real-world, population-based experiment, in which the only difference between insured and uninsured groups would be the luck of the draw. “This was an unprecedented chance to gauge the effect of insurance expansions in the context of a randomized lottery, which had never been done before,” she says. For a health policy researcher, it was like hitting the lottery.

Ordinarily, public health studies take months of planning to rigorously consider every aspect of experimental design before they kick into gear. Baicker and her colleague, MIT economics professor Amy Finkelstein, knew they didn’t have that luxury. “I thought, ‘Oh my God. This was a drop-everything-and-must-look-into-it moment,’” says Finkelstein.

A quick study

Already, the state was drawing names for participants. “We frantically called anyone we knew who had ever worked in Oregon and got in touch with the right people,” says Baicker. Partnering with local researchers in Oregon, the team began distributing surveys to applicants almost immediately, even as they were still developing the experimental design for the study. “We hit the ground running a week after the lottery started.”

Their urgency paid off. The study they eventually produced, the first part of which was released in July 2011 as an NBER working paper, is already being called a classic in the field of health economics. It has also fundamentally changed the way Medicaid is viewed.

Challenging conventional wisdom

While the average person might consider it obvious that people with health insurance would fare better than those without, the benefits of Medicaid have actually been a hotly debated policy question. Some say that despite flaws in the system, having Medicaid gives people access to more care, and saves money by reducing inefficient care. Others counter that with emergency room visits and free care, low-income people without insurance are able to fare as well as those with Medicaid. (Some have even claimed that because of the poor quality of care offered through Medicaid, having coverage may actually be worse than not having it.)

Baicker, Finkelstein, and colleagues have shown definitively that both positions have flaws. For their study, they mailed questionnaires to more than 70,000 applicants to the Oregon program, targeting 35,000 who were selected in the lottery for access to Medicaid and 35,000 who were not. Using these surveys, along with hospital admissions data and credit reports, they were able to demonstrate that those who gained access to Medicaid utilized more health care, reported better physical and mental well-being, and were in better financial circumstances than those who didn’t—with fewer bills sent to collection, and a decline in the average amount of medical collections of about $400.

Clear benefits, quickly

The benefits that Medicaid recipients reported are impressive. In addition to having fewer unpaid bills and bills sent to collection, recipients were 10 percent more likely to screen negative for depression and 25 percent more likely to report their health as “good, very good, or excellent.” Even more dramatically, recipients reported themselves happier overall after receiving coverage, with a 32 percent increase in the number of respondents calling themselves “very happy” or “pretty happy” (as opposed to “not too happy”) after just one year.

Subjects’ health decisions also shifted dramatically, with many reporting substantial changes in their treatments as soon as they acquired insurance—for example, by taking their full dose of medication, now that they could afford it. “They were trying to make a prescription stretch by taking a smaller dose, or taking a dose every other day. Friends and family would try and help by sharing leftover prescriptions. It was only after insurance afforded a prescribed dose that they reported experiencing the health benefits of treatment,” says Heidi Allen, a research scientist with Providence Health & Services in Oregon, one of the lead investigators who coordinated qualitative in-person interviews as part of a follow-up study that has bolstered the original results and provided surprising new anecdotal evidence of Medicaid’s benefits.

In other cases, recipients went through with long-delayed surgeries they couldn’t afford without coverage. “One woman had been sent out of the emergency room with a broken ankle in an ACE bandage years earlier,” says Allen. “The first thing she did when she got Medicaid was get orthopedic surgery to repair the damage. For her, that was life changing. Now she can exercise and she is losing weight.” For those without insurance, meanwhile, the stress of not being able to afford care influenced even simple decisions in daily life. “One man said, ‘I would go for that bike ride—but, man, if I crashed that bike, I wouldn’t be able to afford to go to the doctor,’” says Allen. “People are forced into making difficult choices—do I get high-quality child care, or get my neighbor to watch my kids? Do I pay half my rent so I can afford my medication? That kind of living is really stressful and difficult.”

No free lunch

These benefits do not come without a price tag. The study shows that expanding Medicaid increases overall health costs, since recipients in Oregon with insurance utilized more health care than those without insurance. Over the course of a year, for example, recipients were 30 percent more likely to undergo some hospitalization, with a 45 percent increase in the number of procedures they underwent. According to the researchers’ back-of-the-envelope calculation, those with Medicaid spent about $778 more per year than those without, a 25 percent increase.

Preliminary results from the study do not support the argument, often made by insurance advocates, that those costs are offset by insured people making fewer emergency room visits. “People have argued we should expand Medicaid and it will pay for itself,” says Finkelstein. “Wouldn’t it be nice if it were a free lunch, but our initial results suggest it is probably not the case.”

The researchers did find that compliance with preventive care—such as mammograms, cholesterol testing, and Pap smears—went up for insurance recipients, which could lead to a decrease in health care costs down the line, but there was no decline in spending in the first year. Furthermore, they don’t believe their results necessarily reflect a “pent-up demand” for care that would wane over time, since the spending remained constant over the yearlong study.

Profound policy implications

The findings could reverberate for a long time. The new national health care law mandates an expansion of Medicaid by 2014, but that law has been challenged both in the courts and in Congress, where Medicaid has become a political football in the debate over cutting the national budget.

With the new study, while policymakers can still argue that Medicaid is not worth the cost and debate whether expanding Medicaid is good policy, says Baicker, the study “makes it much harder to argue that Medicaid shouldn’t be funded because it doesn’t help beneficiaries at all.”

As additional results analyzing new data are released, the study will continue to provide invaluable insights, not only in terms of numbers but also by putting a human face on a debate that has so far played out in Washington in abstract terms. “The country is preparing to insure 30 million more people, and we haven’t got a clear sense of what that means,” says David Cutler, Otto Eckstein Professor of Applied Economics at Harvard’s Kennedy School of Government. “Their research shows that while Medicaid isn’t a perfect program, it is better than not having coverage. Now we know that these programs are essential in two ways: health impact and financial impact. That should give us pause in cutting them.”

With the debate over health care in Washington still simmering, Baicker is careful to point out the limitations of her study. For example, it doesn’t weigh the benefits of Medicaid against those of subsidizing private insurance, a comparison Baicker says is not possible with current data. “It’s hard to imagine a circumstance in which people were randomized between Medicaid and private insurance,” she says. Moreover, it may not be valid to extrapolate the results for the relatively small population to which insurance was expanded in Oregon to the nation as a whole.

What the study does do is give lawmakers wrestling with the health care debate a more complete and complex picture of what they are getting for their money. Cost-benefit analyses are economists’ bread and butter, but some of the benefits found in the study may be harder to quantify than others, including recipients’ increased financial and emotional well-being. After all, says Baicker, “you don’t get fire insurance to prevent your house from burning down, but to ensure you aren’t thrown into financial ruin if it does.”

Ensuring that Medicaid recipients are protected from runaway medical bills may spur other benefits, from increased productivity to lower crime to family stability —as well as reduced stress for beneficiaries. “Those benefits are harder to measure and policymakers might think about them differently,” says Baicker.

The multitrillion-dollar question is whether all those benefits are worth the price. That’s a question no study, no matter how groundbreaking, can answer. “Economics can’t tell lawmakers what their policy priorities should be, but it can help inform decision making,” Baicker says. “Our study shows that Medicaid does, in fact, use resources, but that it also significantly helps people. As a society, we need to weigh the program’s financial costs against the tangible and intangible improvements in beneficiaries’ lives.”

Michael Blanding is a Boston-based magazine writer and a fellow with the Safra Center on Ethics at Harvard University.