In Oregon, the Data that Can Guide the Medicare Expansion Debate

Long before the Affordable Care Act expanded Medicaid to provide health insurance to more low-income households, debate raged over the value of such an expansion. Would it save states millions of dollars, as some claimed? Or will free health care result in expensive overuse? Worse still, would low Medicare reimbursement rates make the program worse for patients than having no health insurance at all?

Evidence has been hard to come by, but a groundbreaking randomized study launched in 2008 suggests that, in the short run, providing Medicaid coverage increases health care usage and public spending. However, it also has significant health benefits, Visiting Scholar Amy Finkelstein told undergraduates at a Friedman Forum held May 13, 2016. 

Finkelstein, the Ford Professor of Economics at Harvard University, is co-principal investigator of the Oregon Health Insurance Experiment, the first randomized controlled trial (RCT) to study the impact of covering the uninsured. The project took advantage of a fortuitous natural experiment: Oregon announced an expansion of its Medicaid program, but there wasn’t enough funding to cover everyone eligible, so it was decided that a lottery was the only fair way to allocate coverage. 

This gave researchers a rare (and ethically sound) opportunity to compare those who received Medicaid health insurance and those who did not, in a statistically identical population. It was a way to empirically test opposing arguments about the true costs of providing insurance to the poor—and also assess the benefits.

One common conjecture was that a Medicaid expansion would drive up usage and costs. When insurance coverage lowers out-of-pocket costs for health care, “basic economics says if you lower the price, demand will increase, and people will use more, because they are not facing the true price on the margin,” Finkelstein said. 

This turned out to be true: they found a 2 percentage point increase in probability that a person in the Medicaid group went to the hospital. For covered individuals, the probability of a hospital admission was 30 percent higher; outpatient visits went up 35 percent; and prescriptions were 15 percent higher. Compliance with prescribed care also increased. 

The cost counterargument is that providing health insurance could actually decrease overall spending by reducing uncompensated care and expensive and inappropriate reliance on emergency rooms. But the study showed that overall, health costs for the covered group went up 25 percent—with a surprising 40 percent increase in ER visits in the covered group.

It’s not clear why people did not shift away from emergency care. “Medicaid makes ER free but also makes the doctor’s office free. It’s not just that you can’t get in to see a doctor, because [ordered] tests and drug prescriptions also go up,” Finkelstein commented. “The offset from the ER to the doctor does not dominate.” 

The study looked at administrative data that tallied use of the health care system, but also used clinical tests to capture health status outcomes and conducted surveys to assess well-being. In interviews, Medicaid enrollees reported higher rates of diagnosis and treatment for high blood pressure, high cholesterol, diabetes, and depression. Clinically, however, the study found no detectable impact on blood pressure, cholesterol, or diabetes. 

Notably, however, the Medicaid group showed a significant 30 percent decline in depression. And self-reported health improved, with nearly 70 percent rating their health as good, very good, or excellent and more than 80 percent saying their health was the same or improving. 

Insurance also helped stabilize participants’ financial health. “The probability of facing out-of- pocket expenses goes down, along with the chance of facing collections for medical debt. The chance of catastrophic expenditures goes essentially to zero,” she said. Fewer people reported borrowing money or skipped bills. Researchers found no effect on earnings and employment.

In summary, “The idea that Medicaid is worthless or worse than no insurance is categorically not true, and we reject that,” Finkelstein said. It increases utilization, improves perceived access and quality, reduces financial strain, virtually eliminated catastrophic medical costs, and improved self-reported health and reduced depression. It also increased health spending by 25 percent. 

She said the findings were relevant for the ACA expansion of Medicaid, but cautioned against extrapolating too far. The study followed enrollees for two years so far, and long-run effects may be different. The Oregon population is likely not representative of the entire nation, and there may be differences arising from voluntary rather than mandatory enrollment. 

Randomized control trials are only one method Finkelstein and colleagues use, and it’s not perfect, but it can be a valuable tool for answering questions about the cost and benefits of such policies. “The beautiful thing about a randomization is that you don’t have to assume or cross your fingers and hope that the groups are the same; they actually are, on both the observable things you could have controlled for, and on things you can’t.”

“The results really have the ability to surprise you,” she added. 

The study generated extensive (and often opinionated) media coverage. “That’s great for me and it made my mom proud,” she joked, “but it’s also rather sad. Oregon should not get that much media attention; it got it because it (the study) was way too rare.” 

While RCTs are the gold standard used in 80 percent of studies of medical treatments and drugs, only 18 percent of health care delivery interventions are studied this way. To encourage more work in this area, Finkelstein and Harvard colleague Lawrence Katz have founded J-Pal North America, a branch of the Jameel Poverty Action Lab, which has used RCTs extensively to study development and economic issues around the world. J-PAL North America focuses using randomization to study issues in the US health care delivery system, such as the effectiveness of interventions with “super-utilizers” caught in a cycle of frequent hospital admissions. 

“Post-graduation, there are opportunities like this o apply your rigorous economics training to exciting research and real-world policy problems,” she told the audience. She encouraged students to start with her own project.  “All our data and results are available and posted online. You can use it for your undergrad theses. You can find mistakes we’ve made and write to us about that.” 

“You guys just get fabulous training; I’m not sure you appreciate it. The ability to think rigorously and carefully as an economist is a wonderful gift, no matter what you do in life.”

—Toni Shears