The Washington PostDemocracy Dies in Darkness

Here’s what the Oregon Medicaid study really said

The Oregon Medicaid experiment is an academic miracle born out of a human tragedy.

A few years back, Oregon found the money to add 10,000 residents to the state’s Medicaid program. The only problem was that there were 90,000 residents who qualified for the program and desperately wanted in. So the state held a lottery. Welcome to the American health-care system. Greatest in the world, folks.

But 80,000 Oregonians’ loss was science’s gain. The lottery gave researchers an opportunity that’s almost never available in policymaking: They could create a randomized controlled study -- the absolute gold-standard of experimental design -- comparing the health outcomes of the lucky Oregonians who received Medicaid to those who didn’t. It would be the first time that kind of study had even been used to compare the insured and the uninsured.

The initial batch of results was released in August 2012. The data covered the first year of the Medicaid expansion and found that the folks on Medicaid were getting more care, reporting better health (both physical and mental), and seeing fewer financial problems than the people who weren’t on Medicaid.

The second set of results was released Wednesday. The data now covers two years and, importantly, includes clinical measures of health rather than relying on the reports of the study participants. These results are more mixed, but also more telling.

Here’s what we can say with certainty: Medicaid works as health insurance.

That might seem obvious. It's actually not. A big criticism of Medicaid is that it pays doctors so little that it's essentially worthless because no doctor will see you. But the Oregon residents who won the Medicaid lottery got much more health care -- including preventive health care -- than the residents who lost it. They also saw catastrophic health costs basically vanish.

"People who gained access to Medicaid did use more health care," says Harvard's Katherine Baicker, one of the study's authors. "We can eliminate the story that Medicaid is so lousy you can’t get in to see a doctor.”

But the study didn’t see much improvement in the health indicators it was tracking. Blood pressure and cholesterol readings were mostly unchanged. Diagnosis of diabetes went way up, and the use of medicine to control diabetes also went up, but, again, there wasn’t much difference on the relevant blood tests. The big exception, surprisingly, was mental health: depression rates fell by 30 percent.

So here’s what happened in the first two years of the Oregon Medicaid experiment: Medicaid proved itself good health insurance. The people who got Medicaid used more health care, and seem to have done so smartly -- they got preventive care, they got their diabetes diagnosed and began managing it, they treated their depression, and so on. But the health care itself didn’t work as well as we hoped -- at least not in terms of cutting rates of hypertension and cholesterol.

There are a number of possible spins you can put on that finding. One is that the study was simply too small, with too few sick people, to show the kind of quick health changes the researchers were looking for. Sharply increasing the number of people who are managing their diabetes and mental health, getting colonoscopies and mammograms, and making regular trips to the doctor sure seems like the kind of thing that will improve long-term health outcomes. Other studies with a less rigorous -- but still credible -- design and a longer timeframe have shown that states that expanded Medicaid saw a six percent drop in death rates among the newly insured group.

Another is that the Medicaid enrollees are getting bad health care. But the study mostly seems to blow up that argument. Medicaid’s critics have long worried that the program pays so little that the people on it don’t actually get care. But Medicaid recipients were getting care, and the care they were getting appears to have made sense.

A third and more radical interpretation is that health care -- or at least the kind of routine health care insurance buys you -- simply doesn’t work that well. After all, it’s not as if the Oregon study contradicts past randomized studies pitting the insured against the uninsured. It’s the only study of its kind. In that way, it’s a unicorn.

My view, after speaking with a number of the study’s authors and outside experts, is a mixture of one and three. There’s voluminous evidence that managing diabetes and treating depression and being able to go to the doctor improves health. You have to be willing to throw quite a lot of existing theory and evidence out the window to believe that stuff won’t pay off down the road.

A bit of introspection can be helpful here. I’m lucky enough to be insured. I’ve seen my doctor a few times over the last two years. None of those visits had any measurable impact on my health. But if something had been wrong on one of those visits, the story would be very different. I have health insurance not because it improves my health on any given day, but because regular access to the medical system will, presumably, improve my health over time.

But there really is mounting evidence that the average unit of health care we receive isn’t worth much. That doesn’t mean it’s all waste, at least as we traditionally think of it. A lot of doctor’s visits end with the physician saying “let’s just wait and see,” and then the problem clears up on its own. That visit didn't improve anyone's health. But those visits are necessary, because sometimes they end with the doctor saying, “we need to get you to the hospital right now.”

That said, a lot of what the health system does actually is waste and we need to get much better and faster at telling which procedures help and which just pad profits. That's something we're already doing with comparative-effectiveness reviews for drugs and devices and surgeries. It's something we need to start doing for health insurance, too.

The problem with the Oregon study is that it doesn’t help us figure out how to make health care or health insurance better. We don’t know if the results speak to the health care you get through all health insurance or just Medicaid or if they're just an artifact of the study's timeframe and sample size. We don’t know if different ways of designing insurance programs would lead to radically different care outcomes (we actually tried a randomized study on that question in the 70s, and the answer, at least then, was “not really”). And so we don’t know whether we’re seeing a problem in Medicaid, an inconvenient truth about medical care, or something else. Worse, we don't know what to do next.

But we could find out. Studies like this one don’t need to be so rare. The government could put a very small amount of money -- say, one percent of the federal budget -- toward designing and conducting them across all areas of public policy and the results would help us spend the rest of our dollars much more wisely. There's no reason the Oregon experiment has to be a unicorn.