Is a new study of Medicaid a game changer?
On May 1, 2013, the New England Journal of Medicine published a landmark study with important implications for President Obama’s health care law -- and, in a much less far-reaching way, some of our recent fact-checks.
The topic: Medicaid, the government-run health insurance program for the poor. Under Obama’s law, states can get federal dollars to enroll their uninsured residents in Medicaid. Some politicians argue that taking the federal money now leaves the state on the hook for long-term fiscal commitments it can’t afford.
On April 8, we fact-checked a claim by Sen. Ted Cruz, R-Texas, that went beyond the usual fiscal criticisms. "Expanding Medicaid," he said at a press conference, "will worsen health care options for the most vulnerable among us in Texas."
When we looked at the studies Cruz’s office cited as support, we found a number of problems with his conclusion. On this evidence, we ruled it False.
The new study in the New England Journal of Medicine was published almost a month after our fact-check ran, and unlike the studies we evaluated initially, it was specifically designed to compare health outcomes for Medicaid patients against those who are uninsured.
Did the new study improve Cruz’s case?
We should note that we evaluate claims based only on evidence known at the time. We don’t go back and re-evaluate if new facts come to light later, because it doesn’t make sense to give people credit for evidence they couldn’t possibly have known about at the time.
In this case, though, we were curious about whether this study changed the debate on this important issue going forward. So we decided to take a fresh look.
The new paper represents the second phase of a study coauthored by Katherine Baicker of Harvard University and Amy N. Finkelstein of the Massachusetts Institute of Technology.
In 2008, Oregon expanded its Medicaid rolls, but only had enough funding to cover a portion of residents who wanted to sign up. To distribute spots in the program, officials set up a lottery.
The researchers compared 6,387 lottery winners to 5,842 who didn’t get a slot, tracking factors such as blood pressure, cholesterol, blood sugar, depression, and out-of-pocket medical spending.
The study produced mixed results. That has led both critics and advocates of expanding Medicaid to rush to portray the study’s results as evidence for their side.
Supporters of expanding Medicaid are pointing to the findings that beneficiaries had less depression and more financial security. Catastrophic out-of-pocket medical expenditures, for example, were "nearly eliminated."
Opponents of expanding Medicaid focused on the paper’s conclusion that the new coverage "generated no significant improvements in measured physical health outcomes," including blood pressure, cholesterol and blood sugar.
We’ve seen a lot of thoughtful analysis of the findings from across the ideological spectrum, including pieces by Avik Roy, Tyler Cowen, Megan McArdle, Ray Fisman, Ashish Jha, and Austin Frakt. However, rather than analyze every aspect of the study, we will take a narrower approach: Would Cruz’s original comment -- that "expanding Medicaid will worsen health care options for the most vulnerable" -- have been correct if this study had been published when he said it?
There’s little question that the study was a disappointment to those who support the expansion of Medicaid. The Oregon study gives critics of the expansion much better evidence than they had previously to argue that expanding Medicaid will not necessarily produce actual health benefits.
It’s true that depression decreased, a result that is good on its own, as well as a potentially positive factor on a patient’s future physical health.
But the lack of clear and unmistakable improvements in three metrics that are signals of better health (blood pressure, cholesterol and blood sugar) was an unpleasant surprise to those who support expanding Medicaid.
Despite all this, the paper’s findings don’t support the claim Cruz made, that people actually fared worse under Medicaid. The study seemed to show that people didn’t get much better -- not that they did any worse. (Cruz’s office did not respond to an inquiry for this story.)
Even if you look at the statistically insignificant results, most of the measures moved in the right direction, which hardly provides clear evidence that things got "worse." All three blood pressure measures went in the right direction. Total cholesterol went up slightly, but this was balanced by an upward spike in "good cholesterol" and a decrease in the percentage of subjects with high cholesterol, both of which are the desired results. And blood sugar levels ticked up by a very small amount, but that was balanced by a drop in the percentage of participants who had high levels of blood sugar.
So, even the most negative reading of these results does not suggest that Medicaid patients became worse off in any consistent way. The results were, at worst, mixed -- better results for depression, no statistically significant improvement in several measurements of physical health."
"I think it would be fair to say that our study provides no evidence that Medicaid beneficiaries are worse off," Baicker, the study co-author, told PolitiFact. "It provides evidence that they are better off in some dimensions, but it fails to provide evidence that they are better off in other dimensions. However, in no dimension does it provide evidence that they are worse off." She added that the study found that beneficiaries did consume more health care, meaning that the program costs more money. "It's up to policy makers to weigh those costs against the benefits," she said.
Had the Oregon study been published before Cruz made his statement -- and had he voiced a less sweeping indictment of Medicaid -- he would have had a better shot at scoring well on the Truth-O-Meter. But he didn’t. If he, or another politician, makes a different claim, we’d be open to checking it. But even if we were to rethink our ruling on Cruz’s original comment, the new study gives us no reason to go higher than False.