Share

Sign up for the Faster Lane
email newsletter
May 6, 2013
Print Friendly

Op-Ed: What does the Oregon Medicaid Experiment mean for Kentucky?

Results do not make governor’s decision any easier

Results do not make governor’s decision any easier

By Jose M. Fernandez
Assistant Professor of Economics
University of Louisville

LOUISVILLE, Ky. (May 6, 2013) — The New England Journal of Medicine recently published the finding of the Oregon Medicaid Experiment. Many experts believed the results of this study would provide valuable insights to states still considering Medicaid expansion as part of the Affordable Care Act (a.k.a. Obamacare). However, the results created more questions than answers.

Jose Fernandez is assistant professor of economics at the University of Louisville.

Jose Fernandez is assistant professor of economics at the University of Louisville.

Gov. Steve Beshear has yet to decide if Kentucky will expand Medicaid as part of the Affordable Care Act. The results from the Oregon Medicaid Experiment, published May 1 in the New England Journal of Medicine did not make his decision any easier.

The Oregon Medicaid Experiment was created not out of academic interest, but unfortunately because of state government budget shortfalls. A few years ago, Oregon found funds to expand Medicaid by roughly 10,000 individuals, but the state had more than 90,000 people who qualified for the program. The equitable solution was to hold a lottery.

What resulted from the lottery was a massive randomized controlled trial evaluating the effects of health insurance on health. The study follows 6,387 adults who were randomly selected to be able to apply for Medicaid coverage and 5,842 adults who were not selected for two years. The results found that Medicaid coverage increase health care spending (primary care, specialist and ER visits) and dramatically reduced the financial hardship for these individuals. Medicaid patients spent an average of $1,172 more than uninsured patients.

Further, there is a significant improvement in mental health status as the recently insured report 9 percent fewer cases of depression. However, there are no statistically discernible effects on physical health between the Medicaid group and the uninsured group (hypertension, cholesterol, hypercholesterolemia and diabetes).

At first these results seem to present an indictment against Medicaid expansion (with the exception of the positive results to mental health and financial stability). However, let’s delve a bit further into the results.

♦ These results are consistent with the result found from the famous Rand Health Insurance study carried out from 1974 -1982, which also found an increase in health expenditures as health insurance coverage increased, but no discernible difference in physical health outcomes.

♦ The randomization process only supplied access to health insurance. Only 60 percent of those individuals who won the lottery actually enrolled into Medicaid. This result looms quite large for several reasons:

(1) The lack of insurance uptake gives some credence to the behavior obstacles associated with our health. We only seem to want health insurance when we are sick.

This short-run thinking can lead to long-run problems.

(2) The selection effect caused by this behavior can affect the statistical results. Only individuals who are really sick will take part in the experiment causing the treatment group’s outcomes to be worse off.

(3) What does this lack of insurance uptake say about the health insurance mandate?  On the one hand, people enjoy their freedom to choose, but on the other hand, why would you not choose an option that leaves you unambiguously better off? You don’t have to use the insurance if you have not need, but it is better to have it than to need it.

♦ We need to be careful when interpreting null results. That is, to say something is statistically insignificant does not immediately imply it does not work. Instead, when something is statistically insignificant we say we don’t have enough evidence to distinguish the effect. Statistical insignificance can result for two reasons.

First, the power associated with the test is too small. The power of a test refers to cases when the point estimate is clinically significant, but the sample size is too small to find a statistically significant effect. An alternative reason is that the level of imprecision is too high.

For example, a wonder drug is found to have very large positive effects, but only for a small number of people. The mean will be positive, but the variance of the treatment effect will be large. Given my previous statement about the lack of health insurance uptake calls into question if these results had enough statistical power to claim that health insurance does not improve physical health.

♦ Oregon and Kentucky Medicaid reimburse primary care physicians at nearly identical rates versus private insurance: Oregon (64 percent) and Kentucky (62 percent). Therefore, the results of this study could mean much more to Kentucky than to other states

Returning to the initial question as to if Kentucky should expand Medicaid coverage and should the Oregon results influence this decision? I think the results are important to Kentucky. The governor must decide what are the state’s priorities with respect to health insurance.

More than 250,000 Kentuckians will benefit from expansion, but the studies are clear that health care expenditures will rise and are estimated to rise by 6.3 percent in Kentucky according to Center on Budget and Policy Priorities. However, the expansion reduces the probability of medical bankruptcy to nearly zero and will reduce the looming mental health problem among the uninsured.

 

Print Friendly
Join The Discussion