Healthcare policy should be most interested in improving the health of the country. We aren’t looking to just spend as much as we can on healthcare. We are looking to make sure our system serves as many citizens as possible. We want a system that meets all of our health needs and makes it easier for more people to live healthy lives. So when our presidential candidates discuss plans for universal healthcare it is always packed together with the assumption that universal healthcare would improve the health of more people in our country. After all, more healthcare seems to mean better health. Unfortunately that isn’t as certain as most of us would think.
(It seems helpful to frame policy questions around certainty/probability. We have some desired goal — in this case health outcomes — and we have a broad menu of possible policies that have different estimated probabilities about achieving our outcomes. These estimated probabilities can be informed by past research and other countries.)
There have been two major long-term experiments on the impact of free healthcare on health outcomes in the United States: (1) the 1982 RAND Health Insurance Experiment, and (2) the 2008 Oregon Health Insurance Experiment. Both had similar experimental designs. They both randomly assigned subjects to different insurance plans. The Oregon experiment assigned low-income people to a free Medicaid plan or no plan according to a lottery. The RAND experiment assigned folks of varying income levels to one of five groups: fully insured, cost sharing with 25% covered by families, cost sharing with 50% covered by families, cost sharing with 75% covered by families, and a regular HMO with the entire cost covered by the families.
The Oregon experiment came from a natural policy plan that would have happened without the study. The state opened universal coverage medicaid up to a lottery of roughly 90,000 people. It then studied the health outcomes of the lottery participants depending on whether they got the coverage or not. The study did not have any explicit controls and it only looked at low-income Oregonians. The RAND study was more explicitly organized as an experiment and, with that, it was able to control the random assignment of coverage along many variables — income, rural or urban, etc. The RAND study also only covered 7,700 families.
The results of both experiments are complex and have been re-interpreted many times over. Both saw health improvements in certain problems for the fully insured groups. The RAND study saw that “ free care led to improvements in hypertension, dental health, vision, and selected serious symptoms” and the Oregon study found that “Medicaid increased health care utilization, reduced financial strain, and reduced depression.” However, neither study saw a statistically significant impact on health outcomes and well-accepted health markers like cholesterol or blood pressure.
Now remember, to me, this is all about a probability game. When evaluating universal healthcare what degree of certainty do we have that universal healthcare would improve health? When you take the two most important studies on the question, both studies show that we can’t be as certain about health improvement as we would intuitively assume.
This is not to say that there aren’t flaws in the studies and there is plenty to say that they may not be very informative about the success of a plan like Bernie Sanders’ “Medicare for All”. After all, neither study looked at lifetime health outcomes, and both had relatively small research periods given the length of a life. Maybe there is some small chance that long-term universal healthcare could make a positive impact on health outcomes. I’m still agnostic about this question, and, full disclosure, I support universal healthcare because of different reasons.
My main point is that we need to be more cautious about the assumptions we are making when assessing universal healthcare, because the odds are that it wouldn’t make us healthier.