Does Medicaid Kill?

Several people have written to ask what I think of this piece by Avik Roy on Medicaid outcomes, which discusses a study indicating that people on Medicaid do worse than people with no insurance at all.  I suspect that many of them were hoping to catch me in a bit of hypocrisy, as I rushed to support the study with the findings that undercut the rationale for national healthcare.  And indeed, if you think these are right, you should be horrified by PPACA, which got more than half of its coverage boost by putting people into a program that will kill some of them before their time.

What do I think?  This is not actually a surprising new finding; it turns up off and on in the literature on insurance and health care outcomes.  I'm glad that Roy blogged the study, but I'm skeptical that this reflects the relative benefits of Medicaid, for the same reasons that I outlined in regard to studies of the uninsured.  Unobserved variable bias is incredibly hard to combat in these studies. 

The authors controlled for many, many possible variables, including a long list of comorbidities.  But consider the pretty convincing evidence that people with lower social status do worse on health outcomes than those higher up the SES ladder; the evidence that blacks do worse on a range of health care measures, particularly premature birth and infant mortality, even when we control everything that we can think of; and now even a study that finds people with fewer social contacts tend to die younger. 

Some of these things may reflect actual practical reductions in the quality of your care, because you have fewer people to check up on you, monitor your doctors, or make sure you take your medicine and drive you to PT.  Some of it probably reflects that fact that people with poor impulse control don't take care of themselves either financially or physically, and so are disproportionately likely to end up on Medicaid.  But some of it genuinely seems to reflect the effect of primate status-competition on the brain and body.   It's not a crazy stretch to think that people with fewer social contacts and lower social status are more likely to end up in the state-run health care system, dragging down its performance metrics.

Moreover, everyone I talked to when I wrote my piece on mortality and insurance agreed that Medicaid did have measurably positive outcomes in a few cases; prenatal care and hypertension control were the two most often cited.  Even if you accept that Medicaid patients are likely to do worse in surgery, they might still be faring better on certain preventative measures.

That said, I take seriously Roy's warning not to reject the notion that Medicaid might be worse than no insurance simply because it violates our "common sense" intuition.  Policy history is full of "common sense" policies that didn't work, and our intuition that Medicaid must be better than nothing is not obligated to actually be correct. 

Everything I know about Medicaid confirms that it is a terrible program on many levels, with a dysfunctional payment system and a byzantine bureaucracy, and procedures that vary wildly from one state to the next.  While I assume it is probably better than having no insurance, I wouldn't be surprised to find out that it's not much better. And whether or not you think it is actually killing people who would be better off with no insurance at all, there's no question that it could be massively improved.