Obamacare Case Is Not Life or Death

By Megan McArdle

Wednesday, March 04, 2015

Earlier I wrote about the King v. Burwell case, which was heard before the Supreme Court today and will decide whether insurance subsidies can be offered on federally run health-care exchanges. Specifically, I wrote that the folks who are screaming that a ruling against the government would deliver a mortal blow to the legitimacy of the court are themselves the ones with a legitimacy problem. "Nice court you have there ... it'd be a shame if anything happens to it" is not the tone that responsible citizens should take with one of our most important institutions.

This morning, someone on Twitter explained that this case really is different because if the Supreme Court rules the wrong way, thousands of people will die. I find this explanation wholly unconvincing, for two reasons.

First of all, this is not somehow different from other controversial cases in the court's past. If you think that abortion is murder, then the Supreme Court killed millions of people when it legalized the practice. Thousands more died because the Supreme Court reauthorized the death penalty, after a brief attempt to rule the practice unconstitutional. The Supreme Court rules on all sorts of issues, from wars to environmental regulations, that potentially save or cost lives. Few people argue that this means the court is illegitimate, nor should they; this is, in fact, the court's appointed job.

For the record, I am pro-choice and anti-death penalty. But this is neither here nor there: The point is that if you think you are the first person outraged because the Supreme Court is ruling the wrong way on matters of life and death, then you need to crack open a history textbook.

Unless you think that conservatives are entitled to declare the Supreme Court illegitimate because of Roe v. Wade, then you should not make similar threats about a ruling that you deplore. As I said earlier, there are things even more important than Obamacare, and holding the country together is one of them.

But I thought I'd take a minute to deal with an even more fundamental question: How sure are we that the Affordable Care Act is saving lives?

At this point, the answer is "not very." I think it's quite possible that five years hence, the mortality rate charts will show a sharp inflection point starting around 2014. However, I also think it's quite possible that five years hence, the mortality rate charts will show ... nothing at all. Forget those eye-popping statistics we've all heard about how a lack of insurance costs tens of thousands of lives a year. When you look at a broad array of studies, the evidence is surprisingly mixed, with some studies showing large effects, and others showing no improvement in mortality. Two randomized controlled trials, the gold standard for medical research, have been done on the benefits of generous health coverage; both showed no significantmedical impact from paying for people's health care. Oh, you can pull out subgroups and say that hypertension got better among these folks, but there's a risk with that sort of thing that surprised researchers are simply committing the Texassharpshooter fallacy.

Now, that doesn't mean there's no effect -- the studies are hard to do, and even a very well done study (which both of these were) can show spurious effects simply by random chance. Moreover, it's pretty clear that there are significant nonmedical impacts; people like having health insurance, and it protects them from the negative financial impact of large medical bills. This is not the most surprising result, either: If you pay for stuff that people buy, they will have more money in their pockets, and they'll be pretty happy about that. But it is an impact. Whether you think that was worth spending $1 trillion on is an exercise that must be left to the reader.

I know what you're thinking: This must be baloney. Obviously, health insurance makes us healthier. How could it not? It buys us health care.

All I can say is that history is filled with obvious things that turned out not to be so. Our intuitions are a poor substitute for evidence -- and people's intuitions are doing a simply enormous amount of lifting in the interpretation of the various studies about the benefits of health insurance.*

Let me suggest some ways in which our intuitions might be wrong. For starters, most people do not go without insurance for lengthy periods; most of the uninsured are without insurance for a relatively brief time, which may be too short to affect health outcomes. People without insurance might be finding other ways to access basic care, which is the sort that is most likely to work (experimental treatments sound marvelous, but the reason they're "experimental" is that we don't yet know whether they work). Too much health care can be as bad as too little, as every procedure or hospital stay carries with it a risk that something will go wrong, and these effects could cancel out the benefits of having someone pay for your hypertension drugs. People with good health insurance might take more risks ("the Peltzman effect") because they're no longer worried about catastrophic medical bills.

I don't know if any or all of these explanations are the correct ones. There might be some other factor I haven't considered, and of course, health insurance may well reduce mortality rates, leaving us nothing to explain away. My intuition says it does, but the effect is probably very small, because if it were big, it should show up much more unambiguously in the data. But my read of the evidence includes the possibility that giving people insurance doesn't save any lives at all.

The point is that while Obamacare's supporters have an entirely understandable assumption that it's obviously saving lives, that assumption is still open to question. So my interlocutor is wrong on both counts: The Supreme Court has made rulings in the past that potentially involved large numbers of deaths, some of which the Illegitimators support with no apparent sense of irony. But it's not actually clear that this will be one of them.

   *For example, the studies that show huge negative impacts from being uninsured also show that Medicaid kills you -- that is, Medicaid patients have even worse outcomes than the uninsured. The authors are quick to explain this as a matter of omitted variable bias: Medicaid patients may superficially look like other people, but in fact they're poorer and sicker than the rest of the population. I find this entirely plausible, since it seems unlikely that having Medicaid kills you. However, the authors seem much less motivated to look for omitted variable bias when they are considering their entirely intuitive finding that the uninsured are more likely to die -- even though it seems very likely indeed that people who have poor impulse control and difficulty sticking to a routine are also more likely to end up unemployed and uninsured ... and to skip their medications, ride motorcycles without a helmet and otherwise endanger their health.