The Truth About Medicaid

With the Senate health care bill finally out in the open, a tediously familiar game is unfolding once again. The game is to defend massive Medicaid cuts by claiming that Medicaid is useless anyway. In fact, maybe worse than useless. You see, people on Medicaid don’t have mortality rates any better than people with no insurance at all.

This is tiresome, especially since it invariably comes from folks who have private insurance and would sooner cut off their big toes than give it up. My guess is that if they were suddenly poverty stricken, not a single one of them would choose to go uninsured rather than accept Medicaid.

But that’s just my frustration talking. Instead, let’s talk facts. There are two big reasons why I find this nonsense so annoying.

First, improving mortality is hardly the only goal of medical care. For most of us, it’s not even the main goal: the vast majority of doctor visits aren’t for life-threatening conditions. So even if it were true that Medicaid did nothing to extend lives, it still does plenty. If you get a toothache fixed, or your migraine headaches treated, or a meniscus repaired, it doesn’t affect your lifespan at all. But it sure makes you feel better. And if, in addition, it does this without subjecting you to $10,000 in debt and a horde of bill collectors on your ass 24/7, that’s pretty damn helpful too. Unless you’re explicitly looking for a cynical reason to claim that Medicaid is useless, you just can’t pretend that this stuff doesn’t matter.

Second, the average age of adult Medicaid recipients is 38. At that age, there isn’t much mortality in the first place: less than 2 per thousand each year. To meaningfully distinguish mortality rates given such a low baseline you’d need a huge study. And it better be a pretty good one. General mortality (as opposed to mortality for, say, a specific chemotherapy drug) can only be properly estimated if you follow a population for five or ten years. So it needs to be a long-term study. And it needs to account for the fact that the Medicaid population is different from the general population: poorer, sicker,¹ more stressed, more disabled, etc. I don’t know if any studies have ever been done that meet these criteria.

However, there have been studies. Warts and all, they’re the best we have. The most famous is the Oregon Experiment, which is especially useful because it used genuinely identical treatment and control groups. However, each group consisted of only 10,000 people, which means the expected mortality in each group was 18 per year. That’s so low that the authors of the study couldn’t reach any conclusions at all about whether Medicaid had any effect on mortality. They did, however, draw null conclusions about other health indicators like blood pressure and cholesterol. The problem, again, is that the treatment group was so small that it’s hard to put much stock in this, especially since the study only lasted two years. However, if you nonetheless take their null conclusions about blood pressure and cholesterol seriously, you also have to take seriously their extremely positive results about depression, self-reported health, decrease in pain, and lack of financial catastrophe.

Either way, that’s only one study. There have been others. A trio of researchers reviewed Medicaid studies recently in the New England Journal of Medicine, and came to these conclusions:

The Oregon experiment found…significant increases in the rate of diagnosis of diabetes…along with a near-doubling of use of diabetes medications…. Meanwhile, the Oregon study found…a 30% relative reduction in rates of depressive symptoms.

….Multiple analyses have found improved self-reported health after the ACA’s coverage expansion, either in broad national trends or Medicaid expansion studies….Self-reported health is a validated measure of the risk of death….One study compared three states implementing large Medicaid expansions in the early 2000s to neighboring states that didn’t expand Medicaid, finding a significant 6% decrease in mortality over 5 years of follow-up….A more recent analysis of Medicaid’s mortality effects was one life saved for every 239 to 316 adults gaining coverage.

….One head-to-head quasi-experimental study of Medicaid versus private insurance, based on Arkansas’s decision to use ACA dollars to buy private coverage for low-income adults, found minimal differences. [That is, Medicaid was as effective as private insurance.]

That’s pretty impressive, and it doesn’t even count aching teeth, repaired cartilage, treated migraines, or any of the hundreds of other routine things that don’t show up in studies of chronic conditions. Nor does it validate the frequent claim that Medicaid is useless because doctors won’t see Medicaid patients: most studies show that Medicaid patients did indeed see doctors when they needed to and were generally pleased with their treatment.

That said, there’s a pretty obvious reason that Medicaid might not always produce dramatic differences compared to people with no insurance: by hook or by crook, the uninsured do get treatment for serious conditions. The main difference is that they have to pay for it. Here are the results of the Oregon experiment on the finances of the folks who received Medicaid:

Adults on Medicaid had 81 percent fewer catastrophic expenses. And in this case, the methodology of the Oregon Experiment is fine. A sample of 10,000 is plenty big and a timeframe of two years is plenty long. These are not things that affect only a tiny percentage of people, or that need years to show an effect. In addition to its medical benefits, Medicaid is a financial godsend for poor people.

One final thing: the folks who claim that Medicaid is useless never provide any feasible mechanism for this. Statistical studies are fine, but if they really do show null results for Medicaid coverage, there has to be some plausible reason why it produces no results. If we assume that medical care in general has positive results, why wouldn’t Medicaid? It may not provide the quality of care of private insurance, but it’s hardly the domain of quacks and shysters. It pays for care by real doctors and real hospitals, and it would be remarkable if that care really did no one any good. So what’s the theory here?

None of this means that Medicaid is perfect. It reimburses too little and really does seem to fail on some measures. It’s a serious drain on state budgets. And it has other problems as well that we ought to address. Liberals and conservatives alike should be dedicated to continually improving it.

Of course, speaking for myself, I’d love to eliminate Medicaid. And Medicare. And Obamacare. And employer insurance. A simpleminded national health care system would cover more people, almost certainly cost less than our current hodgepodge, and produce better results than what we get now. But that’s pie in the sky in this particular political moment. Given the reality of what we have, Medicaid is a critical part of our health care hodgepodge. Slashing it in order to give a big tax cut to the rich is obscene.

¹You may be surprised to learn that low reimbursement rates aren’t the only reason doctors are reluctant to see Medicaid patients. It’s a big factor, but nearly as important is that Medicaid patients take up so much time. This is because they’re generally sicker than other patients. So doctors are getting paid half as much for patients who have much more complex medical needs than average. The general stress of being poor accounts for most of this, but lack of consistent health care over their lifetime probably accounts for some of it too.