Yes, Health Insurance Saves Lives — Part Two

With thanks to Ross Douthat for replying to my post yesterday about Medicaid and mortality, I want to take issue with the following arguments.“[W]hen partisans of the Affordable Care Act predicted that the law would have a significant impact on mortality, were they really just staking that claim on the specific phenomenon of people who (hypothetically) postpone an emergency room visit during a heart attack because they lack insurance? My impression was that the argument was much more holistic, emphasizing insurance’s positive impact on health across the board. (Here’s Ezra Klein, for instance, arguing for a strong mortality-insurance link: “The uninsured are less likely to seek early care. They are less likely to get good care. They are less likely to return for follow-up care. They are less likely to be able to afford the maintenance of chronic conditions.”) And if so, I don’t see how you can draw the kind of clear division that Beutler does between studies that measure excess deaths and studies that measure health indicators. Over the long run, a significant link between insurance and mortality would pretty much have to be associated with precisely the kind of significant link between insurance and health indicators that the Oregon research failed to find.”

Emphasis mine, and there I’d say two things. First, I think it’s reasonable to expect that if insurance and mortality are linked, then insurance and key health indicators will be linked as well. But it doesn’t absolutely have to be the case. As health economist Austin Frakt noted in an email to me, “it’s possible insurance saves lives apart from its impact on management of chronic conditions.” People can die for lack of insurance for a variety of reasons (more on that below) and you wouldn’t necessarily see it reflected in aggregate data as higher blood pressure, glycated hemoglobin or anything else.

But moreover, even if Douthat’s claim were unimpeachably true, the Oregon research is still not sufficient grounds upon which to deny a link between insurance and mortality — which was a point I tried to get across yesterday.

Stepping back a bit, though, I think what’s going on here is that Douthat has mistaken advocate passion for argumentative substance. Click over to Klein’s post and you’ll find a write up of a study linking uninsurance to mortality — the same link, I noted, that the Oregon Medicaid study wasn’t designed to address. Klein’s claims might appear to be more holistic. But they’re ultimately about the same phenomenon my thought experiment was intended to illustrate, and that excess death studies have actually measured, with respect to a variety of ailments.

I wrote about this last year, when Mitt Romney claimed people don’t die because they lack health insurance.

One of the researchers who found an insurance-mortality link explained to me how data match what for many liberals was an intuitive sense that Romney was wrong: “We know that women with cervical cancer who are uninsured get their cancer detected later…. We know that people with heart disease don’t take their medicine because they can’t afford it…and sometimes die.”

And indeed most, though not all, of the research suggests this adds up to some tens of thousands of deaths a year. That’s a tenth of a percent or less of the total number of uninsured people in the U.S., so it’s not something that’s likely to happen to a typical uninsured person. But there are a ton of uninsured people in the country, and so a marginal phenomenon like this can nevertheless translate into a large cumulative number of deaths.

If preventing these deaths was the only goal of covering the uninsured, then the debate over Medicaid and the ACA would be a sort of bloodless but necessary one about whether it makes sense to spend hundreds of billions a decade to accomplish it. Obviously, the argument for universal coverage is much more extensive than that. But Medicaid skeptics, including Douthat, want to home in on the narrower issue, because eradicating the link between insurance and outcomes — particularly mortality — strengthens the conservative case for limiting public provision of insurance to catastrophic coverage. The problem is, Douthat has cited one underpowered and off point study to refute or call into question years of research on the mortality question, rather than asking whether research on the mortality question ought to give him pause about reading too much into the Oregon story. That smells like confirmation bias to me.

Maybe time and more thorough research will vindicate his suspicions. But until then I call foul on attempts to marshal Oregon as an argument for providing fewer poor people less comprehensive health care.