Network News

X My Profile
View More Activity

All you ever wanted to know about the research on health insurance and health -- and more

I think I've said quite enough on the question of whether health-care insurance reduces the risk of death. But one of the first things I did when looking into the subject was call Stan Dorn, the author of the Urban Institute study (pdf) that estimated 18,000 people died in 2006 because they didn't have health-care insurance. At the time, he said he was writing a response of his own, and would be in touch when he finished. Last night, he sent it along. It follows in full, and serves as a good introduction to the literature on this topic.

I have a few additional comments explaining why Megan McArdle’s article does not effectively refute the substantial evidence linking health insurance and mortality. She fairly characterizes as “a trifle elderly” 1993 and 1994 studies that the Institute of Medicine (IOM) used in 2002 to calculate that 18,000 adults died in 2000 because they lacked health insurance. However, McArdle completely ignores the IOM’s 2009 follow-up report finding that “the body of evidence on the effects of uninsurance on adults’ health has strengthened considerably since 2002. Numerous studies have addressed some of the methodological shortcomings of past research… The quality and consistency of the recent research findings is striking.”

Testifying to Congress last year on behalf of the IOM, John Ayanian, a professor of medicine and health care policy at Harvard Medical School, reaffirmed, “Uninsured adults are 25 percent more likely to die prematurely than insured adults overall, and with serious conditions such as heart disease, diabetes or cancer, their risk of premature death can be 40 to 50 percent higher.” The conclusion that uninsurance increases risk of death by 25 percent was the basis of IOM’s 2002 finding of 18,000 deaths resulting from a lack of health coverage as well as our 2008 updating of IOM’s calculations.

Research on the link between mortality and insurance goes far beyond the few studies discussed by McArdle. The IOM conducted rigorous, comprehensive literature reviews in both 2002 and 2009. The first report described 19 peer-reviewed journal articles finding a statistically significant relationship between insurance status and mortality, compared to 5 that found no such relationship. The 2009 update described 30 additional studies that found a robust link between insurance status and health outcomes (including mortality), compared to 9 that found no such link.

The IOM observed that insurance especially matters for adults with chronic illness, who comprise 40 percent of the uninsured: “health insurance is clearly most beneficial for adults who need medical attention, particularly for adults with common chronic conditions or acute conditions for which effective treatments are available.” Both IOM reports cited abundant research finding that the uninsured with chronic health problems receive later diagnoses and less care, so a lack of insurance coverage substantially increases mortality rates among cancer patients, heart disease patients, older adults, and hospitalized patients in general.

Recent research discussed in the 2009 IOM report imposed tight controls that addressed the methodological challenges raised by McArdle. And several noteworthy studies, both before and after the 2009 report, took advantage of “natural experiments” to isolate the effects of health coverage on health status and mortality:

1) When California terminated Medicaid coverage for childless adults in 1982, “excess deaths [for hypertensive patients] were evident within 6 months of losing insurance, and the estimated risk of dying was increased by 40%.” Link.

2) When New Jersey eliminated its subsidies of hospital care for the uninsured in 1994, death rates among the hospitalized uninsured rose by 41 to 57 percent, while such rates were unchanged in other states. Link.

3) When the near-elderly uninsured receive Medicare, they experience significant improvements in their control of cardiovascular disease and diabetes, fewer declines in health status, and greater overall health. Further, death rates among acutely ill, hospitalized patients decline by 20 percent when people turn 65 and qualify for Medicare, as noted in your post on Friday afternoon. (The latter effect is not limited to the uninsured gaining coverage, however.)

4) The uninsured in severe automobile accidents receive 20 percent less care than the insured and die at rates 39 percent higher.

As found by the IOM’s literature reviews, the research evidence is not unanimous. For example, Richard Kronick’s solid study, referenced by McCardle, found no relationship between mortality and health insurance status. However, the main point of Kronick’s study is that some of the earlier research may have overstated the effect of insurance on mortality by omitting important variables. Kronick’s study had its own problems because, as his paper alludes, he was not able to address a critically important methodological issue—namely, that people in poor health are more likely to seek health insurance, which obscures any positive relationship between health insurance and health status. Studies that adjust for this factor have found a statistically and quantitatively significant relationship between lack of insurance and increased mortality risk.

After conducting thorough reviews of the research, IOM in 2002 and 2009, McWilliams in 2009, and Hadley in 2003 all concluded that the clear preponderance of findings from well-designed studies strongly link insurance coverage and mortality rates. McCardle erred by presenting the Kronick study as the gold standard for research on this issue to the exclusion of all studies published since 1994 that go against her argument.

Of course, the number of adults who die because they lack health insurance cannot be defined with absolute precision. Our earlier report found that improving the IOM’s methodology would increase the number of estimated deaths by more than 15 percent. And a recent Harvard Medical School study concluded that, because health care now controls disease more effectively than in the past, the lack of health insurance increases mortality by 40 percent, rather than the 25 percent estimate used by the IOM and our earlier study.

The mortality estimates reported by the IOM in 2002, updated in our 2008 report, are best viewed as indicating the likely magnitude of loss of life that results from tens of millions of Americans lacking health coverage. The exact number of fatalities may be somewhat higher or lower, but it is surely quite large.

At bottom, McCardle suggests that health is not crucially affected by the access to health care that insurance provides. I doubt that many of us fortunate enough to have health insurance would drop coverage based on the strength of her arguments.

By Ezra Klein  |  February 16, 2010; 7:00 AM ET
Categories:  Health Coverage  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Tab dump
Next: Valuing life


Excellent response. Now can we all agree that McArdle is not a worthy commentator and ignore her? Please.

Posted by: Kew100 | February 16, 2010 7:37 AM | Report abuse

it seems that you wouldnt need a study, just common know that people who are ill, need medical care, and live longer with it.
and if they cant receive it, they become more ill, and have a good chance of not surviving.
think of wounds, of diabetes, of serious auto-immune diseases, of bad strep throats, allergic reactions, infections....of everything.
would ms mccardle like to give it up her health insurance and participate in this study, so she can see if she lives longer, when she is ill and cant get to a doctor?
i dont think so.

Posted by: jkaren | February 16, 2010 9:24 AM | Report abuse

Seriously, McCardle has completely thrown her lot in with the anti-science, anti-expertise teabagging fringe. No number of IOM reports will convince these people - they really believe that universal health care is an assault on freedom and I am really convinced that this is not the era for trying to even have debates with people like McCardle. The only hope for success in this day and age is simply to ram through ideas as fast as we can and let demonstrable success do the talking (although even success won't really change their minds, just look at the ridiculous language about Medicare; we'll just have to live with the cold comfort that our ideas are objectively better and make people's lives better even if many of them are ingrates about it).

Posted by: reader44 | February 16, 2010 9:54 AM | Report abuse

I'm late to this party, but here are two points:

1) I suggest that the debate as described is a little unsophisticated. Providing access to heath care clearly has benefits. How much is the debate you've been focused on. The best metric to be evaluated is NOT lives saved. Its number of life years extended. The question that SHOULD be asked is not does health insurance save lives, but is this really such a great expenditure ($100 billion/year) that its a clear moral no-brainer to do this, versus other things that could have better impact with the same or fewer dollars. The answer to that question is a lot less clear. Back of the envelope math suggests the following: Assume a high estimate of 50,000 deaths prevented each year. Assume that these avoided deaths added 20 years of life on average. (Note that these are generous assumptions.) That's a million life years saved, at an expenditure of $100 billion a year. That's $100,000/year life saved. That's borderline cost-effective based on health economics literature/standards. If you assume 25,000 deaths or lower, or only 10 years saved on average or lower, then it starts bordering on cost-ineffective. This ignores the likely double-counting of deaths avoided on an annual basis, i.e. is it a different 50,000 people saved next year? There's probably some overlap. Overall, its borderline cost-effective. Which gets to...

2) The more relevant point is that there are clearly are number of public health interventions that would cost significantly less than $100 billion/year that would save more lives. Food policy. Obesity policy. Sociodeterminants of health. The recent study on modest reductions in salt intake suggested life savings on par with the numbers cited here. These would require a mere fraction of the spend. Yet none of those things are being pursued, and we're instead focused on the more tangible, but costly issue of extending health coverage. Its worthy to be sure, but its unclear that this should be prioritized before 10 or 20 others with similar impact that require significantly less than $100 billion per year.

Posted by: wisewon | February 16, 2010 10:58 AM | Report abuse

I think you could also do a mortality rate study using education level, and type of job and find similar results. Usually the same people who don't have health care problably don't have a job that provides health insurance. Also have you seen the numbers of people who don't take the time to fill out paper work and apply for state insurance programs they would qualify for. Individual responsibility and initiative has made this country great. People need to start taking responsibility for themselves and their actions. We don't need a nanny state because of a segment of our population that does not have the initiative to take care of their situation.

Posted by: CommonSense61 | February 16, 2010 11:05 AM | Report abuse

These are pieces of information gleaned from sources, or research or second hand research, that can be disputed with the same "method" this information is presented: Some "facts" put together and accepted despite contrary opinion, possibly (or probably) taken out of context, etc.

This is not the "definitive" explanation of any of this.
The "fact" is the majority of the country has health insurance (estimates have been 300 million); a percentage of the country does not have insurance (estimates at 30 million - which may include non-citizens, those who have elected NOT to have insurance, etc.).
This is the fact. Whatever issues Mr. Klein wants to promote here are in dire need of examination and attention: Help the uninsured, who agree to pay some sort of premium and do this in incremental ways.
Why has this administration wasted everyone's time? Because President Obama decided for us that he wanted to fulfill whatever personal agenda for whatever reasons for his own ego or paybacks.

Posted by: pjcafe | February 16, 2010 12:35 PM | Report abuse

The only place an individual can buy care a la carte is in an efficient market fantasy world.

List prices for insurable medical services are absurdly inflated.

Insurance and Medicare negotiate/impose large discounts, so list price is irrelevant to the vast majority of people.

And there is no incentive to reduce list prices, since anyone who comes in as a sick individual is in no position to bargain, and if she is unable to pay the full amount, the hospital is better off starting with a high bar for a future negotiation, settlement or bankruptcy.

There are many reasons, like subsidies and bargaining power, why the employer-paid option is cheaper for Ms. McCardle... one of them is that the market doesn't offer an economical alternative.

Posted by: curmudgeonlytroll | February 16, 2010 12:40 PM | Report abuse

Ms. McCardle made the point, I believe it was yesterday, that she was not trying to argue that Health Insurance does not save lives, but that there were few studies that showed how many lives it saves. I think common sense does suggest that people with high quality health insurance live longer, and better, but that a) is not evidence, and b) does not tell me how many people would die if they went uninsured.

Posted by: jnewman418 | February 16, 2010 1:23 PM | Report abuse

Now that McMegan has been fully, completely refuted, the Glibertarians are falling back on their backup arguments.

The most entertaining one is the "it now costs ($100,000? $9,000,000?) per saved life, so it is not "cost-effective" to give insurance to lower middle class people."

Of course, the number seems to vary wildly, and all other beneficial effects of broadened health insurance coverage (lower absenteeism, lower long term care costs, reducing medical cost-related bankruptcies) are being ignored.

But for the moment, let's entertain that cost-effectiveness model. And immediately start 1) analyzing all defense and homeland security expenditures by that model and 2) assessing all drug purchases by that model, and aggressively negotiating down the price of all drugs to deliver that cost-effectiveness.

Oh, does McMegan and Glibertarian team never, ever question defense expenditures? And have they previously argued against negotiating with PHARma since they don't want to slow down that incredible stream of innovation?

But I thought libertarianism is the One True Faith that can be consistent in all respects. Oh, never mind.

I think Daniel Gross has this strain of Libertarianism right (and lumps McMegan and Tyler in with Broder and R. Samuelson): "Conservatism is that haunting fear that someone, somewhere, may be collecting on their social insurance."

Posted by: Dollared | February 16, 2010 2:25 PM | Report abuse

Mr. Dorn:
You say: "At bottom, McCardle suggests that health is not crucially affected by the access to health care that insurance provides."

McArdle explained three days ago that
"But I have not asserted that insuring the uninsured wouldn't save anyone's lives, so I don't know why they're making this argument while linking to me. What I said is, the studies so far done often cannot exclude the possibility that there is no effect--this is true of one of the two studies that IOM/Urban relied upon, and also of the largest observational study done to date, which found no effect. That is not the same as saying there is no effect. ... What I am saying is that we don't know how big the effect is."

It seems you are mischaracterizing her argument. (Please correct me if I am wrong, of course.)

On the larger issue, however, don't you think that the question of the size of the effect from extending insurance coverage to the uninsured is a wholly legitimate, indeed vital, issue relevant to the current health care debates?

Posted by: WmOckham | February 16, 2010 4:41 PM | Report abuse

This is a reply to WmOckham, from Stan Dorn.

Thanks for pointing to Ms. McArdle's recent post. I was addressing her article in the Atlantic, where she said, among other things, "Quite possibly, lack of health insurance has no more impact on your health than lack of flood insurance," and "it’s possible that, by blocking the proposed expansion of health care through Medicare, Senator Lieberman, rather than committing the industrial-scale slaughter Klein fears, might not have harmed anyone at all." The article is available at I read it to say that no solid evidence establishes that the lack of insurance causes a loss of life. In truth, a substantial body of evidence shows precisely such a causal impact.

And I agree that establishing the general magnitude of this effect is worthwhile. That's why we did the 2008 paper finding that, conservatively estimated, using IOM's precise methods, roughly 22,000 people died in 2006 because they were uninsured; but that using a slightly different (and improved) variant of the IOM methodology yielded an estimate of 27,000 deaths. But even if the number were as low as 15,000 a year, that would be enough to warrant policy change, in my view.

Posted by: stan_1036 | February 16, 2010 10:26 PM | Report abuse

The article hyperlinked in 1) where California terminated Medicaid coverage for childless adults mentions:

"Meyers et al conducted a pilot study of 25 physicians in a primary care research network in Washington, DC, finding that physicians reported making changes in clinical management of patients based on insurance status. In particular, in nearly one half of visits with uninsured patients, physicians reported a change from their preferred management, whereas such a change occurred only in less than one third (29%) of those who were publicly insured and one quarter (19%) of those privately insured (P = .01)."

In other words, access to the doctor and medical care was not the issue. The issue is with the treatment that the doctor proscribes based on insurance status.

As there are usually cheaper alternatives to the expensive solutions that doctors often recommend, such as older, less expensive medicines, dietary changes, exercise, etc., the problem may not be insurance coverage but doctor training about cost effective solutions to medical problems.

Posted by: MiltonRecht | February 17, 2010 12:43 AM | Report abuse

At its heart, I truly believe Megan and her readers know full well
that insurance helps and that is why they pay for it themselves. They
just don't want any of THEIR money going to pay for anyone else.
Despite being self-professed Randites, rather than arguing for
selfishness, they attempt to argue that their money wouldn't really
help other people anyway (even though these same purchases seem to
help them when they buy it for themselves).

In the process they butcher quite a bit of science and research and
make pretty bad arguments with pretty bad logic. Once painted into
the corner, they start backtracking, redefining past arguments, and
quibbling about small details. We are seeing plenty of that right

But none of it matters. It doesn't matter if we're discussing health
care, job training, food stamps, health/weight-loss programs, etc. I
bet all (or most) of these topics have been written about my Megan,
and I bet that in each case she argues that they don't really work.

This is because she'd rather by butter boats and lemon zesters than
help out people in need

Posted by: nylund | February 17, 2010 2:34 AM | Report abuse

They make ridiculous claims like this not because they think they can convince anyone that they are true, but to induce opinion leaders such as yourself to waste their time debunking them rather than building consensus for something real. It's a diversion attack. And it works. Look at the climate "debate".

Posted by: chase-truth | February 17, 2010 8:00 AM | Report abuse

Mr. Dorn,

Thanks. I think that's a reasonable response to my questions.

Posted by: WmOckham | February 17, 2010 9:39 AM | Report abuse

Ah yes, the statisticians quibble, while Americans die.

Don't you think that if Ronald Reagan were telling this story, he would provide some real live examples.

That is what is missing in this story - true stories.

Glenn Beck, that chipmunk in jackboots, weeps for America, but who is weeping for these 18,000 Americans.

I found the comments of Mr. Milton Recht (Falsch?) particularly curious.
Admittedly providers too often treat the plan and not the patient. But does that make the converse true, no treatment because no treatment is necessary? I don't think so.

I administer an employer sponsored health plan. We regularly get complaints from members who were turned away from the doctor's office because some computer glitch showed that they had no health insurance. One woman had a lump in her breast that the doctor could feel. He would not order a mammogram because he called the wrong carrier number and was told the member had no health insurance. He sent her home with instructions to call back when it was straightened out.

The daughter of another member was prepped for surgery to remove a tumor and was sent home, when it was discovered that she had aged off the plan after scheduling the surgery.

The fact that people die without health care is just one of many reasons for universal coverage.

You could argue, for example, how much we all pay for "uncompensated care" or how much we pay to cover people who are working for employers who don't provide health insurance.

But should we really be arguing about people dieing?

Have we sunk that low?

Posted by: Jimmy1920 | February 17, 2010 2:06 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company