The University of Michigan
Q&A with David Meltzer, M.D., Ph.D.
David Meltzer, M.D., Ph.D., Associate Professor in the Department of Medicine, Graduate School of Public Policy and Department of Economics at the University of Chicago, discusses why health insurance coverage expansions may not be the best way to improve the health status of the uninsured. A health economics and outcomes researcher, Meltzer is a member of the Institute of Medicine's Subcommittee on Health Outcomes for the Uninsured, and co-authored "What Do We Really Know About Whether Health Insurance Affects Health?" for ERIU.
Q: It's widely perceived that health insurance coverage impacts health status or health outcomes. However, your work indicates that this is not the case. Why?
Our work doesn't argue that health insurance does not impact health, only
that much of the evidence that claims to show that is less conclusive
than one would like. The literature clearly shows that health insurance
coverage is correlated with health status, so people who are better insured
tend to be in better health. The question is, 'What drives that correlation?'
And, 'Is there a causal relationship that people who have better insurance
have better health because they have insurance?' And that's a lot harder
Q: Why is that harder to know?
One doesn't know why someone has or doesn't have insurance. They could
have insurance for reasons that also might affect their health. They might
be more affluent. They might be ill. They might have better social contacts
or be more educated, which makes it easier in some ways to get access
to insurance. Forces that are likely to get you health insurance are also
likely the forces that might make you have better health.
Q: What are the shortcomings of research that have people believing that health insurance coverage impacts health status?
There are probably a thousand studies in the literature that show the
correlation, but less than a dozen really have a strategy that gets around
interpreting this relationship as more than just being a correlation.
It's this ability to discern a causal effect, which is the crucial problem
in the observational studies.
Q: You cite a handful of studies that try to control for some of these forces. What do they show?
A: The studies that have been done of so-called natural experiments or policy evaluations have in general suggested there are health benefits to expansions of insurance. The studies do suggest a connection. But the point is there are relatively few of them. They've looked at these expansions in relatively narrow contexts. A lot of them are focused around kids.
Nevertheless, what we've got does seem to suggest that
health insurance makes a difference; that it does improve health. But
that's not enough of an answer. We'd be shocked if you could spend a whole
ton of money on health insurance and it doesn't do something. The real
question is: 'What is it doing? And is it worthwhile compared to some
of the alternatives?' If we had better studies, we could probably really
make a big difference.
Q: What is the danger in health care policymakers relying on the observational studies that have found a coincidental relationship between health insurance and health, as opposed to a "causal" or more direct relationship?
The danger is you're going to conclude that the effects are other than
they are, and you'll choose to spend money on an intervention that will
not have as large benefits relative to cost as other interventions you
might have chosen. Maybe we face the choice between further raising the
income threshold that makes poor families eligible for coverage as opposed
to targeting families based on where they live, expanding coverage to
just below Medicare age, or funding public hospitals. Any of those presumably
will have different costs and different benefits. And if we base our analyses
on studies that are potentially biased, then we may make the wrong decisions.
Q: In its recent "Care Without Coverage" report, the Institute of Medicine concluded that health coverage effects health status. Is the IOM finding largely based on observational studies?
A: Yes. The study concludes that the weight of the evidence based largely, but not exclusively, on observational studies suggests that there are substantial health effects of health insurance. I don't disagree with the spirit of the conclusion. The IOM committee was sufficiently convinced by the observational studies that it felt it was urgent to push ahead a policy agenda. I don't necessarily disagree with that, but I feel it's a little like the off-label use of a drug. We don't have randomized clinical trials to tell us that this is the right thing to do, but the evidence we have suggests the drug may work for patients with other conditions. It's all we've got, so we forge ahead. That's not unreasonable, but it is important to understand the strength of the evidence we have as we move ahead if we are to realize the best possible outcomes from our investments.
Do I really believe in the end that we'll discover that
health insurance will improve health? I do believe that, but it's a belief.
And I'm quite confident that beliefs won't be the way to identify the
perfect health insurance policy. Certainly, it may get us to a better
place than where we are, but ultimately the right answer is in systematic
Q: Do we know how health insurance coverage or what aspects of coverage can improve health for those studies that show there is a positive impact?
A lot of examples we see are from maternal and child health because that's
where the big expansions have been. There we've seen some health effects.
Some of the work that's been done on the establishment of Medicare suggests
that it makes a big difference to older people. Those are some examples
we know about. A theme that comes through that is very interesting is
that in a lot of these studies among adults one sees improvements in the
control of high blood pressure. That's interesting because the biggest
declines in mortality in the U.S. over the past couple of decades have
been in cardiovascular disease. One of the crucial things in that is controlling
high blood pressure. There is probably a whole series of very fundamental
insights into health and medicine that really come out of thinking about
how health insurance affects health that we've only begun to scratch the
What are other possible primary factors to improving health status?
Q: How are observational studies useful?
They demonstrate differences, and when one sees differences one asks questions
about why those exist. When you do a drug trial, you often do it on animals
before you do it on people. Even though that's not exactly the information
you want, you need to be able to get some basic answers to questions to
figure out how to design the study you ultimately want to do. You can
think of an observational study as a strong first step toward designing
So how do we get the best bang for our buck when it comes to improving
We need to spend more money on research that evaluates insurance, the
effectiveness of health care technology, and the operation of the health
care system. We are spending way too little on research, particularly
health services research. The doubling of the NIH budget is a great start,
but this is an area of profound social productivity where the public sector
is essential because it plays a dominant role.
Q: How important is it to cover the uninsured?
There are ways to spend money on the uninsured, on improving their coverage,
that seem likely to be very favorable social investments compared to other
social investments that we make. I think there are very likely programs
to expand coverage for the uninsured that would turn out to be worthwhile
in terms of the health benefits that they would produce relative to their
costs. But identifying what the ideal one is, is something that requires
more research than we have done yet.
Funded by The Robert Wood Johnson Foundation, ERIU is a five-year program shedding new light on the causes and consequences of lack of coverage, and the crucial role that health insurance plays in shaping the U.S. labor market. The Foundation does not endorse the findings of this or other independent research projects.