Can you highlight the purpose and findings of your seminal study with University of Michigan economist Helen Levy?

MELTZER: Our work focused on one very specific question: How good is the evidence that health insurance improves health? Our conclusion is that the evidence is far more limited than people think. And, if your goal is improving health, then we have to think of a whole series of policy options beyond improving coverage to reach that goal. This includes things like the direct provision of health care. The practical implication of our work is that we need better research to understand the effects of lacking health insurance on health.

So, are you saying that because there isn’t hard evidence making a causal link between insurance and health that having coverage doesn’t matter?

MELTZER: Certainly there have been people who have chosen to interpret what we did that way. That was not our message. Our message is not that health insurance doesn’t matter but that we lack the data to determine for certain whether it does, for what populations, and to what degree.  Most of us believe that it probably does have a positive effect, at least in certain populations. But what is the magnitude of that positive effect? Is it sufficiently large given the costs compared to other things that we can do to address the health needs of people who currently don’t have health insurance?

What’s the problem with the evidence that does exist?  What kind of evidence are you looking for exactly? 

MELTZER: The essential problem is that most studies cannot fully control for the reasons people do or do not have health insurance, and that those reasons may themselves explain why people with health insurance have better (or worse) health than people without health insurance.  Ideally studies would randomize specific subgroups of people to receive health insurance or not, but no such studies exist. “Natural experiments,” for example public policy changes that expand insurance, are typically the best available strategy to assess the effects of heath insurance on health.

Has evidence emerged since your 2001 study to indicate a more causal effect?

MELTZER: The message has gotten out that while we have plenty of studies that show the association between not having health insurance and poor health, what we need are high quality studies – such as of these natural experiments – to give us a more accurate estimate of the role of health insurance on health. Having an accurate estimate is important because, as we all know, obtaining coverage is going to cost a lot. There are other things that we can do that also cost a lot and that can also improve health. We need to decide which ones of these should be the greatest national priority. The best way to do that is through high quality evidence.

Certainly we often have to make policy decisions without great evidence. But, it’s the responsibility of the researchers to do their best to try to make sure that, when evidence can be provided, it is provided. I think the most positive outcome of our paper is that it has forced researchers who are working in this area to think harder about whether they can develop causal estimates about the effect.

You and Helen Levy recently did an update of the topic for the Annual Review of Public Health. What does it show?

MELTZER: The central question of how health insurance affects health remains largely unanswered without a sufficient level of detail to inform policy decisions. The evidence available to date convincingly demonstrates that health insurance improves the health of vulnerable people such as infants, children, and people with AIDS.  Having coverage also can improve specific measures of health such as control of high blood pressure for adults, especially those who have lower incomes. For near-elderly people, there is conflicting evidence about the effects of insurance on health.

What about the evidence for others?

MELTZER: For most of the people at risk of being uninsured (adults ages 19 to 50), we have limited reliable evidence on how health insurance affects health. This lack of evidence and the resulting lack of consensus indicate that to summarize the effects of health insurance on health is, inevitably, to misrepresent it. Definitive answers to the multitude of important questions about how specific health insurance policy options affect health are likely to be forthcoming only with investment of substantial resources in social experiments.

How would you do a study?

MELTZER: There are some major things that we can do to do a better job of answering this, but it will cost some money. One example is another RAND health insurance experiment that really looks at health insurance.  The original landmark RAND health insurance experiment did not ask the question of what the effect of not having health insurance was on people. It looked at the effect of having health insurance that varied in the degree of co-payments. This is a very different question. So, you know, it’s not like we don’t have unemployed, uninsured people in America. We can randomize the health insurance. What a great study that would be!

So what benefits does health insurance afford?

MELTZER: It affords a lot of benefits. Health insurance decreases the likelihood people go bankrupt. It decreases anxiety; it makes the health care system easier to run. It wouldn’t surprise me if it makes the health care system more efficient in certain ways because we spend all sorts of money trying to figure out how to maximize revenue from the people we can get it from and use that to pay for the people from whom we can’t get money from.

So what’s the message for policymakers?
 
MELTZER: We are researchers and our most important message for policy makers is that we should focus on producing higher quality evidence from natural and randomized experiments about the causal effects of health insurance on health. In fact, even if you choose health insurance for other reasons, the types of questions you are going to be asking are going to be exactly the same questions. If you decide we may want to have health insurance because we don’t want to have people going bankrupt or we want to help the health care system stop worrying so much about payer mix and things like that, then, you still need to know which benefits should be offered.

It is a lot like the debate about comparative effectiveness. We spend so much money on health, and so little money figuring out when it is worthwhile. And, in the same way, we spend a lot of money on health insurance in this country. But how much do we spend figuring out when it’s really worthwhile?

What’s your message to researchers?

MELTZER: We are saying to researchers that if you are fortunate enough to be making it as a researcher, and being able to study things like health insurance, spend your time on studies that are useful – ones that can seriously assess causal association of policy-relevant changes in health insurance on potentially important outcomes, and not just studies that show associations. Now, is the fact that we don’t have these answers five years later or in the papers a failure? No. There are a number of good new articles that have come out since 2001. There is no way to say whether it would have gone more or less, without the article we wrote. But the real test is what does the next five years look like? Do we start to see movement towards these more rigorous studies? Do we start to get more answers to inform the policy today?

What can we expect realistically from health coverage, the achievements of health coverage?

MELTZER: One thing we can certainly expect is that people get more access to care. I would hope we would find a more even distribution of the financial burden. Do I think we would improve health in certain measures? Yes, I suspect we would. Depending on how you organize that health insurance, I could imagine a health care system that potentially worked more efficiently, particularly under certain environments.

Does coverage expansion make sense? If so, why?

MELTZER: I think to some extent, the senior decision has been made and the kids’ decision is cheap and largely already been made. I think coverage expansion for the near elderly is still an open question and it is a beautiful opportunity for a social experiment. It would be a very good investment of public dollars to try to convene a bunch of experts to think about how one would design a set of strategies to try to efficiently expand insurance coverage, and those strategies should include the possibility of social experiments that randomly assign people or groups without health insurance to receive it. You have to decide at what level to do them – at the level of the individual, the city, the township, the state, the county, etc.? This is exactly where we should be generating data.

If you had a large pot of money to improve health, how would you spend it?

Back to top