Jonathan Gruber on the RAND HIE
What is Insurance?
by Jonathan Gruber
wbur.org (Boston’s NPR)
Commonhealth (a blog)
March 9, 2007
The debate over health care reform implementation in Massachusetts has become focused on the critical issue of defining the level of “minimum creditable coverage”, the minimum level of insurance generosity that will satisfy the individual mandate. This is not a decision that can be avoided; without such a definition, the mandate has no teeth.
But this decision has become needlessly controversial because the debate has moved away from the fundamental goal of insurance: to insure individuals against unexpected medical risk.
This definition of insurance has nothing to do with the typical American conception of health insurance. Most Americans think of health insurance as medical prepayment: you pay an up-front premium and in return all of your medical expenses are covered (what experts refer to as “first dollar coverage”). But such a system has an inherent flaw: any time something is free, it will be overused. This should not be a controversial statement to anyone who has ever gone to an all-you-can-eat buffet. Having paid at the door, you always end up eating more than if you were paying for each item your ordered.
But isn’t medical care different from food? Don’t individuals only use medical care when they really need it? The answer to this question is clearly no. There is substantial evidence of overuse of medical care by the insured. The classic piece of evidence for this proposition is the famous RAND Health Insurance Experiment (HIE) in the 1970s. In this experiment, individuals were randomly assigned to insurance plans that were more or less generous, ranging from first dollar coverage to policies where indivduials paid 95% of the cost of their medical utilization. Once individuals had spent $1000 (about $4000 in today’s dollars), however, they were fully insured — so no one was going uninsured in these plans.
The lessons of the HIE are clear, yet have been conveniently misinterpreted by parties on both sides of the debate. I review these lessons in a paper prepared for the Kaiser Family Foundation, available at http://www.kff.org/insurance/7566.cfm. First, medical care is price sensitive: those who paid nothing for their care used about one-third more than those who paid almost the full cost at the point of service. Second, on average, the extra medical care consumed by those who received care for free had no beneficial impact on health: individuals in the first dollar coverage plan were no more healthy, on average, than those in the full cost plan. Third, there is some evidence (although it is not very precise) that there were some health benefits for low-income, chronically ill populations; most of this benefit was found in screening for chronic illness and maintaining treatments for that illness. The 25 years of research since the results of the HIE were reported have done nothing to overturn these central conclusions, although no study has so carefully studied the health impacts of insurance generosity.
At the same time, there is a large literature in health services research which shows that having insurance, relative to having none, is important for health outcomes. The resolution between these seemingly conflicting results is that it matters whether individuals have some insurance, but, given insurance coverage, for the typical person it does not really matter how generous that insurance is. Once again, coming back to the buffet analogy, it is clearly harmful to not allow individuals to eat — but less critical that you allow them to eat as much as they want.
These results have two clear implications for health policy. First, it is vital that we provide health insurance coverage to the uninsured as a means of improving their health. Second, that insurance need not be as generous as first dollar coverage. There is no clear benefit to covering all medical costs, and a clear cost in terms of encouraging excessive use of medical care. Insured individuals should therefore bear some of the costs of their medical care utilization, but with a limit when costs rise to a sizeable share of income. In addition, insurance should provide reduced cost sharing for prevention and maintenance of chronic illness.
This suggests that the structure of the minimum creditable coverage plans that are being discussed makes a lot of sense. All of the plans being considered provide some up front medical care that individuals can use to get preventative care and be evaluated for more serious medical disease. If individual are found to be chronically ill, they can then buy up to more generous plans that are more appropriate to their health status.
Ideally, insurance plans would be more flexibly designed to accommodate the lessons of the HIE, for example offering different levels of out-of-pocket costs by income level or by illness level. There is some exciting work being done now on designing such “value-based insurance plans”, whereby co-payments are targeted to the needs of the particular individual being insured. But such plans are still in their infancy, so that today the choice remains between offering first-dollar coverage or a plan where individuals pay more of their costs up front. The evidence is clear that the latter design provides real insurance, in the true sense of the word.
Jonathan Gruber, Professor of Economics at MIT and member of the Connector Board
http://blogs.wbur.org/commonhealth/?p=23#comments
Comment:
By Don McCanne, MD
Dr. Gruber states that the RAND Health Insurance Experiment (RAND HIE) has been “conveniently misinterpreted by parties on both sides of the debate.” He then takes one side, stating that the lessons are that “medical care is price sensitive,” that “on average, the extra medical care consumed by those who received care for free had no beneficial impact on health, but that “there were some health benefits for low-income, chronically ill populations.” It would be incorrect to characterize this as a misinterpretation, because these statements are, “on average,” true.
But there is another side as to the importance of the findings of the RAND HIE, which also should not be characterized as a misinterpretation.
It is crucial to understand the nature of the population studied. The subjects were a gainfully employed workforce and their families, who were observed for a few years. Thus this was a study of relatively healthy workers and their healthy families during healthy years of their lives. Being employed, they had discretionary income sources that could be used for health care cost sharing.
The intrinsic validity of this study can be applied to similar healthy populations with reliable incomes, but it does not have extrinsic validity in a system of “universal coverage.” A system that would cover everyone must take into consideration the entire market basket of health problems, from conception to death. Even if sectors are segregated based on age (Medicare), income (Medicaid), employment (employer-sponsored coverage), or whatever, the remaining uninsured have a multitude of problems which result in their being uninsured in the first place. The RAND HIE conclusions would apply “on average” only to the employer-sponsored group, but not to these others, and certainly not to the uninsured.
There is also considerable question as to the overall impact on total health care spending. Although the 80 percent of us who are healthy would reduce utilization, we consume only 20 percent of health care. A modest reduction of only one-fifth of our health care bill would not have a major impact. The 20 percent of individuals who use 80 percent of health care, the group that was not studied in the RAND HIE, have little opportunity to selectively decline health care that seems necessary because of their health care problems.
Besides, is the RAND HIE “on average” conclusion satisfactory? Individuals in the RAND HIE with hypertension had a 10 percent higher death rate when cost sharing was required. The families of those who died receive little consolation in the fact that these deaths did not change the statistics of the overall beneficial outcomes of this healthy population.
Is health care about the majority of us who are healthy, or is it about prevention and management of disorders that can lead to disability or death?
Since we are not static beings, how can we possibly determine which of us will remain healthy and be well served by a private insurance plan which has an affordable premium because it doesn’t cover expenses that we won’t be facing anyway? But the real question is the reciprocal. Which of us will develop major medical problems that will require affordable access to health care made possible only by selecting a more comprehensive plan, but with premiums that we can’t afford?
So what is the real lesson of the RAND HIE? Cheap insurance works for healthy people. Period.
Since the RAND HIE was completed in the 1970s, there has been a plethora of studies in the health policy arena. The studies on cost sharing have clearly indicated that erecting financial barriers to care maim and kill people with health care needs.
The fundamental flaw with using private health plans to expand coverage to everyone is that private plans work only for healthy subsets such as employer-sponsored group plans. If those with health care needs are to be included with reasonable levels of coverage, the premiums become unaffordable.
Healthy individuals are now segregated into low risk pools (e.g., employer-sponsored) which, regardless, have become so expensive that they are now barely affordable. If we are to have universal coverage, these healthy individuals must also contribute to the pools covering the high-cost, high-risk individuals, even though that isn’t affordable either. Rather than compounding the nightmare of administrative inefficiencies, it would be much simpler to combine everyone into a single risk pool that is equitably funded and therefore affordable for everyone.
Single payer would do it. But in his efforts to make the antiquated private plans work, Dr. Gruber states, “… we can’t insist everyone who has no insurance get the policy that optimizes their health.” Further, “Let’s get them into the system and get them real insurance, and then maybe they’ll be interested in buying something better.” (The Boston Globe, March 5)
Why don’t we start out with something better? Real insurance. Single payer.
Additional Comment:
By Leonard Rodberg, PhD, Chair, Department of Urban Studies, Queens College/CUNY, responds to the comments on the RAND HIE made by Jonathan Gruber, Professor of Economics, MIT:
Prof. Gruber, like many other American economists, should remove his chauvinist blinders and take a look at the rest of the world where there has been a real world “experiment” going on for years, one that is far more extensive than the very limited Rand HIE. Countries around the world offer what we would call “first dollar coverage”, that is, little or no financial barrier to seeing a doctor. The results: (i) no overuse of medical services and (ii) lower spending than the US. The conclusion might be that eliminating cost-sharing would actually save us money!
For Dr. Gruber’s original statement and our responses:
http://blogs.wbur.org/commonhealth/?p=23#comments