The RAND Experiment: Still Apologizing After All These Years, Part I

As health policy studies go, the RAND Health Insurance Experiment is the gold standard. Conducted 25 years ago at a cost of $50 million (close to $300 million in today's medical marketplace), this study sorted families into health plans with different deductibles as well as an HMO, and carefully monitored the results.

The findings: (a) patients are responsive to out-of-pocket costs (the more they have to pay, the less health care they buy); (b) changes in the amount of spending have no apparent impact on health care outcomes in most cases; and (c) judging from the difference in behavior between HMO doctors and fee-for-service doctors, physicians are also very responsive to economic incentives.

So what does this mean? That's not clear. My view is very different from the economists who did the study. Since I am chairing this session, I'll go first.

My Take: In showing that both patients and doctors respond to economic incentives, the RAND study completely debunked the idea that health systems can be understood without reference to economic models. Two very important policy implications follow. First, the study put the final nail in the coffin (at least for several decades) of naive national health insurance (NHI), a la Michael Moore. If health care were completely free, spending would soar, with no improvement in health status. (NHI, by the way, was a very popular idea at the time and a principal motivation behind the entire experiment.) Second, the study opened the door to a myriad of market-based solutions to health policy problems. We could have full-service diabetes centers vigorously competing for diabetic patients, each managing his own risk-adjusted Health Savings Account. In fact, we could revolutionize the chronic care industry, creating specialized markets for sick people. These possibilities, which are almost endless, would not be possible, however, if patients and doctors did not respond to economic incentives.

The RAND take: None of the thoughts in the preceding paragraph appear in the "selective memories" of the original researchers, recently published by RAND. "Selective" is the right word. They do not even mention national health insurance (despite the rising clamor for it on the left). Nor do they mention the technocratic view of health care they so totally eviscerated (even though this is still the dominant view among the laity). As for cost sharing, their prose is full of caution–pointing to dangers for the poor and the sick and to more recent studies showing what can go wrong. There is no mention of the creative ways in which cost sharing is being used in chronic care — either in private plans or in Medicaid.

Clearly we need a better division of labor. If the experiment is ever repeated, let Joe Newhouse and his colleagues conduct the study. Let Regi Herzlinger and me explain the importance of the results.

More on this in a future Alert.

For some very, very "Selective Memories" by RAND researchers, Emmett B. Keeler, Joseph P. Newhouse and Robert H. Brook, go to:
http://www.rand.org/publications/randreview/issues/summer2007/health.html

12 thoughts on “The RAND Experiment: Still Apologizing After All These Years, Part I”

  1. The issue is not whether physicians and others respond to economic incentives. The question is whether they SHOULD respond to economic incentives. US medical practioners respond very well to economics. In an environment where practioners control both demand and supply, they act the way one would expect…skyrocketing prices.

    Should a health care system be subject to supply and demand forces? Why should ability to pay define access to basic medical care? In a private market based system, participants seek to gain an advantage and spend accordingly to gain an advantage. Add up all this admin spending and it comes to 30 cents on the dollar? That amounts to a lot of dollars that should be focused on medical care delivery not advertising budgets, paperwork, etc.

  2. The missing ingredients that need to be added to make more politically palatable a reformed private healthcare system are direct subsidies, based on a sliding scale tied to income, that would help the poor to pay for high deductible real insurance premiums (instead of the prepaid managed care model that most so-called health insurance follows) and subsidized contributions to HSAs for them. As a 60 year old consumer with high deductible coverage and HSA, I have seen firsthand how dramatically incentives shift and how much more responsive the healthcare system can be when gatekeepers and insurance company benefit managers are removed and my care is managed by me and my doctor. It's "managed care" that frustrates everyone, and there'll be a LOT more of it if we go to single payer system. We need to design our model to work efficiently for the 85% who already have resources dedicated to the funding of their healthcare, then bring in the remainder with subsidies. After capturing market efficiencies, our costs should go down and quality up. If our auto insurance worked like our health "insurance", whenever you pulled into Exxon, you wouldn't care what the pump price was because you'd be responsible for a fixed copay regardless, but the pump jock would have to verify your coverage and find out how many gallons you could have. And you'd file scores of claims every year. How efficient is that?

  3. Thanks for your continuing reports. Actually, there seems to be a flourishing industry in cancer treatment with specialised centers competing for patients already. Cancer Treatment Centers of America was the first to incorporate nutrition and homeopathy/holistic medicine as consumer alternatives. Specialized treatment centers as you suggest could compete and offer more efficient and less expensive treatments creating a consumer driven marketplace.

    More and more people are going to Costa Rica for cosmetic and dental surgery. The doctors there compete with each other and are willing to negotiate prices. Prices are about 60% lower and the surgeons are excellent. Many of the dentists and surgeons studied at some of the most prestigious medical schools in the United States. The hospitals are modern and up to date. Who says that the U.S. has the best medical care in the world? Not in my books. I have a lot of air miles and I can get from Dallas to Costa Rica in about 3 hours.

  4. Efficiency versus fairness…market based systems are efficient but are they fair? If all market efficiencies were acceptable, then Adam Smith would rule the day…but he does not. Why?…because in some cases, markets are not fair…and health care is one of those markets.

    Why should a person’s survival that has nothing to do with his behaviour or actions be based on his ability to pay? Is that fair?

    Consuming gasoline should be based on market principles, health care should not. There are different realities for health care and the private system needs modification to account for them. One approach is subsidies, another is single payer, and there are many variations. But to blindly believe the private system is the preferred approach is narrow minded and for many Americans, dangerous. The private system works if one is wealthy, scary if not.

  5. I’d be interested in your assessment of the recently published Commonwealth Fund study ranking the US very low compared to other developed nations in terms of preventing deaths. From what I can tell by reading information on its website, the Commonwealth is another left wing foundation whose president, Karen Davis, writes as though she were composing talking points for the Clinton campaign.

  6. If you’re writing up more about the selective memories, I wish you would take aim at the claim that the poor were made worse off by cost sharing. Newhouse, in Free For All?, discusses the fact that the gains to the poor mostly came from the fact that undiagnosed cases of hypertension were picked up in the initial free physical exam provided under the free plan. Subsequent use of medication was unaffected by cost share. He commented that free insurance is a costly way to get free exams and that more targeted measures would likely be more efficient. This cuts no ice with the RAND detractors, who still run around saying that it showed that the poor are harmed if they don’t get free care. And they always lump dental in, which most people don’t consider part of health insurance, at least until the concerted recent effort to make it so.

    Some specific quotes from Newhouse in the book Free For All? 1996 Harvard Press paperback edition, pages 351-352, are

    “For dental and vision care, the benefits observed on the free plan could almost certainly be achieved more cheaply by a targeted benefit…the issue is whether the modest gains we found under free care justify the cost of a targeted benefit.”

    “Virtually all of the improvement in blood pressure control brought about by free care occurred as a result of better identification of hypertensives…Control, once the person was diagnosed, was not measurably affected by cost sharing.”

    He goes on to say that the experiment shed some light on the benefits of screening for hypertension, pointing out that the improvement from simply screening for hypertensives created an “improvement that was more than half as large as the improvement caused by free care. These results indicate that a screening examination might be an attractive alternative to free care.”

    As for the “we were so surprised that utilization fell across the board and not in any one area which means that people cut health care willy nilly and therefore don’t know what they are doing” argument, I’ve always thought that this was a lame explanation. It is much more likely that health care is so individualized that people can cut in ways researchers both can’t see and don’t expect. The results from the Medicaid Cash & Counseling experiments support this kind of reasoning. So do the RAND ER usage figures (pages 154-158 of Free for All?) and a 2007 study of emergency department by Wharam et al. in the March 14, 2007 issue of JAMA. People apparently do much more effective triage when their money is on the line. Similar results seem to be showing up in the preliminary results from commercial consumer-directed account based plans. In one case, specialist use went up but costs went down. In other cases, compliance in taking medications increased.

    If these results hold up, researchers may need to reconsider some of the long held research templates. It may be, for example, that specialist care for the chronically ill reduces costs and that people who pay the cost use specialists appropriately. This has profound implications for policy specialists who unceasingly chant the “we need more primary care” mantra.

  7. I have been preaching your gospel for 40 plus years. Someone (I forget who) once told me that free care was worth exactly what you paid for it. When neither the doctor nor the patient has no skin in the game, quality disappears and the personality of what was once a wonderful profession morphs into an impersonal business.

  8. What I have found interesting from an experience working in a staff model HMO setting is that costs, as well as quality of patient care — as measured by the Healthcare Effectiveness Data and Information Set (HEDIS) — can be positively impacted through proper economic incentives applied to providers.

    The problem with this model is that it is so unpopular with most of the public who have gotten used to the co-pay mentality and do not react well to the “insurance company control” that they say they experience in an HMO setting.

    The good news is that consumer-directed plans seem to be showing that consumers who have a financial stake in their health care will make rational decisions. Now all we need to make it happen on a wide-spread basis are more provider models like the full-service diabetes centers envisioned in the blog piece that will vigorously compete for patients.

  9. Good comments…
    Sounds like it’s time for primary care physicians to work for their patients and not the insurers and the government…

Comments are closed.