Note: In part 1 of this post, we offered a more precise definition of rationing. In part 2 below, we describe the type of health plan we would like for ourselves and our families.

In addition to misusing the term “rationing,” another important impediment to productive discussions of the prudent use of health care services is the perspective taken by the discussants.

To assert that “we” are going to have to ration health care services, without specifying exactly who “we” are or who will be subject to rationing, leaves an opening for opponents to fill in with worst case scenarios, regardless of their accuracy. Prudence, as a virtue, involves the application of skill and good judgment in use of resources toward a good end.

The 2009 (and ongoing) controversies over mammography for women under age 50 illustrates the wrong way to talk to the public about the prudent use of health care services. Virtually every editorial on the subject conflated two very different issues: (1) the risks and benefits of mammography to individual women; and (2) the “societal” cost of saving an additional life. The latter discussion often implied that it might be bad for “society” for women under 50 to get mammograms because the money could be spent more productively on something else.

We suggest that “societal” rhetoric has the potential to do enormous harm. When researchers say that an expensive screening saves “only five lives per million dollars spent,” what the public hears is, “My life isn’t worth saving.”

Our simple recommendation is to stop discussing rationing as an activity that “we” are going to do to “you,” and instead, discuss the prudent use of health care services in terms of the health plan that we would choose for ourselves and our families. The following discussion provides an example. Along the way, we provide a categorization of health care services that might be helpful in its own right.

The Information Problem

Any discussion of the health plan we would like to join must be prefaced by a brief comment on the deplorable state of knowledge about the effectiveness of medical and surgical treatments. A chart published recently by British Medical Journal-affiliated Clinical Evidence identifies “treatments that work and for which the benefits outweigh the harms, especially treatments that may be underused;” and 2) highlights “treatments that do not work or for which harms outweigh benefits”. For 3,000 selected treatments with randomized control trial data, 11 percent are identified as beneficial, 24 percent likely to be beneficial, 7 percent have considerable harm-benefit tradeoffs, and 5 percent are likely to be ineffective or harmful. A full 50 percent, however, were categorized as being of “unknown effectiveness”.

Services That Have No Probabilistic Risk

The purpose of insurance is to spread the cost of risky events among those who are subject to the risk. Using insurance to pay for riskless events or to force contributions by those who are not subject to the risk may be an effective way to redistribute income, but it is not insurance.

For example, there is no risk associated with the consumption of preventive care, cosmetic surgery, or contraception. Coverage of prostate cancer screening, for example, transfers income from women and children to men who are age-eligible for screening. Income also is transferred from men who choose not to get screened or who have trouble accessing screening services to men who do get screened. The latter transfer is likely to be regressive because consumers of preventive care are likely to be better-off individuals or families with greater access to information, transportation, and health care infrastructure, with the ability to take time off from work or have someone cover household demands.

Thus, we would prefer that our health plan limit its coverage to insurable events. That, of course, is now illegal under the PPACA legislation, which mandates full coverage of preventive care and contraception. Those mandates are a form of floor rationing.

If there are positive externalities associated with consumption of a preventive service such as inoculation against an infectious disease, then it may be efficient for the government to subsidize the price of that preventive service; however, a price subsidy for a specific service can be offered or not, independently of the insurance coverage decision.

We are, nonetheless, open to arguments that coverage of preventive care suffers from market failure of the following form. Suppose that some preventive care was cost saving, netting more than a dollar in reduced care costs for each dollar spent on prevention.

Cost-savings services, along with improved clinical outcomes, underlie much of the enthusiasm behind “value-based health insurance design.” Value-based insurance reduces the point-of-purchase cost-sharing for services that have greater clinical benefits for specific patients.

Suppose further that health plans are reluctant to cover even cost-effective preventive care because the enrollee may be in a different health plan when the benefits of the care are realized. Mandated universal coverage of preventive care might be a way to capture those spillover benefits for society as a whole.

Despite not wanting our health plan to cover preventive care, we would like our plan to negotiate discounted prices for preventive care from reputable providers for all, whether or not one is receiving an income transfer for reasons of justice to facilitate such coverage.

Services That Provide No Medical Benefits

We would prefer that our health plan not cover services that have been demonstrated to have no medical benefit. (Notice that we are not asking that the health plan covers only services with demonstrated positive medical benefits. As indicated by the chart referenced above, that surely would be the cheapest health plan ever offered.) By “no medical benefit” we mean that there are neither physiological nor psychological benefits to the service. A frequently-cited example of a service with no medical benefit is arthroscopic knee surgery for osteoarthritis, the subject of two randomized trials. CMS has suspended payment for the procedure, yet approximately one million such surgeries still take place every year at an estimated cost of $7,000 each.

Cohen, et al. add the example of surgery in 70-year-old men with a new diagnosis of prostate cancer, as compared with watchful waiting. Another example is a virtual colonoscopy performed on an asymptomatic forty-nine year old President of the United States. Of course, under our preferred health plan, individuals would remain free to consume medically useless care, but would pay the total cost out of their own pockets.

A diagnostic test that confirms the presence of a disease for which there is no treatment may confer no physiological benefit to the patient, but the confirmation of the diagnosis could provide the patient some psychological reassurance, e.g., the ability to order one’s life according to priorities derived from the new information. We do not object to that type of coverage, but some cost sharing might be in order, as discussed below.

Services For Which The Expected Harms Exceed The Expected Benefits

We would prefer that our health plan not cover services for which the expected harms exceed the expected benefits to the patient. An example of risks that exceed benefits is lung-volume reduction surgery, the subject of a randomized trial required by CMS. The procedure did have medical benefits and CMS agreed to cover the cost, but it also carried a ten percent mortality risk. The New York Times reported that “many patients and doctors … seemed to lose their enthusiasm” for the procedure.

Another example is electron beam computed tomographic (CT) scan for coronarycalcium, also performed on an asymptomatic forty-nine year old President of the United States. Many medical procedures are dangerous, but we refer here to those that are “inefficiently dangerous” — e.g., the risks to the patient exceed the expected benefits.

If the expected benefits and harm are unknown, we would prefer partial coverage that reflects the best available evidence on likely benefit and likely harm, in addition to the usual considerations of moral hazard and risk protection, discussed below. We also would like our health plan to participate in post-marketing surveillance programs to improve the knowledge base for treatment effectiveness and efficacy.

Services For Which Expected Benefits Exceed Expected Harms

Hopefully, this category includes the bulk of insured medical care. There are two important sub-categories: services for which a lower cost alternative treatment exists that produces the same expected health outcome, and services for which no such alternative exists. In the first case, we would like for our health plan to require step therapy, in which the least costly treatments are tried first before moving to more expensive treatments. Step therapy coverage is fairly common in pharmaceuticals, but we would like to see it expanded to a broader range of medical treatments. If patients want to skip a step with no medical justification, then they should pay the extra (differential) cost of the more expensive treatment out of their own pocket.

A broad movement towards step therapy coverage would not stifle the development of new treatments, but it certainly would direct development towards treatments that provide the same (or better) expected health benefits at lower cost. Step therapy might be encouraged by movement to care systems coordinated by a primary care provider. We are indifferent regarding the organizational mechanism by which step therapy is implemented.

We also would like our health plan to require that health services be provided by the least expensive personnel capable of providing the service — another form of step therapy.

Amount Of Coverage

When no alternative therapy exists, then the question becomes, “How much coverage should the health plan offer?” Questions regarding the proper level of coverage for specific services bring us to the topic of moral hazard — the additional consumption of health care services that results from insurance coverage.

John Nyman distinguishes efficient moral hazard from inefficient moral hazard in the following way. Suppose that you have liver disease and you have no health insurance. Presumably you would spend some money on treatment, but that amount might be limited by the fact that you don’t have the cash required for full treatment (which might include a liver transplant), and you might have difficulty obtaining a loan to pay for the care.

Now, suppose that your health plan covers the full cost of treatment, so you and your doctor choose a course of treatment accompanied by various ancillary services. Presumably, given the possibility of benefit and no personal monetary cost, you would consume more services than if you were uninsured.

But suppose instead that your health plan wrote you a check for the average cost of treating liver disease under full insurance and you decided to spend some of the check on treatment and some of the check on a vacation or a new automobile. Nyman refers to the amount you spend on health care as efficient moral hazard. The money spent on the vacation or car is the inefficient portion of moral hazard. It is not inefficient in the broadest sense, because vacation and cars provide some additional utility, but it is inefficient relative to health coverage. Nyman also notes that in some cases, insurance provides a financing mechanism for health care services that otherwise would not have been available.

We would like for our health plan to cover the efficient portion of moral hazard. Obviously, determining the efficient level of moral hazard for various health conditions constitutes an important and ongoing research agenda.

We also would favor coverage that reflected the likely benefits of the service. The categories “likely to be beneficial” and “unlikely to be beneficial” in the Clinical Evidence review referenced earlier would be a good starting point.

Finally, we would like our health plan to provide its members with information on provider quality and total cost of care, and to contribute to community efforts to collect and disseminate those same data. We also would like our plan to experiment with payment reforms such as bundled payments that could improve provider incentives and transparent pricing for consumers.


As seen in Part I, much of the current discussion of rationing is an impediment to a badly needed discussion of the prudent use of health care services. Terms such as “societal good” are too vague to be helpful. They not only invite antagonistic interpretations, but as the public is exposed to more pronouncements such as, “the U.S. health care system wastes $750 billion a year,” they are likely to ask why restrictions on the allocation of medically effective care even are being discussed at all.

They might also wonder why we continue to pour hundreds of billions of subsidy dollars a year into consumption of services on which most people think we spend too much. Government subsidies of health insurance and health care consumption are not limited to efficient health plans and effective medical care. When inefficiency leads to higher costs, the government subsidizes inefficiency. A proposal to ration effective care while subsidizing inefficient consumption is like a proposal to ration the supply of water to firemen while throwing gasoline on the fire.

Instead of “rationing,” we should be discussing specific examples of market failure that lead to the inefficient consumption of health care services, including the unjustified imposition of floors and ceilings on consumption choices that otherwise would be available and preferred by consumers. We also should call economic injustice by its proper name, rather than calling it rationing, and address it through appropriate transfers. It also would be helpful to point out that floor rationing can improve consumer welfare in the presence of some types of market failure such as poor information or externalities.

When we describe to other people the health plan we would like to join, a frequent reaction is, “Isn’t that the type of health plan that everyone would like to join?” The answer is “No.” Clearly, there are patients who, when handed the randomized trials on arthroscopic knee surgery for osteoarthritis, will say, “But my neighbor got that surgery. Why can’t I have it?” Clearly, there are patients who will feel that they have been subjected either to rationing or gross injustice if their health plan doesn’t cover medically ineffective treatments.

This leads to our recommendation for progress on health care costs in the U.S. What we need are demonstrations. We need to assemble the right employers, the right health plans, the right providers, and perhaps most importantly, the right consumers into a health plan that offers rational coverage of services obtained from efficient and ethical providers and care systems. The demonstrations do not have to be large scale. The demonstration groups simply need to be left alone to determine what total cost differentials result.

If the total cost differentials are miniscule, that approach can be abandoned. If the cost differentials are substantial, with no deleterious health effects, then we expect that some conversion of reluctant or skeptical consumers would follow. However, we are not recommending rationing. If some consumers want to join health plans that allow them to roam aimlessly from one specialist and one treatment to another, regardless of the demonstrated benefits and harms, then they should have that option. All we ask is that they pay the marginal cost of that type of health plan out of their own pocket.

Another reaction to our preferred health plan is that it already is offered. The usual suggestions are staff model HMOs. The news is welcomed, but we remain skeptical because even those plans may be constrained by the preferences of enrollees who prefer their neighbor’s personal experience to the results from randomized trials.

Demonstrations are not the only pathway to reform. Simultaneous aggressive attention to market failure, including reduction of price-distorting subsidies, more aggressive enforcement of the antitrust laws, and cost-effective production of better information are important as well. Economic injustice remains a serious problem due, in part, to the design of our current programs. But in the long run, transformational health care reform in the U.S. is more likely to be the product of demonstrated success rather than legislative mandates. Health services researchers and policymakers can contribute more than research and policy proposals. They can lead by example.