It is much easier to achieve consensus about the claim that there is too much waste in the U.S. health care system than it is to find someone who is willing to take responsibility for this state of affairs. Finger-pointing is inevitable in a health care system with so many players whose interests are diverse and often conflicting. For people fed up with the health care system, and daunted by its complexity, simple solutions — like those that Robin Hanson proposes– are tempting. The principles that should guide health care reform in the United States are simple and universal, such as preserving choice, getting the incentives right, and ensuring markets work the way they should. But successful implementation will not be simple. It will require better information, rational incentives, and public education.
Hanson may well be guilty of overstating his indictment of health care in the United States. As he recognizes, we should not place too much credence in the finding that health outcomes at the national level bear little relationship to national health expenditures. Even better-designed studies, such as analyses of individual observational data, are themselves subject to methodological flaws that call their findings into question. Very few convincingly correct for the effects of unmeasured but important aspects of health. But some of the best studies, such as those of Elliott Fisher and his Dartmouth colleagues, show that regions in which Medicare patients are treated more aggressively have higher expenditures and no better outcomes. These findings do not establish that more aggressive styles of medical care provide no benefit. They lack sufficient detail about the underlying health of patients in different parts of the country to support such a sweeping claim. Nevertheless, if there is anyone who is still complacent about the state of the U.S. health care system, he or she must engage in some heavy rationalization. How to explain, for example, why it is so difficult to find any evidence that more spending generates better outcomes across regions of the United States? And why do disease-specific analyses comparing citizens of different countries fail to show that their greater medical expenditures routinely buy Americans better health outcomes?
Hanson’s diagnosis, therefore, is not particularly controversial. His solution is. Like any attempt to trim excessive spending, it must confront one paramount fact: the benefits of medical care are highly variable. The antibiotic ampicillin is life-saving for meningitis and of dubious value in treating a child’s bacterial ear infection. It only offers the benefits of a placebo, along with adverse reactions like diarrhea, when used to treat the common cold. Ampicillin is frequently used for each of these conditions, and the benefit of ampicillin when used to treat the common cold can be markedly different from the average benefit, which includes its use in far more serious conditions. Of course, variability in benefits is not limited to inexpensive treatments like ampicillin.
Take biological treatments for cancer, which cost tens to hundreds of thousands of dollars per course of therapy. Tarceva increases median survival by about two months in previously untreated non-small-cell lung cancer but only by about 10 days in pancreatic cancer patients. The doses of Avastin used to treat lung and breast cancer are much higher than the dose in colorectal cancer, in which it is more effective. Preventive interventions, which are used in predominantly healthy populations, are expensive in aggregate because they are so widely used. Their benefits are greatest when used by people at high risk of the disease that they are designed to prevent; heart attack survivors can expect large survival benefits from cholesterol reduction, while cholesterol reduction in young women who have high cholesterol levels but no other risk factors for heart disease confers benefits so small that they have never been measured directly. Heterogeneous benefits are the rule, not the exception, in medical care, and they pose both a challenge and opportunity for reducing inefficiency: it is hard to develop policies to reduce selectively the care that is of low marginal benefit, but if we succeed in doing so, we should be able to preserve health outcomes while reducing expenditures.
Hanson’s statement that “most any way to implement [a 50%] cut would likely give big gains” is an invitation not only to ignore the variability in benefits but to encourage policies that would heedlessly cut high-value benefits along with the low-value marginal benefits. Average benefits can be very large, as David Cutler’s work suggests. Yet as Hanson argues, aggressive cuts may well be possible and desirable. How can we selectively cut care that is of low marginal value?
The first step is better information. Many economists and policy analysts downplay the importance of better information, even though economists are almost all familiar with Kenneth Arrow’s explanation that the unique characteristics of markets for medical care – and the reasons that market failure is inherent in this sector – stem from informational failure. The information needed to make decisions about eliminating low-value care is relatively straightforward: we need more information about what works and, in particular, about the value of specific interventions. It’s not difficult to identify interventions at the extremes of value. Think of basic care for high blood pressure as very high value, and perhaps the use of left ventricular assist devices for heart failure as an expensive, very low value intervention. But choosing where to cut means making decisions in the vast gray area between these extremes, and this requires comprehensive information that is simply unavailable today. My co-authors and I, among others, have argued that this information would be best provided by a well-funded federal program that would ensure that the necessary studies are carried out (Emanuel, E., V. R. Fuchs, A.M. Garber (2007). “Essential elements of a technology and outcomes assessment initiative.” JAMA 298(11)). One of its important functions would be to make the information available in accessible form to the general public, so that patients would have the information they need to evaluate management options themselves.
Information can be a powerful tool but it will have little impact if the incentives to use it are inadequate. Hanson identifies government and corporate subsidies to medical care as the problem, but these are incentives for the overconsumption of medical care generally, not for the overuse of low-value care. The changes needed to promote high-value care are inevitably more complex, and we are still learning which approaches will work best. Health plans are gaining experience with health insurance benefit design, such as benefit-based copayments and the selective use of disease management programs, to promote high-value care. They could also improve incentives by adjusting reimbursement rates in fee-for-service settings, which often promote too much of the wrong kinds of care. Medicare changed its rules for reimbursing chemotherapy after recognizing that many oncologists made most of their money by dispensing chemotherapy, simply because their margins on the drugs were so large (and the reimbursement rates for their services were so low). I do not mean to downplay the complexity of correcting reimbursement rates, or to dismiss the outcry that will occur if incomes of some specialists drop as a result. But we should find better ways to compensate physicians than by paying them to perform unnecessary procedures.
Finally, the full costs of an insured individual’s medical care have been so effectively hidden for so long that many Americans have been lulled into a sense that medical care is, or should be, nearly free. Employers have adopted high deductible health plans as a tool to control expenditures on health insurance for their employees, but equally important for them has been a belief that these plans make their employees more sensitive to the costs of care. Certainly employees will be much more reluctant to spend funds drawn from their own health savings account than from an insurance plan. They are much more likely to make prudent decisions if they take costs into account, and if they have good information to guide their choices.
This is true for people who are enrolled in any kind of health insurance plan. As David Cutler and Dana Goldman have noted, observational studies appear to show that patients cut back indiscriminately on the drugs they use when copayments rise. For the most part, these findings are inconsistent with the results of the Rand Health Insurance Experiment, a study with a superlative design that was carried out in a different era, long before pharmacy benefits managers and pervasive tiered copayments. I don’t believe that increased cost-sharing inevitably leads to poor decisions about health care. But too often health plans have simply increased copayments and deductibles without making it easy for patients to decide where to spend their own money. If we want them to succeed, let’s give them the tools they need.
Alan Garber is the Henry J. Kaiser Jr. Professor and professor of medicine at Stanford University, and is a staff physician at the VA Palo Alto Health Care System.