This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

P E R S P E C T I V E S
V A R I A T I O N S
W E B  E X C L U S I V E   

7 October 2004  
  

Perspective: 
More Variation In Use Of Care, 
More Flat-Of-The-Curve Medicine

Why does it occur? What should be done about it?

By Victor R. Fuchs

ABSTRACT:

Variation in use of health care is ubiquitous in the United States. It is attributable to exogenous differences in supply of medical resources; to identified and unidentified economic, social, and cultural factors; and to the idiosyncratic beliefs of physicians. It is perpetuated by the parochial character of much clinical practice. Patients in high-intensity areas do not appear to have better health outcomes: Much care is “flat of the curve.” A more robust scientific foundation for clinical decisions could help to reduce variations, but major reform of health care financing is probably necessary to achieve substantial improvement in the organization and delivery of care.

The three contributions that provide the starting point for my commentary investigate variation in intensity of services provided to Medicare beneficiaries.¹ The authors of these three papers explore the consequences of variation, try to explain it, and offer recommendations for its reduction. The papers differ in the units across which variation is measured: academic medical centers (AMCs), hospital referral regions (HRRs), or “well-known” hospitals. They also differ in the study populations: patients with serious chronic illnesses, patients undergoing major orthopedic surgery, and patients in the last six months of life.

Data sources. The primary sources of data are individual records of Medicare beneficiaries. The sample sizes are large, and the records provide much information concerning patients’ characteristics, use of medical care, and health outcomes. I am hesitant to criticize the data because, for many questions, they are the “only game in town.” One caveat, however, deserves mention. The Medicare data cover only a portion of the services provided to patients. Variation in the use of services not covered by Medicare may not parallel variation in covered services. Sometimes the differences can be dramatic. For example, Jonathan Skinner and Weiping Zhou found that per capita Medicare spending in 1996–98 for seniors in the lowest income decile was 30 percent higher than for those in the highest decile.² Alex Chen and José Escarce, using the more comprehensive measure of services in the Medical Expenditure Panel Survey (MEPS) for the same period, estimated that wealthy seniors accounted for 83 percent more spending per capita than poor seniors.³

Certainty of variation. The least controversial result of the three studies is that there is much variation in the amount of care delivered to what appear to be comparable patients. This result has been reported many times in the past, by John Wennberg and colleagues and by others. There should be no residual doubt in the health policy community that the variations are real phenomena; they are not the result of comparing apples with oranges. Moreover, patterns of variation persist over time. If A shows more intensity of care than B for a particular patient population in a given year, there is a good chance that A will also show more intensity one year or even ten years later.

Role of intensity. Slightly more controversial is the conclusion that patients in high-intensity areas do not experience better health outcomes than those in low-intensity areas. Some readers will question this result; I find it quite credible. It accords with my reading of aggregate data in the United States and other high-income countries during the past forty years: Differences in intensity of care play, at most, a minor role in explaining cross-section differences in health outcomes.

The reasons are reasonably clear. First, truly effective interventions tend to diffuse widely and rapidly—for example, antibiotics for combating infectious diseases and diuretics for reducing hypertension. Second, where supplies of physicians, specialty care, or hospitals are tight, some informed rationing occurs. Third, and most important, cross-section differences in health outcomes are primarily determined by nonmedical factors: genetics, the physical and psychosocial environments, socioeconomic factors, and personal behavior.

The bottom line is that a considerable amount of the care delivered in the United States is “flat-of-the-curve” medicine. This term, popularized by Alain Enthoven, describes a level of intensity of care that provides no incremental health benefit.⁴ This conclusion is not refuted by two propositions that are popular in current health policy discussions: (1) the total value of health care is greater than its total cost, and (2) the value of the increase in health that Americans have experienced since 1950 is greater than the increase in health care spending.

Exhibit 1 shows the relationship between health and intensity of care at two different times, t and n years later ( t + n ). In both periods, the level of intensity is at the flat of the curve, points A and B. Nevertheless, the value of health at A (H_A) could well be greater than the cost (C_A). Similarly, the value of the increase in health from time t to t + n, measured by H_B–H_A, could well be greater than the increase in cost, measured by C_B–C_A. But it is the flat-of-the-curve conclusion that is particularly relevant to health policy. At any given time, policy usually involves choosing between more care or less; good decisions require comparing incremental benefit and incremental cost.

One caveat to this conclusion is its reliance on survival rates, life expectancy, or some other measure of mortality. Measures of quality of life could show some incremental benefit in high-intensity areas. Reliable measures of improvement in quality of life, however, are difficult to obtain. They are not nearly as objective as mortality rates and are subject to bias. For example, surgical patients who have experienced a great deal of trauma and cost are often reluctant to acknowledge to themselves or to researchers that it was all in vain.

Quality of life. An important consideration in policy analyses of quality-of-life claims is the source of payment for more intensive care. If payment is borne collectively, through private or public coverage, the standard of evidence should be much higher than if consumers are paying with their own funds. Each day Americans spend billions of dollars for goods and services that they think will add to their quality of life, even though experts might think otherwise—for example, they flock to movies that professional reviewers have judged as terrible; they consult astrologers, psychics, and faith healers; they buy brand-name over-the-counter drugs when generic versions are available for half the price; and the list goes on. The free expression of consumers’ preferences, backed by their own pocketbooks, is not usually considered a major social problem. But collective payment for medical care of questionable value is. When standard outcome measures say that the curve is flat, it is incumbent on those delivering care at that level to justify collective payment for it.

The strongest rationale for equal access to care is that everyone ought to have an equal chance to live, regardless of economic circumstances. But when quality of life is the object of high-intensity care, the egalitarian imperative for collectively funding such care loses much of its force. Quality of life for the poor could be improved in many other areas; disproportionate allocation of resources to medical care does a disservice to the poor and to society as a whole.

Reasons for variation and flat-of-the-curve medicine. Why is there so much variation in use and so much flat-of-the-curve medicine in the United States? According to the three studies, the principal explanation is that excess supply induces additional demand. In my research, I found evidence of supply-induced demand.⁵ I am not, therefore, inclined to reject this hypothesis in principle, but the Dartmouth studies do not provide rigorous tests of it. It is not enough to show a correlation between supply and utilization. One must explain why supply differs and establish that the differences are independent of any differences in demand attributable to economic, social, and cultural factors. My surmise is that supply does matter, but not to the extent suggested by the authors of these studies.

The authors also imply that regardless of origin, variations in medical practice tend to persist because of the parochial approach of much clinical practice. I agree. Consider the following hypothetical (but not unrealistic) exchange between a new resident and a chief of surgery. New Resident: “How long should a patient stay in the hospital after operation X?” Chief of Surgery: “In our hospital we keep the patient for N days.” There would probably be no discussion of the fact that in other hospitals, the standard might be N+1 or N–1 days. Also, there would probably be no appraisal of the scientific basis (or lack thereof) for the answer, and there would be no consideration of the benefits and costs of N+1 or N–1.

The story about length-of-stay could be repeated regarding indications for surgery, use of diagnostic tests, and most clinical decisions. Conformity with local practice is reinforced by a desire to follow the example of physicians who command respect or authority. Conformity also flows from patients’ expectations, which are partially formed by the treatment received by neighbors and friends. Deviations from local standards are always suspect, must be justified, and are more vulnerable to ex post criticism if things turn out badly (even if the reasons for the bad outcome lie elsewhere).

Policy considerations. The two policy recommendations that emerge from these studies are not unreasonable, but to me they fall far short. First, the authors suggest more patient involvement in clinical decision making. If patients want to be involved, who could object? But some might prefer to let the physician decide. Should they be denied that option? Also, the authors seem to assume that patient involvement will result in less care in high-intensity situations. But it may also result in more care when intensity is low. The overall effect on spending is not obvious. The other major recommendation is for more research to provide a firmer scientific foundation for clinical decisions. Amen! But no one should imagine that this problem is easily remedied.

The technical difficulties involved in providing a more robust scientific basis for the myriad clinical decisions that determine the cost of medical care are formidable. Given the rapid rate of introduction of new medical technologies and the consequent rapid growth of health spending, there is urgent need for a large, private, nonprofit institution to assess the cost-effectiveness of various interventions.⁶ It must have a large, steady source of funding and be as free from political and professional pressures as possible. Its primary function would be to help develop and disseminate systematic knowledge about the cost-effectiveness of medical technologies, including drugs, diagnostic procedures, and surgical interventions. Its second important function would be to provide legitimacy for the cost-effective practice of medicine. Many directors of health plans and many physicians know that they could be practicing in a more cost-effective way, but they are inhibited from doing so by fear of malpractice suits and by pressure from patients and peers.

With regard to policy, the most pressing issue to address is the role of incentives. Financial arrangements surely influence the choices made by patients, physicians, hospital administrators, and others. Consider the role that insurance plays in underwriting flat-of-the-curve medicine. A fully insured patient wants care up to the point at which there is no incremental benefit. A physician concerned only with the welfare of such a patient would recommend care up to that point. Or consider how reimbursement affects clinical decision making: much more than is generally appreciated because once standards of care are established in a setting, the individual physician honestly believes that he or she is delivering appropriate care. But reimbursement has a big influence on the standards.

These three studies and others show the need for improvements in the organization and delivery of care. To achieve those improvements, however, I am increasingly convinced that the United States must undertake major reform of health care financing.

Financial support from the Robert Wood Johnson Foundation and comments from Ezekiel Emanuel, Alain Enthoven, and Alan Garber are gratefully acknowledged.

NOTES

1. E.S. Fisher et al., “Variations in the Longitudinal Efficiency of Academic Medical Centers”; J.E. Wennberg et al., “Use of Medicare Claims Data to Monitor Provider-Specific Performance among Patients with Severe Chronic Illness”; and J.M. Weinstein et al., “Trends and Geographic Variations in Major Surgergy for Degenerative Diseases of the Hip, Knee, and Spine,” all available at Health Affairs, 7 October 2004, content.healthaffairs.org/cgi/content/full/hlthaff.var.104/DC1.
2. J. Skinner and W. Zhou, “The Measurement and Evolution of Health Inequality: Evidence from the U.S. Medicare Population” (Paper presented at a symposium in honor of Eugene Smolensky, Berkeley, California, 12–13 December 2003).
3. A.Y. Chen and J.J. Escarce, “Quantifying Income-Related Inequality in Health Care Delivery in the United States,” Medical Care 42, no. 1 (2004): 38–47.
4. A.C. Enthoven, “Shattuck Lecture—Cutting Cost without Cutting the Quality of Care,” New England Journal of Medicine 298, no. 22 (1978): 1229–1238; and A.C. Enthoven, Health Plan: The Only Practical Solution to the Soaring Cost of Medical Care (Reading, Mass., and Menlo Park, Calif.: Addison Wesley, 1980).
5. V.R. Fuchs and M.J. Kramer, “Determinants of Expenditures for Physicians’ Services in the United States 1948–68,” NBER Occasional Paper no. 117 (New York: National Bureau of Economic Research, 1973); and V.R. Fuchs, “The Supply of Surgeons and the Demand for Operations,” Journal of Human Resources 13 Supp. (1978): 35–56.
6. V.R. Fuchs, “Health System Reform: A Different Approach,” Journal of the American Medical Association 272, no. 7 (1994): 560–563; and V.R. Fuchs and A.M. Garber, “Health and Medical Care,” in Agenda for the Nation, ed. H.J. Aaron, J.M. Lindsay, and P.S. Nivola (Washington: Brookings Institution Press, 2003), 145–181.

Victor R. Fuchs (fuchs{at}newage3.stanford.edu) is the Henry J. Kaiser Jr. Professor Emeritus, Stanford University, and a research associate at the National Bureau of Economic Research in Stanford, California.

DOI: 10.1377/hlthaff.var.104
©2004 Project HOPE–The People-to-People Health Foundation, Inc.