take time for researchers to reach agreement in cases where practice variation is the result of real scientific uncertainty. The full potential savings from this strategy might therefore be realized only over the long

term.

Assuming that future research can resolve disagreements over appropriate treatments, there would remain the task of inducing physicians to modify their practices voluntarily on the basis of the new findings. Some success in changing practices has been reported when physicians have been introduced to guidelines through structured face-to-face educational programs conducted by respected peers. 15 Some other efforts that relied only on printed materials to communicate practice recommendations have had disappointing results. Providers could be aware of and even approve the recommendations without making significant changes in practice. It is possible that some physicians may encounter barriers in implementing even guidelines with which they nominally agree. These may include concerns about malpractice liability, lack of the substitute skills or the special equipment needed to follow the guidelines, economic incentives, or pressure from patients. 16 These barriers might be overcome with more vigorous educational efforts. Still, countervailing economic and professional pressures may limit the willingness or ability of physicians to comply voluntarily with treatment guidelines.

One alternative is to use the results of outcomes research as the basis for mandatory, rather than voluntary, guidelines—that is, as a way of strengthening or broadening current utilization control programs. Proposals to do so have met strong opposition from the medical community, on the grounds that medicine cannot be reduced to a "cookbook" and that to compel physicians to comply with fixed practice rules would stifle innovation. In addition, there would remain the problem of achieving sufficient savings to offset the administrative costs of review systems.

Another option is to replace service-by-service utilization review with general comparisons of each physician's practice patterns to those of his or her peers. Physicians who, over time, consistently furnished or ordered more of certain services than others in the peer group would be targeted for closer scrutiny, to determine whether patterns of inappropriate utilization existed. Physicians found to be outliers might be the focus of special educational efforts in the hopes of

15 See Mark R. Chassin, "Standards of Care in Medicine," Inquiry 25 (4) (Winter 1988): 437-453.

16 Jonathan Lomas, et al., "Do Practice Guidelines Guide Practice? The Effect of a Consensus Statement on the Practice of Physicians," New England Journal of Medicine 321 (19) (November 9, 1989): 1306-1311; and Jacqueline Kosecoff, et al., "Effects of the National Institutes of Health Consensus Development Program on Physician Practice," Journal of the American Medical Association 258 (19) (November 20, 1987): 2708-2713.

inducing voluntary change. Continued noncompliance might trigger requirements that individual services receive prior authorization or could even lead to exclusion from participation in a given public or private insurance program.

How much could be saved if all inappropriate services were eliminated? Some studies have found very high rates of unnecessary care. For example, Chassin et al., in a thirteen-site study, found that 17 percent of all coronary angiographies were unnecessary; for other procedures, the rate of inappropriate use was as high as 32 percent. They also found, however, that the unnecessary care explained only a small fraction of variations in utilization across geographic areas. If none of the inappropriate angiographies had been performed, the area with the highest use of this procedure would still have had more than twice the number of angiographies as the lowest-use area. The authors suggest that other factors must play a part in this difference: disease incidence, differences in the point at which primary care physicians decide to refer patients to specialists, or cultural or social differences in the stage at which patients sought care. 17 Another multi-site study has found that, while practice style may explain differences in utilization of certain specific procedures, it does not explain overall differences in per capita use of medical care in different areas. At the aggregate level, standard socioeconomic factors could explain much of the difference in use and intensity of services. 18

These preliminary studies suggest that there could be underutilization of services in some areas, while there is overutilization of the same services in other areas. Treatment research could pinpoint, not only cases in which unnecessary services could be eliminated, but also cases in which patients have had insufficient access (whether physical or financial) to necessary care. It is for this reason that some proponents of outcomes research have emphasized its potential impact on quality, rather than its potential for cost savings. Precisely because there is uncertainty about the relative efficacy of many treatments, it may be too early to say whether optimal medical treatment would involve more or fewer services than are currently furnished.

Supply Controls

If utilization controls or practice guidelines succeed in limiting unnecessary care, the full potential savings from any reduction in the number of services delivered may be realized only if there is a proportionate reduction in the resources used to provide those services. For example, changes in medical practice in the late 1970s and early 1980s led to a decline in inpatient hospital admissions without a corresponding reduction in hospital capacity. The result in many areas has been underutilized facilities spreading their fixed costs across a declining number of patients; while there are fewer patients, the cost for each patient rises because the unused capacity must still be paid for.

In addition, the existence of excess capacity may generate continuing pressures to find some new way of using that capacity and restoring utilization to its previous levels. 19 The view that the use of medical services could rise to fill any underused resources led to what was perhaps the dominant approach to cost containment in the 1970s: health planning, the regulation of facility construction and other capital expenditures.

In 1964, New York became the first State to establish a certificate-of-need (CON) program, under which proposals to build a new facility or expand an existing one had to be approved by a government agency. Other States followed, and a 1972 amendment to the Social Security Act provided that facilities in those States proceeding with construction without obtaining a CON could be denied Medicare and Medicaid reimbursement for their capital expenditures. Finally, the Health Planning and Resources Development Act in 1974 required all States to establish similar programs. This requirement was repealed in 1986, along with all Federal support for State health planning programs. States may continue to operate programs on their own; 39 States and the District of Columbia still do so. However, Medicare reimbursement is no longer contingent on State approval of capital expenditures, and a number of States have now limited their reviews to nursing home construction.20

Several factors contributed to the reversal of policy on health planning. In part, it fell victim to the gen

19 The view that hospital admissions rise in proportion to hospital bed capacity was originally advanced by Milton Roemer, in "Bed Supply and Hospital Utilization: A Natural Experiment," Hospitals 35 (21) (November 1, 1961): 36-42; Some more recent studies have concluded that the relation between supply and utilization may not be as straightforward as "Roemer's law" would suggest. W. Ross Brewer and Mary Anne Freedman, "Causes and Implications of Variation in Hospital Utilization," Journal of Public Health Policy 3 (4) (December 1982): 445–454.

20 American Hospital Association, State Issues Forum, State Health Planning Report (Chicago: July 1989).

eral preference for market as opposed to regulatory solutions during the early 1980s. From a Federal perspective, the adoption in 1983 of Medicare's prospective payment system (PPS) for inpatient hospital services was expected to offer a different way of limiting health care resources; this approach is discussed further in the next section.21 Underlying this shift, however, were claims that health planning had been tried and had failed, largely because of conflicting political pressures. In many areas, the oversupply of facilities was such that savings would have required, not just limits on new construction, but closure or consolidation of existing facilities. Few States were able to overcome the political resistance to such closures. Attempts to limit duplication of services or the spread of new technologies often faced similar barriers; attempts to plan for the rational distribution of resources on a regional basis had to confront providers' fears of losing to competitors and individual communities' desires for the most up-to-date facilities. 22

CON programs did have some successes, particularly in constraining the growth in nursing home beds. Because State Medicaid programs are the major source of payment for nursing home care, States had a strong motive to overcome the political barriers to supply constraint. In at least some States, the CON process was explicitly seen as a Medicaid cost-containment measure; the determination of the number of nursing home beds needed was related to the maximum number of patients the State was prepared to cover. 23 Even in this case, however, any savings were achieved by holding growth in bed supply below the rate of growth in the aged population. States generally did not close down existing capacity.

Recent concern about the rate of medical care cost increases has led to some calls for a revival of health planning, and it is conceivable that these concerns might eventually be sufficient to overcome the political barriers faced by health planners in the past. However, not all of the problems with health planning are political ones. Effective planning may require a fuller understanding of the workings of the health care system than is currently available. That system is a dynamic one, and decisions that seemed sensible in the late 1970s have sometimes had unpredictable effects. For example, most planning programs focused

21 The inclusion of capital expenditures in PPS payments has been repeatedly postponed. Hospitals are instead paid for Medicare capital expenses on a reasonable cost basis, subject to a fixed percentage discount (15 percent beginning January 1, 1990).

22 For an overview of the barriers to health planning, see Lawrence D. Brown, "Common Sense Meets Implementation: Certificate-of-Need Regulation in the States," Journal of Health Politics, Policy and Law 8 (3) (Fall 1983): 480-494. (Hereafter cited as Common Sense Meets Implementation.)

23 Judith Feder and William Scanlon, "Regulating the Bed Supply in Nursing Homes," Milbank Quarterly 58 (1) (1980): 54–88.

on institutional services in hospitals and nursing homes, because these were the major sources of expenditure, and did little to control the capital expenditures of community-based physicians or clinics. The resulting growth in the availability of high-technology facilities outside hospitals is one of the reasons that recent reductions in inpatient utilization have been offset by increased outpatient costs. (Some States are now applying uniform rules across settings.)

Moreover, a community's needs may change unpredictably. New York was more successful than most States in controlling inpatient bed supply; it was one of the few States in which hospital closures occurred on a planned basis. While the number of community hospital beds nationally dropped 1.1 percent between 1977 and 1987, the number in New York dropped 9.9 percent. 24 New demands on these facilities in the 1980s, such as the appearance of AIDS (acquired immune deficiency syndrome) and the rise in drug-related problems, have led to serious overcrowding in some New York hospitals. The reported crisis in New York illustrates one of the potential constraints on the planning process. On the one hand, it may be necessary to maintain enough excess capacity to meet unforeseen needs or random fluctuations in demand. On the other hand, this excess capacity is costly to maintain and may itself generate demand. If the supply of a given kind of service is sufficient that no one ever has to stand in line for it, then the savings from health planning may be limited.

The fullest potential savings from health planning would require a more controversial step: limiting the supply of health resources to the point at which patients may have to wait for some period to obtain needed but non-emergency services. The result is "queueing," the delays in surgery or high-cost diagnostic procedures that are alleged to occur to some extent in Canada and to a greater extent in the United Kingdom. The degree to which queueing actually occurs in either country's health system has often been debated by those who favor or oppose adoption of a similar system here. Some people say that essential care may be unavailable, while others argue that resource limits merely oblige providers to set priorities and avoid unnecessary services.

Whatever the extent to which resources have been limited elsewhere, rationing of supply in the United States might raise concerns that are not as significant in countries where the entire population participates in a single insurance program. In those countries, everyone is in the same queue, and one's place in line is chiefly determined by the urgency or duration of

24 American Hospital Association, Hospital Statistics, 1978 and 1988 editions.

one's need. (There are exceptions: one can step out of line in the United Kingdom by finding a private provider, and there are anecdotal accounts that some Canadians with sufficient resources may seek care in the United States.) When queueing has occurred in the United States, however, places in line may have been determined by financial resources.

The facilities in New York reporting the greatest overcrowding have been those serving the poor and the uninsured. Similar effects may have resulted from health planning's major success, the control of nursing home bed supply. Because Medicaid payment is generally less than that available from private patients, nursing homes in areas with limited bed supply and high occupancy rates have an incentive to accept a private-pay patient when a vacancy occurs, while Medicaid beneficiaries may be unable to find a place. In 28 States, Medicaid administrators report that beneficiaries awaiting hospital discharge had difficulty finding a nursing home bed. 25 While supply constraints are not the only factors limiting access to care for low-income Americans, they may exacerbate existing problems. The acceptability of health planning as a cost control strategy may, then, depend in part on the extent to which supply limitations are accompanied by efforts to make distribution of limited resources more equitatie.

One other issue should be raised in the context of a discussion of health resources: the debate over the possible oversupply of physicians and the potential consequences of physician supply on health care costs. In 1980, the Graduate Medical Education National Advisory Committee (GMENAC) reported that the United States would have a surplus of 150,000 physicians by the year 2000.26 The extent of the potential surplus has since been the subject of continuing debate. There are questions about the extent to which technology and the aging of the population could increase demand, or the adoption of utilization controls or managed care could decrease it. The number of medical school admissions could decline, or physicians might spend more of their time on administrative activities and less on patient care. 27

25 For a fuller discussion of this problem, see U.S. Library of Congress, Congressional Research Service, Medicaid Source Book: Background Data and Analysis. Report prepared for the House Committee on Energy and Commerce (Committee print 100-AA) (Washington, D.C.: November 1988), 467483. (Hereafter cited as Congressional Research Service, Medicaid Source Book.)

26 Graduate Medical Education National Advisory Committee, Report to the Secretary, U.S. Department of Health and Human Services (Washington, D.C.: 1980).

27 For contrasting views on these issues, see William B. Schwartz, Frank A. Sloan, and Daniel N. Mendelson, "Why There Will Be Little or No Physician Surplus between Now and the Year 2000," New England Journal of Medicine 318 (14) (April 7, 1988): 892-897; Ernest P. Schloss, "Beyond GMENAC-Another Physician Shortage from 2010 to 2030?" New England Journal of Medicine 318 (14) (April 7, 1988): 920-922.

Even less clear than the extent of the future surplus is its possible effect on medical costs. Observations that per capita use of physician services increases in geographic areas with a high ratio of physicians to population have led to the hypothesis of "physicianinduced demand." Just as excess hospital bed capacity may generate more hospital stays, this theory holds that a surplus of physicians all attempting to maintain their incomes would lead-in the absence of any controls to excess delivery of services. Repeated efforts to demonstrate this have been inconclusive. 28 It is not clear that physicians actually modify their medical practice in order to maintain a "target income." Still, if the projected surplus does in fact appear, there might be greater pressures on physicians to increase the number of services they furnish to each patient. Some people believe that it may eventually be necessary to consider reducing the supply of physicians (or curtailing their working hours).

This has actually been attempted in one Canadian province, British Columbia. A physician who wants to participate in the health program that covers all citizens of the province must have a billing account, and since 1985 the number of accounts has been limited (limits vary by specialty and geographic area). A physician who fails to obtain a billing number cannot earn a living as a physician. Critics of the system contend, however, that British Columbia is merely exporting its physician surplus to other provinces or to the United States. 29 Given the political problems health planners in the United States have experienced in trying to close hospitals, it seems unlikely that British Columbia's efforts could be reproduced here, with government regulators telling new medical school graduates to find some other profession. However, there are proposals to achieve the same goal through private means. Some of the more ambitious "managed care" agendas discussed in the final section of this report contemplate enrollment of the entire population in health maintenance organizations (HMOs) or other structured delivery systems that would match their resources to the needs of the enrolled population; this approach would potentially reduce employment opportunities for physicians. 30

28 See Louis F. Rossiter and Gail R. Wilensky, "A Reexamination of the Use of Physician Services: The Role of Physician-Initiated Demand," Inquiry 20 (2) (Summer 1983): 162-172; Kathryn M. Langwell and Lyle M. Nelson, "Physician Payment Systems: A Review of History, Alternatives and Evidence," Medical Care Review 43 (1) (Spring 1986): 5-58.

29 Morris L. Barer, "Regulating Physician Supply: The Evolution of British Columbia's Bill 41," Journal of Health Politics, Policy, and Law 13 (1) (Spring 1988): 1-25.

30 For example, Alain Enthoven has characterized the "buy right" scheme advanced by Walter McClure as requiring that "good-quality, efficient doctors prosper while others are induced to retire." Alain C. Enthoven, "Managed Competition in Health Care and the Unfinished Agenda," Health Care Financing Review, 1986 Annual Supplement, 105–119.

Reimbursement Reform

Proposals for reimbursement reform begin with the premise that traditional payment systems, under which providers receive their full costs or charges for whatever services they choose to furnish, encourage inefficiency and the delivery of unnecessary care.

The simplest type of reform is for payers to set fixed prices for defined units of service, such as a day of inpatient care or a physician office visit. However, this approach may not reduce costs if providers are able to modify the volume or nature of the services they provide to make up for the lost revenue on individual services. For this reason, the focus of reimbursement reform proposals is on developing pricing mechanisms that give providers incentives to control both volume and unit cost.

This is generally accomplished by redefining the commodity the insurer is purchasing. Instead of paying for individual units of service, the insurer makes one payment for an episode of care (as in Medicare's prospective payment system, PPS), for overall treatment of a patient during a given time period (capitation), or for treatment of an entire population (as in Canada's global budgeting system for hospitals). These approaches may be seen as aligned on an ascending scale depending on the degree of aggregation of the unit being purchased, with per-case payment at the low end and payment for an entire patient population at the other. In all cases, however, the aim is to define in advance the total amount of resources the provider may consume in furnishing treatment to a patient or group of patients.

Per-case payment and capitation give the provider an incentive to perform more efficiently in treating individual patients, either reducing the cost of producing each unit of service or reducing the number of units furnished to each patient. These approaches may therefore be seen as alternatives to external utilization controls. Global budgeting defines the total resources available for treating all patients, and may be seen as an alternative to health planning. 31 Reimbursement controls have the same goals as direct regulation of medical practice and supply, but shift the responsibility for decision-making from the third-party payer or the government to the actual providers of care. In order to live within the established rates or budgets, the providers must be self-regulating; they must make

31 In practice, the Canadian system uses both global budgeting and health planning. However, some of the rate regulation systems in the United States have explicitly superseded the health planning system. A facility that has obtained a certificate of need for expansion may proceed only if the rate commission approves the necessary increase in capital costs. For a discussion of the interplay of planning and rate regulation, see Brown, Common Sense Meets Implementation.

the same sorts of treatment and resource allocation decisions that would otherwise have been imposed externally.

As the Medicare program has demonstrated, it is possible for a single payer with sufficient market power to adopt such reimbursement changes on its own.32 The effects of this unilateral approach in a pluralistic system are uncertain. While some providers may be driven to improve their efficiency, others may instead respond to shortfalls in reimbursement from one payer by raising charges to other groups, those without the market power to dictate prices. The possibility of "cost-shifting" may mean that savings for one purchaser are not translated into real reductions in total system expenditures.

In a sufficiently competitive market, the providers' ability to engage in this "cost-shifting" may be limited. A hospital may face, not only payment limits under Medicare and Medicaid, but pressure from private insurers or employer groups to grant price discounts in order to be assured of an adequate market share. Characteristics other than efficiency may determine a provider's success in the face of these competing demands. For example, a suburban non-teaching hospital with few uninsured patients may be at a relative advantage as compared to a center city teaching facility with a heavy uncompensated care load. Individual purchasers who reduce their costs by favoring the suburban hospital may leave the society to find some other means of subsidizing essential facilities that are handicapped in price competition.

A system in which multiple payers negotiate individually with providers may, then, lead either to costshifting or to a situation in which price concerns override other societal goals, such as medical education and charity care. For this reason, some people argue that real efficiency can be achieved only if all payers are paying under the same rules.

Uniform ratesetting is common in other industrialized nations, both those with single-payer health insurance systems (as in Canada) and those where many different entities provide insurance (as in West Germany). The experience in the United States is limited to experiments in a few States beginning in the 1970s. Federal waivers of Medicare and Medicaid rules made it possible for those two payers to participate in the programs on a demonstration basis, while State laws compelled participation by private insurers and individual payers, resulting in an "all-payer" system.

32 As the Medicaid experience has shown, adoption of payment restraints by a payer with too small a market share may reduce access for the payer's enrollees. For example, low reimbursement rates are the major reason physicians decline to participate in the Medicaid program. See Congressional Research Service, Medicaid Source Book, 448-454.

Medicaid law now permits any State to include Medicaid in such a system, and Medicare may be included if the State can show that its system controls costs as effectively as PPS. However, full "all-payer" systems continue only in Maryland and in part of New York State. Several other States operate "partial-payer” systems that include all payers except Medicare. 33 These systems have generally used the price aggregation approaches described above. That is, they either establish a rate for total treatment of a case (as under PPS) or they establish a total budget for a hospital during a year, setting prices for the hospital in such a way as to achieve a target revenue amount.

It has been shown that, in 6 States with ratesetting systems, annual increases in cost per admission were consistently 3 to 4 percentage points below the national average from 1976 to 1984. During the same period, however, other States saw a drop in admissions per capita, while admissions in the ratesetting States were stable. As a result, the difference in growth in per capita rates of spending was not so striking: per capita costs rose at an annual rate of 11.5 percent a year in the ratesetting States and 13 percent a year in other States. 34 In addition, the ratesetting States had much higher costs at the outset than most other States. Some observers have questioned whether ratesetting could have achieved comparable savings in areas where costs were lower to begin with. 35

Evidence from other countries with universal ratesetting systems suggests that greater savings may be possible. In Canada, where the provinces establish global budgets for each hospital, hospital expenditures per capita were one-third lower than in the United States in 1985. (Similar systems in other industrial nations have been less successful.) 36 As admission rates are not markedly lower, there is considerable uncertainty about the sources of the difference. Some of the saving may be in administrative costs, simply because the hospitals do not need to meet the paperwork requirements of multiple payers. The rest of the difference is often attributed to differences in the intensity of the services furnished to each patient. Whether these differences reflect "underservice" in Canada or

33 Maine's system takes hospitals' Medicare revenues into account when determining what the hospitals may charge other payers, thus achieving overall budgetary control without direct Medicare participation. This approach has recently survived a legal challenge by hospitals.

34 Carl J. Schramm, Steven C. Renn, and Brian Biles, "New Perspectives on State Rate-Setting," Health Affairs 5 (3) (Fall 1986): 22-33.

35 Charles L. Eby and Donald R. Cohodes, "What Do We Know About Rate-Setting?" Journal of Health Politics, Policy, and Law 10 (2) (Summer 1985): 299-327.

36 Organisation for Economic Co-Operation and Development, Financing and Delivering Health Care: A Comparative Analysis of OECD Countries (OECD Social Policy Studies No. 4.) (Paris: 1987), 63.