"overservice" in the United States is the subject of continuing debate. 37

In a sense, the statistical evidence may be beside the point. An all-payer system could in theory fix its prices at any level, with the potential consequence of reduced access or quality if the prices are set too low. The available data may thus be taken as indicating, not the savings that could hypothetically be achieved, but the savings that were politically feasible in specific States during a specific period. Continuing pressure by consumers and providers for the adoption of new medical technologies may limit the ability of ratesetting systems to restrain expenditure growth over the long term. Even in Canada, overall medical expenditures outpaced inflation by 2.9 percent a year in the period 1980-87, almost the same as the 3.0 percent annual rate observed in the United States in the same years. 38 The ultimate efficacy of reimbursement controls may depend, in the same way that the success of health planning depends, on the political will to constrain health care consumption.

That political will might in turn depend on perceptions of the impact of reimbursement controls on the quality of care. The effect of Medicare's prospective payment system, for example, has been argued continuously since its implementation in 1983. One of the immediate responses of hospitals to the incentives of the new system was to shorten the average length of stay in the hospital for each Medicare patient (although average length of stay had already been dropping for several years). Opponents of the new system have contended that patients were being discharged "quicker and sicker," transferred to their own homes or to nursing homes at a stage in their recovery when they still required hospital-level care. Because of a lack of satisfactory measures of medical care outcomes for large populations, evidence on this issue remains largely anecdotal. Still, the possibility that there has been a deterioration in quality of care for at least some Medicare patients since the implementation of PPS cannot be ruled out. The hospitals themselves argue that current payment levels are insufficient to maintain adequate quality. At the same time, the Administration and the Prospective Payment Assessment Commission (the independent commission that reviews PPS) have argued that hospitals are still not operating at peak efficiency and that further payment restraint is needed to provide continued incentives for cost reduction. 39

37 For a variety of views on this subject, see the series of articles on Canada's hospital system in Health Affairs 7 (5) (Winter 1988).

38 Schieber and Poullier, International Health Care Expenditure Trends:

1987.

39 U.S. Prospective Payment Assessment Commission, Report and Recommendations to the Secretary, U.S. Department of Health and Human Services (Washington, D.C.: Government Printing Office, March 1989); For a recent review of hospital cost responses to PPS, see Steven H. Sheingold, "The

This debate illustrates one potential dilemma in the strategy of achieving savings by relying on the political process to limit the financial resources available to providers. On the one hand, legislators driven by budgetary concerns may continue to ratchet down spending limits until they have clear evidence that quality has been seriously affected. On the other hand, provider or constituent pressure may lead them to relax those limits before the providers have done everything possible to improve their efficiency. Because no one knows the ideal amount to spend on medical care, some people say that this process can never achieve equilibrium and that cost control efforts should instead depend on the process through which other sectors of the economy achieve "correct" spending levels: the free market. Proposals for encouraging competition in health care represent the last of the strategies to be reviewed in this report.

Competition

The idea of reducing health care costs by promoting competition in the health care marketplace was first advanced in the 1970s. Some analysts, arguing that such initiatives as rate regulation, health planning, and utilization review had been compromised by political interference, contended that the free market was better equipped to control costs than Government was. By the early 1980s, this view had wide currency and had become the official policy of the Reagan Administration. Since then, there has been a continuing debate between advocates of competition and those who favored further regulatory interventions by Government. The debate has been complicated by a lack of agreement over what "competition" consists of. What is the health care market? Who are the purchasers, and what are they buying?

In a simple market, hospitals and physicians would compete directly for the individual consumer's dollar. The consumer would pick the best values just as he or she does when buying any other commodity. As was suggested in the discussion of cost-sharing, it is not clear that consumers are capable of making such evaluations; moreover, many purchasing decisions are made by physicians on their patients' behalf, rather than directly by consumers. Finally, because few people can afford the costs of care for a major illness, most of the consumer's dollar is spent on health insurance, not on medical care itself. As was suggested earlier, this is true even when the insurance plan imposes cost-sharing requirements on enrollees, because most health care costs are incurred by a relatively

First Three Years of PPS: Impact on Medicare Costs," Health Affairs 8 (3) (Fall 1989): 191–204.

Competition among insurers can result in real cost savings only if the insurers have some influence on the costs of health care itself. In this model, insurers compete to offer lower prices by acting as prudent purchasers, proxies for the rational consumer. The insurers are selling a new product, no longer simply insurance, but "insured health care." To some extent, this new insurance market has already arrived. As was suggested earlier, most insurance plans, both public and private, have adopted some utilization control measures. Very few insurers are still passive bill-payers.

Once all insurers have adopted these basic cost control measures, further competition would presumably require more aggressive interventions by insurers in the health care system. Proponents of competition contemplate a marketplace in which insurers develop structured delivery systems, with the highest profits going to those whose networks are most efficient. The prototype for these systems is the HMO. More recently, some insurers have been experimenting with hybrid programs, such as "point-of-service plans," that are less structured and provide somewhat greater flexibility to enrollees.

Health Maintenance Organizations-A health maintenance organization (HMO) is a form of health insurer; like any other insurer, it accepts financial responsibility for a defined set of health care benefits in return for a fixed monthly per capita premium. Unlike other insurers, HMOS directly provide or arrange for health care services, through affiliated physicians, hospitals, and other providers. The enrollees covered by the HMO agree to obtain all services, except emergency and out-of-area care, from or with the authorization of the HMO or its affiliated providers. The

40 Alain Enthoven has summarized the alternatives to price competition: "[S]election of preferred risks, market segmentation, product differentiation that raises the costs of comparing products, discontinuity in coverage, refusal to insure certain individuals or exclusion of coverage for treatment of preexisting medical conditions, biased information regarding coverage and quality, and erection of entry barriers [that is, to new competitors]." Alain C. Enthoven, "Managed Competition of Alternative Delivery Systems," Journal of Health Politics, Policy and Law 13 (2) (Summer 1988): 305–321.

HMO has no liability to pay for unauthorized nonurgent care obtained outside the organization. Ordinarily, the enrollee's point of entry into the system is through a single primary care provider, who functions as a "gatekeeper," determining when a patient may see a specialist or be admitted to the hospital. The HMO exerts further administrative controls on use of services through authorization mechanisms and/or treatment protocols. HMOs also use a variety of other cost-saving techniques, such as negotiated discounts with providers and payment mechanisms that place individual providers at risk for the costs of the services they furnish or order.

The particular cost-saving techniques adopted by HMOs and other "managed care" plans are not fundamentally different from the regulatory approaches described in the preceding sections. An HMO imposes external utilization review on its participating providers and may develop practice guidelines or protocols. Staff or group practice model HMOs (those that employ physicians on a full-time basis) impose supply constraints, limiting available resources to those needed by their membership. Individual practice associations (IPAs, whose physicians practice in their own offices and see a mix of HMO and non-HMO patients) use payment methods that create financial incentives to control utilization, such as capitation or expenditure targets.

One additional cost-saving approach that was once unique to HMOs is "gatekeeping." Under a gatekeeping approach, a patient receives all non-emergency care from, or with the authorization of, a single primary care provider. The provider thus functions as a "gatekeeper," preventing the enrollee from independently accessing specialists or other services and presumably managing the overall care of the patient. The extent to which gatekeeping produces savings over and above those provided by the other costsaving techniques adopted by HMOS is uncertain. The results of one experiment, the SAFECO health plan operated by United HealthCare in the early 1980s, suggest that gatekeeping alone has little effect on overall cost. While primary care providers reduced the number of referrals to specialists, they were unable to control the behavior of the specialists once a referral had occurred. There was no meaningful reduction in hospital admissions, 70 percent of which were controlled by the specialists.41 Greater success

41 Stephen Moore, Diane Martin, and William Richardson, "Does the Primary-Care Gatekeeper Control the Costs of Health Care? Lessons from the SAFECO Experience," New England Journal of Medicine 309 (22) (December 1, 1983): 1400-1404; For the extent to which specialty referrals may determine overall costs, see John K. Glenn, Frank H. Lawler, and Mark S. Hoerl, "Physician Referrals in a Competitive Environment: An Estimate of the Economic Impact of a Referral," Journal of the American Medical Association 258 (14) (October 9, 1987): 1920-1923.

has been reported by some State Medicaid programs, which have established "primary care case management" programs for segments of their covered populations. Gatekeeping reduced such inappropriate behaviors as the use of emergency rooms for primary care. However, the utilization patterns addressed by these programs may be characteristic of Medicaid beneficiaries in the inner city and not of other groups; it is not clear that equivalent savings could be achieved with a general population. There is some evidence that most patients' care is already "managed" by their primary care physicians, at least to the extent that it is managed under formal gatekeeping arrangements.42

Aside from the uncertain effects of gatekeeping, managed care depends on the same kinds of interventions in medical care practice, supply, and financing that might otherwise be attempted on a regulatory basis. The difference is that, instead of relying on the political process to make decisions about the allocation of health care resources, managed care privatizes these decisions. The choice among alternative cost control methods-and the stringency with which these methods will be applied-will be made by the free market. The fundamental contention of proponents of the competitive approach is that the market can impose discipline on the health care system that cannot be imposed through external regulation.

This contention rests on two key assumptions: first, that buyers will, all other things being equal, select the most cost-effective plan; second, that managed care offers greater cost-saving potential than the various regulatory controls described earlier.

One critical factor has made it difficult to generalize about the efficacy of HMOs as a cost-saving approach: the problem of "biased selection" in systems that allow a choice between a conventional health insurance plan and an HMO. Numerous studies of such "dual choice" employer group plans have shown that the members of the group choosing the HMO option used fewer health services before their enrollment than persons who chose a conventional plan. Similar patterns have been observed in Medicare HMO enrollment.43 This does not necessarily mean that HMO enrollees were healthier. Studies using self-reported condition and similar limited measures of health status have found no difference between HMO and indemnity enrollees. It may be, then, that HMO enrollees

42 A.J. Dietrich, et al., "Do Primary Physicians Actually Manage Their Patients' Fee-for-Service Care?" Journal of the American Medical Association 259 (21) (June 3, 1988): 3145-3149.

43 For a review of the evidence, see General Accounting Office, Medicare: Increase in HMO Reimbursement Would Eliminate Potential Savings. Report to the Chairman, Subcommittee on Health, House Committee on Ways and Means (Washington, D.C.: November 1989). [GAO/HRD-90–38]

are simply less prone to seek health services, regardless of their condition.44

In groups that have no HMO option but do offer a choice between high- and low-option plans the common selection pattern is for the higher users of services to choose the more comprehensive plan.45 In most group health programs offering a choice between HMOs and conventional plans, the HMO options offer more comprehensive coverage, with less enrollee cost-sharing, than even a high-option conventional plan. That higher users of services still prefer the conventional plan suggests that non-financial aspects of HMOs affect the decision, such as limited choice of providers, bureaucratic constraints on treatment, or waiting time for non-urgent care. There is stronger evidence of biased selection for staff and group model HMOs, the most restrictive, than for IPAs, which are less likely to disrupt enrollees' traditional ways of obtaining medical care.

Possible solutions to the problem of selection bias will be discussed further below. One immediate consequence, however, is that the differences between the populations in HMOs and conventional plans have made it difficult to determine whether HMOs are actually more efficient than other insurers. Only one major study has corrected adequately for this problem. In a second component of the RAND Health Insurance Experiment (HIE) cited earlier, enrollees were randomly assigned to the Group Health Cooperative of Puget Sound and an equally comprehensive conventional plan; neither plan required cost-sharing. This arrangement allowed comparisons of efficiency with identical benefits and populations with comparable health needs. The results strongly confirmed the cost-saving potential of the HMO. The HMO enrollees had 40 percent fewer hospital admissions; their use of ambulatory services was about the same as that of the conventional enrollees. Overall, costs for the HMO group were estimated to be 28 percent lower than for the control group. 46 There were no perceived effects on quality; measures of health outcomes were generally the same for both groups. 47

** Fred J. Hellinger, "Selection Bias in Health Maintenance Organizations: Analysis of Recent Evidence," Health Care Financing Review 9 (2) (Winter 1987): 55-63.

45 Robert W. Broyles and Michael D. Rosko, "The Demand for Health Insurance and Health Care: A Review of the Empirical Literature," Medical Care Review 45 (2) (Fall 1988): 291–338.

46 Willard G. Manning, et al., "A Controlled Trial of the Effect of a Prepaid Group Practice on Use of Services," New England Journal of Medicine 310 (23) (June 7, 1984): 1505-1510.

47 John E. Ware, Jr., et al., "Comparison of Health Outcomes at a Health Maintenance Organisation With Those of Fee-for-Service Care," Lancet (May 3, 1986): 1017-1022. One group, low-income HMO enrollees with existing health problems, had poorer outcomes, possibly because of difficulty dealing with the HMO's internal bureaucracy.

prevent enrollees from using the emergency room for non-urgent care; this approach is common in health maintenance organizations (HMOs) and has been adopted by some State Medicaid plans. It is not certain, however, that even such narrower measures will deter only unnecessary care.

Finally, and perhaps most important from the perspective of cost reduction, cost-sharing may not modify the course of care once treatment has begun, presumably because the decision-making has generally shifted from the patient to the physician. This finding of the HIE is partly a result of the design of the experiment. Regardless of the level of cost-sharing required, each plan had an out-of-pocket limit, a point beyond which the insurer assumed full responsibility for all further expenses. In the absence of such a limit, enrollees might have been more likely to decline the services ordered by their physicians. At the same time, however, the most severely ill would have been subject to catastrophic financial losses.

In 1978, 10 percent of U.S. families accounted for 67 percent of total health expenditures. U.S. Congress, Congressional Budget Office, Catastrophic Medical Expenses: Patterns in the Non-Elderly, Non-Poor Population (Washington, D.C.: Government Printing Office, December 1982), xviii.

These techniques are sometimes referred to by health insurers as "managed care." Others restrict the term “managed care" to the more aggressive interventions in the health care system represented by HMOs or similar entities. This is the sense in which the term will be used later in this report.

• Concurrent review, under which patients already admitted to the hospital are monitored to ensure the appropriateness of their continued stay;

• Voluntary or mandatory second opinions before elective surgery;

• Case management, under which the payer or the payer's agent attempts to assume control of the overall delivery of services to an individual highcost patient;

• Various approaches for shifting the locus of care from high-cost to low-cost settings. These include requirements that certain surgical procedures be performed on an outpatient basis, or that diag nostic tests ordinarily required for inpatients be conducted before the patient is admitted to the hospital.

Utilization controls, especially pre-admission certification and concurrent review, have become a standard feature of health insurance plans during the 1980s. They are now used in the Medicare program, in 29 State Medicaid programs (as of 1987), and in 72 percent of employer-sponsored health plans (as of 1988), up from 59 percent just a year earlier. Despite the rapid adoption of utilization control systems by both public and private payers, they have received little systematic study, and evidence that they actually reduce spending is limited. Pre-admission review has the strongest track record; one controlled study found that it produced net savings for an average employee group of 7.3 percent, with even higher savings for groups that had very high utilization before the programs were initiated. The evidence on some of the other approaches is less clear. For example, some studies have suggested that voluntary second surgical opinion programs may not deter enough unnecessary surgery to offset the costs of the second opinions themselves; mandatory programs appear to be more successful.9

There are also concerns that even the most successful utilization control approaches focus only on inpatient care and may merely shift the site in which care is delivered without fundamentally changing medical practice.10 If a reduction in inpatient admissions is

7 Phoebe A. Lindsey, "Medicaid Utilization Control Programs: Results of a 1987 Study," Health Care Financing Review 10 (4) (Summer 1989): 79-92; and Jon Gabel, et al., “Employer-Sponsored Health Insurance in America," Health Affairs 8 (2) (Summer 1989): 116-128.

Paul Feldstein, Thomas Wickizer, and John Wheeler, "Private Cost Containment: The Effects of Utilization Review Programs on Health Care Use and Expenditures," New England Journal of Medicine 318 (20) (May 19, 1988): 1310-1314.

For a review of the literature, see Danny Ermann, "Hospital Utilization Review: Past Experience, Future Directions," Journal of Health Politics, Policy and Law 13 (4) (Winter 1988): 683-704.

10 For a discussion of this issue, see Institute of Medicine, Controlling Costs and Changing Patient Care? The Role of Utilization Management (Washington, D.C.: 1989).

followed by an increase in outpatient services, savings may be only temporary; soon costs may begin to rise again as rapidly as before. One observer has argued that, because technologies that were once available only in hospitals are now widely diffused in the community, the hospital is no longer the appropriate focus of cost-containment efforts. At the same time, however, utilization controls for ambulatory services have been slow to develop. In part, this is because most ambulatory services have relatively small prices. The administrative costs of reviewing each service may outweigh any potential savings. 11 Some insurers have begun to require prior authorization for the most costly outpatient services, such as CAT scans or other major diagnostic procedures. Whether such measures are actually producing savings is not yet known.

Utilization controls face another barrier that may be even more important than administrative costs: the subjective nature of medical practice. Each patient is somehow unique, and external reviewers may have difficulty overriding the clinical judgments of individual practitioners in specific cases. This may be especially true when there is little consensus about the most appropriate treatment for a given condition, a problem to be discussed in the next section. In any event, some observers have contended that a persistent physician who is prepared to appeal a denial of authorization will often prevail. (The relative leverage of the individual practitioner may have been enhanced by recent legal decisions subjecting external utilization control agents to malpractice liability for denials of necessary care.) In consequence, utilization review may function as a delaying tactic rather than an absolute control, achieving savings only because some physicians will not take the trouble to protest the reviewers' decisions. The result has been termed "rationing by inconvenience." 12 Such savings as are achieved may diminish over time as physicians become more skillful in dealing with the system.

For this reason, some analysts have suggested that savings over a longer term may depend on the extent to which providers "sign on" to the concept of eliminating unnecessary services. In this view, real utilization control will require voluntary changes in the way physicians practice medicine.

Modifying Practice Styles-Beginning in the 1970s, studies by Wennberg and others showed that there was substantial geographic variation in the rate of use of specific medical or surgical procedures. For example, the rate of tonsillectomies in one area of New

"Jeff C. Goldsmith, "Competition's Impact: A Report from the Front," Health Affairs 7 (3) (Summer 1988): 162-173.

12 Gerald W. Grumet, "Health Care Rationing Through Inconvenience: The Third Party's Secret Weapon," New England Journal of Medicine 321 (9) (August 31, 1989): 607–611.

England was six times higher than the lowest rate in the region. 13 While some of the variations uncovered in "small area analysis" might be attributable to differences in the incidence of illness in different populations, this explanation appeared to be insufficient to account for all the variation; some other factors had to be at work. One hypothesis was that physicians in different areas had different "practice styles." Each community had its own medical culture, its own characteristic way of diagnosing or treating particular diseases or conditions. Physicians adopted the practice style of their community in the absence of firm and objective information about which treatment approach was actually superior.

Other explanations have been offered for small area variations in medical practice; these will be discussed further below. However, the practice style hypothesis has won many supporters and has led to proposals for controlling medical care costs by (a) improving knowledge of the relative efficacy of different medical treatments and (b) disseminating this knowledge to practitioners in the expectation that they will modify their practice styles accordingly. The Omnibus Budget Reconciliation Act of 1989 (P.L. 101-239) establishes a new program within the Department of Health and Human Services for research on the effectiveness of medical treatments and the development of practice guidelines. Not all of the proponents of this initiative view it as a cost-containment measure. Some view it chiefly as a possible way of improving quality of care, and therefore worth pursuing whether or not any cost savings result. The following discussion, however, considers only the potential of medical practice research to reduce costs.

To have a significant impact, guidelines will need to address areas of practice on which there is real disagreement among physicians. There have been some efforts in the past to codify elements of medical practice on which there already existed a consensus. However, if most physicians already agree on the best treatments, promulgating that agreement in the form of guidelines may not have a measurable impact on medical practice. (This appears to have been the case, for example, with a 1984 consensus report on the treatment of high blood pressure. 14) For this reason, the treatment research initiative will focus on conditions for which there is found to be a wide variation in current practice. Because the Nation is just beginning to devote significant resources to research on the outcomes of alternative medical treatments, it may

13 John Wennberg and Alan Gittelsohn, "Variations in Medical Care Among Small Areas," Scientific American 246 (April 1982): 120-134.

14 Martha N. Hill, David M. Levine, and Paul K. Whelton, "Awareness, Use, and Impact of the 1984 Joint National Committee Consensus Report on High Blood Pressure," American Journal of Public Health 78 (9) (September 1988): 1190-1194.