DataWatch Potential Impact Of Managed Care On National Health Spending by Verdon S. Staines Abstract: Illustrative estimates suggest that if all acute health care services were delivered through staff- or group-model health maintenance organizations (HMOs), national health spending might be almost 10 percent lower. If the delivery of all such services (except those now provided by staff- or group-model HMOs) were subject to utilization review arrangements incorporating precertification and concurrent review of inpatient care, spending might be 1 percent lower. The estimates assume no changes in the health care system apart from expansion of these two forms of managed care to cover all insured persons. They also assume that moving to universal managed care would produce a one-time drop in the level of national health spending with no subsequent effect on spending growth. Managed care has attracted considerable interest as a possible way to constrain health care spending. Among workers who were covered by private employment-based insurance in 1990, 95 percent were in plans that used some form of managed care. 1 These included several kinds of health maintenance organizations (HMOs), numerous forms of utilization review, and various arrangements based on specified networks of providers. This DataWatch reports the Congressional Budget Office s (CBO s) illustrative estimates of the reductions in national health spending that might result if all acute health care services now funded through insurance were delivered through two specific forms of managed care: staff- or groupmodel HMOs and utilization review arrangements that incorporate precertification and concurrent review of inpatient care. 2 Strong evidence exists that these forms of managed care reduce costs. 3 Reliable evidence on other forms either is unavailable or fails to show significant cost savings. 4 Forms Of Managed Care Examined Staff- and group-model HMOs are unique among today s managed care organizations in that they have integrated their systems for financing and delivering care. Staff-model HMOs own the clinical facilities that the insured population must use and employ salaried physicians to serve the HMO s members exclusively. Group-model HMOs contract with multi- Verdon Staines is a principal analyst in the Human Resources and Community Development Division of the Congressional Budget Office.
D ATAWATCH 249 specialty physician groups to provide care to the HMO s membership. The HMO usually pays the physician group for the services on a capitation basis. The medical group practice then manages the provision of care independently of the HMO, contracts with the physicians who belong, and pays physicians on a fee-for-service, salaried, or other basis. Utilization review organizations contract with employers and insurers to determine that patients receive medically necessary care of good quality in a cost-effective manner. Review protocols traditionally have included precertification, concurrent review, and retrospective review, although some utilization review organizations now also offer management support, network development, and contract administration. In addition, many insurers have their own internal utilization review systems. Our study examines the effects of using these forms of managed care to cover the entire insured population in all acute health care settings. Study Methods We used simulations, based on the strategy and simplifying assumptions discussed below, to estimate how much lower 1990 levels of spending would havk been if the specified forms of managed care had been in universal (or almost universal) use. 5 The simulation strategy involves four main steps. (1) The, simulated effects are assumed all to occur via the direct effects that managed care has on personal health care spending for acute care services that are covered by public or private insurance arrangements. (2) Available data are used to estimate how the population in 1990 was distributed by primary source of insurance and by the level of effectiveness of the managed care arrangements under which their care was delivered. 6 (3) This distribution is used to help estimate the corresponding distribution in 1990 for personal health care spending on acute care services that were covered by public or private insurance. (4) Differences in the latter distribution under universal managed care are then simulated. Simplifying assumptions underlie each of these steps. In the first step, assuming that managed care affects national health expenditures only via its direct effects on personal health care spending disregards any additional effects from the remaining components of national health expenditures (administrative costs, public health activities, research, and construction). Administrative costs are ignored because satisfactory evidence is lacking to quantify how these costs vary among different forms of managed care, although it is likely that more intensive forms such as staff- and groupmodel HMOs generate higher administrative costs that would offset to some degree the simulated reductions in personal health expenditures. 7 Reductions in personal health care spending associated with universal
250 HEALTH AFFAIRS Supplement 1993 managed care also could affect construction programs in the longer term. Moreover, the analysis assumes that managed care arrangements would have no effect on spending for nonacute care or care that is not covered by health insurance. The analysis therefore excludes, as spending on nonacute care, all nursing home spending except under Medicare and spending on care for people without insurance. To allow for coinsurance payments, private insurance funding is assumed to represent 75 percent of all spending on privately insured care; out-of-pocket spending beyond that level of coinsurance is also excluded. This leaves 71 percent of total 1990 personal health care spending, or $475 billion, that might be reduced by extending managed care to the entire population (Exhibit 1). In the second step, additional assumptions are invoked to sort out types of managed care and insurance (Exhibit 2). The most important is that four groups exist, based on assumed level of effectiveness in reducing health care costs per person: group- and staff-model HMOs; effective forms of utilization review that incorporate precertification and concurrent review for hospitalization; other forms of managed care that might possibly have some effect (including HMOs modeled on individual practice associations, Exhibit 1 National Health Expenditures, By Type Of Service And By Expenditures Potentially Affected By Managed Care, 1990 Type of health service All expenditures (billions) Share of expenditures potentially affected by managed care a Billions of dollars Percentage of total National health expenditures $666.2 $475.0 71% Personal health care expenditures 585.3 475.0 81 Hospitals 256.0 256.0 100 Physicians 125.7 125.7 100 Dentists 34.0 20.4 60 Ocher professionals 31.6 31.6 100 Home health 6.9 6.9 100 Drugs and medical nondurables 54.6 16.4 30 Vision products and durables 12.1 4.2 35 Nursing home care 53.1 2.5 5 Other personal health care 11.3 11.3 100 Administration and net cost of private health insurance 38.7 0 0 Public health activity 19.3 0 0 Research 12.4 0 0 Construction 10.4 0 0 Source: Congressional Budget Office calculations based on K.R. Levit et al., National Health Expenditures, 1990, Health Care Financing Review (Fall 1991): 29-54. a Expenditures potentially affected by managed care arrangements consist of those personal health care expenditures that are for acute care services and that are covered by public or private insurance arrangements.
D ATAWATCH 251 Exhibit 2 Distribution Of U.S. Population, By Primary Source Of Health Insurance Coverage And Form Of Managed Care, 1990 Form of managed care Staff- or Effective Other No Primary source of group- utilization managed managed insurance coverage Total model HMO review a care care All sources 259.6 16.7 65.5 119.4 57.9 Medicare 32.1 1.3 0.0 30.8 0. 0 Medicaid 15.4 0.3 0.0 1.3 13.9 Private or other public 176.9 15.2 65.5 87.3 8.8 No insurance 35.2 0.0 0.0 0.0 35.2 Sources: Congressional Budget Office calculations based on data from the Social Security Administration Office of the Actuary; Census Bureau, Current Population Survey, March 1990 Supplement; Health Care Financing Administration; Group Health Association of America; C.B. Sullivan and T. Rice, The Health Insurance Picture in 1990. Health Affairs (Summer 1991): 104-115; and Physician Payment Review Commission, Annual Report to Congress, 1992. Note: Millions of persons. HMO is health maintenance organization. a Defined as those that incorporate precertification and concurrent review of inpatient care. network-model HMOs, point-of-service plans, preferred provider organizations, normal Medicare care, and non-hmo forms of Medicaid managed care); and arrangements that involve no managed care. The third step estimates how spending for insured acute care services is distributed not only by primary insurer but also by managed care category (Exhibit 3). This step rests on crucial assumptions, some of which are needed because various forms of managed care had been widely adopted by 1990. One assumption is that average spending per person would have been the same for the population across all four managed care categories in 1990 if no managed care arrangements had been in effect then, except for systematic differences related to the proportions of people in each category who were age sixty-five or older or who had no health insurance. This implies that, apart from the effects of age and insurance status, observed spending per person in 1990 was lowest in the managed care categories most effective in reducing costs and that relative expenditures tracked the assumed effectiveness of the categories. In addition, two sets of assumptions (Alternatives 1 and 2) are made about the cost reductions associated with each category of managed care. Both alternatives include the same assumptions for staff-model and groupmodel HMOs. Specifically, these HMOs are assumed to reduce spending by 15 percent compared with spending under traditional private health insurance and Medicare and by 7.5 percent compared with traditional Medicaid programs. (The effect assumed is smaller than that estimated by the RAND Health Insurance Experiment, partly because that evidence came from a single, well-established HMO and partly because hospital admissions and
252 HEALTH AFFAIRS Supplement 1993 Exhibit 3 Estimated Distribution Of 1990 Personal Health Expenditures For Insured Acute Care, By Primary Insurer, Managed Care Category, And Assumed Effectiveness Of Managed Care, 1990 Primary source of insurance coverage Alternative 1 b Total Form of managed care Staff- or Effective group- utilization model HMO review a Other managed care No managed care All sources $475.0 $26.5 $106.8 $254.3 $87.4 Medicare 108.9 3.9 0.0 105.0 0.0 Medicaid 47.2 0.7 0.0 3.9 42.6 Private or other public 289.1 21.9 106.8 145.4 15.0 No insurance 29.8 0.0 0.0 0.0 29.8 Alternative 2 c All sources 475.0 26.0 108.0 254.5 86.5 Medicare 108.9 3.8 0.0 105.1 0.0 Medicaid 47.2 0.7 0.0 3.9 42.6 Private or other public 289.7 21.5 108.0 145.4 14.7 No insurance 29.2 0.0 0.0 0.0 29.2 Source: Congressional Budget Office calculations based on Exhibits 1 and 2; and K.R. Levit et al., National Health Expenditures, 1990, Health Care Financing Review (Fall 1991): 29-54. Note: Billions of dollars. HMO is health maintenance organization. a Defined as those that incorporate precertification and concurrent review of inpatient care. b Assumes a 4 percent reduction for effective utilization review under Medicare and private insurance. c Assumes a 1 percent reduction for effective utilization review under Medicare and private insurance. lengths-of-stay have declined in subsequent years.) 8 The alternatives differ in their assumptions for effective utilization review and for other forms of managed care, where the evidence is more limited (Exhibit 4). Alternative 1 assumes that effective forms of utilization review-those incorporating precertification and concurrent review of inpatient care-reduce spending by 4 percent from its levels under traditional health insurance and Medicare and by 2 percent from its level under Medicaid; and other forms of managed care reduce spending by 2 percent from traditional health insurance and Medicare levels and by 1 percent from Medicaid levels. Alternative 2 assumes that effective forms of utilization review reduce spending by 1 percent from traditional health insurance and Medicare levels and by 0.5 percent from Medicaid levels; and other forms of managed care reduce spending by zero percent. In the fourth step, the simulations incorporate three additional assumptions. They assume (1) no changes in the health care system apart from expansion of these two forms of managed care to cover the entire insured population; (2) that moving to universal managed care would produce a one-time drop in the level of national health expenditures; and (3) that effective utilization review would not displace existing staff- and group-
D ATAWATCH 253 Exhibit 4 Reductions In Potentially Manageable Personal Health Expenditures That Are Assumed To Result From Alternative Managed Care Arrangements, By Primary Insurer, 1990 Primary source of insurance coverage Form of managed care Staff- or Effective group- utilization model HMO review a Other managed care No managed care Alternative 1 Medicare 15.0% 4.0% 2.0% 0.0% Medicaid 7.5 2.0 1.0 0.0 Private or other public 15.0 4.0 2.0 0.0 No insurance 0.0 0.0 0.0 0.0 Alternative 2 Medicare 15.0 1.0 0.0 0.0 Medicaid 7.5 0.5 0.0 0.0 Private or other public 15.0 1.0 0.0 0.0 No insurance 0.0 0.0 0.0 0.0 Source: Congressional Budget Office calculations. Note: HMO is health maintenance organization. a Defined as those that incorporate precertification and concurrent review of inpatient care. model HMOs that reduce costs by a larger proportion. Study Results We organize our findings around two questions: What are the maximum reductions in health care spending that might flow from the universal adoption of staff- or group-model HMOs or effective forms of utilization review within the current health care system? And how would these savings be distributed among primary insurers? Maximum reductions in spending. According to our simulations, if all insured care had been delivered through staff- or group-model HMOs in 1996, total health care costs might have been cut by as much as $64 billion (see Alternative 2, Exhibit 5). Given the growth in national health spending since 1990, the same proportional savings in 1992 would amount to approximately $78 billion, based on CBO projections. 9 Using effective forms of utilization review to deliver all health care (except services currently provided through more cost-effective staff- and group-model HMOs) could have yielded substantially smaller potential savings. Distribution of savings. If staff- or group-model HMOs had been used universally in 1990, about two-thirds of the savings would have come from care for people whose primary insurance coverage was private. Care for Medicare enrollees would have contributed another one-fourth of the savings, with care for Medicaid beneficiaries contributing the remainder.
254 HEALTH AFFAIRS Supplement 1993 Exhibit 5 Estimated Savings In Health Expenditures Under Various Forms Of Managed Care And By Assumed Effectiveness Of Managed Care, In Billions Of Dollars And As Percentage Of Total Spending, 1990 Alternative 1 a Alternative 2 b Form of managed care Form of managed care Staff- or Effective Staff- or Effective group- utilization group- utilization Estimated dollars saved model HMO review c model HMO review c All sources of coverage $51.2 $6.6 $64.2 $2.9 Medicare 13.9 2.1 16.4 1.1 Medicaid 3.4 0.9 3.5 0.2 Private or other public 33.8 3.6 44.2 1.6 No insurance 0.0 0.0 0.0 0.0 Percent of total spending saved All personal health spending 8.7% 1.1% 11.0% 0.5% National health expenditures 7.7 1.0 9.6 0.4 Source: Congressional Budget Office calculations. Note: HMO is health maintenance organization. a Assumes a 4 percent reduction for effective utilization review under Medicare and private insurance. b Assumes a 1 percent reduction for effective utilization review under Medicare and private insurance. c Defined as incorporating precertification and concurrent review of inpatient care. If managed care incorporating effective forms of utilization review had replaced all other delivery arrangements except staff-model or group-model HMOs throughout 1990, more than half of the savings would again have come from care for people with primarily private insurance coverage. Care for Medicare enrollees would have contributed about one-third of the total savings, with care for Medicaid beneficiaries contributing the remainder. Not surprisingly, the choice of assumptions about the effectiveness of particular categories of managed care affects the estimates of total savings. For universal adoption of staff- and group-model HMOs, estimated savings are larger under the assumptions of Alternative 2, but for (almost) universal adoption of effective utilization review, estimated total savings are larger under the assumptions of Alternative 1. What accounts for this asymmetry? Both sets of assumptions imply that the same reduction in total personal health spending would have resulted if the health care system moved from a hypothetical situation in which no one was covered by any managed care in 1990 to a hypothetical situation in which everyone was enrolled in staffor group-model HMOs in 1990. Because Alternative 1 assumes relatively larger effects for both effective utilization review and other forms of managed care, however, it implicitly assumes that the managed care actually in place in 1990 had already achieved more of the potential savings that would have resulted from universal coverage by staff- or group-model HMOs. So moving to complete coverage by such HMOs yields smaller
D ATAWATCH 255 estimated savings under Alternative 1 than under Alternative 2. Similarly, mandating the use of effective utilization review in place of delivery systems that are less cost-effective is equivalent to shifting people to the category for effective utilization review from those for other forms of managed care and no managed care. Doing so would achieve larger savings in total health care spending, as the proportional savings assumed for utilization review are larger than those for the two less-effective categories. Assumptions for Alternative 1 are larger for effective utilization review and other forms of managed care than are the assumptions for Alternative 2. They also entail larger differentials in effectiveness between the category for effective utilization review and those for other forms of managed care and no managed care. Thus the Alternative 1 assumptions yield larger estimates of the savings from adopting effective utilization review. Discussion And Caveats As these estimates illustrate, potential savings as large as almost 10 percent of national health expenditures $78 billion in 1992 might result from universal use of HMOs. Much smaller savings-no greater than about 1 percent of national health expenditures and possibly considerably less-might result if effective utilization review governed the delivery of all health care for insured people who are not already enrolled in staff- or group-model HMOs. Most of the spending reductions from universal adoption of staff- and group-model HMOs would relate to care for people currently insured through the private sector, with the majority of the remaming savings accruing to Medicare. For several reasons, these illustrative estimates should be interpreted with considerable caution. One key caveat is that the estimates should not be generalized to other forms of managed care for which there is no evidence. Other qualifications arise because the results are no more reliable than the data and assumptions on which they are based. As noted earlier, the evidence for assumptions about the effectiveness of various forms of managed care is often quite limited. Similarly, because available data do not provide a sound basis for simulating how more widespread use of managed care would affect administrative costs, the results do not reflect any such effects and therefore tend to overstate the potential savings from expanding managed care. Classifying individuals by the primary source of their insurance coverage and then allocating expenditures among sources of insurance on that basis oversimplifies the true flow of health care financing. Furthermore, data about the distribution of individuals among the various forms of managed care are also relatively limited and not always fully consistent. The estimation methods also assume that care for people without insurance
256 HEALTH AFFAIRS Supplement 1993 coverage is not managed. That assumption would introduce errors to the extent that managed care providers render uncompensated care or that uninsured persons pay for managed care from their own resources. Although the simulation assumes otherwise, the impact of managed care on health spending per person might be different if it were applied to all consumers, providers, and services rather than to just part of the health care system. For example, if the change to universal managed care were part of a comprehensive restructuring of the health care system that included incentives to choose efficient arrangements, the managed care component of the package might have a larger impact than if it were adopted on its own. 10 That is because, under the present structure of competition among insurers, managed care arrangements may not be delivering their full potential of cost savings. It is at least as likely, however, that mandating universal adoption of managed care might have smaller effects than suggested by estimates based on experience. If all consumers and providers were required to adopt managed care arrangements, the new participants levels of commitment to the processes and values implicit in managed care approaches might be less than those of the current participants, who have voluntarily chosen these arrangements. Mandating managed care of any type for the entire population would imply significant changes in the current health care system. Limitations could be placed on people s existing choice of providers, health insurance coverage, and treatment alternatives; also, access to new technologies might be restricted or permitted only after a longer waiting time. In other words, managed care, when effective, does impose constraints on patients and providers, and some people view those constraints as undesirable. A reason for caution when interpreting these estimates is that they relate to the level of health costs, not the rate of increase. The limited available evidence suggests that managed care does not affect the underlying rate of growth in those costs, and the simulations assume that mandating managed care would not have an enduring effect on the rate. 11 It might slow that growth, however, if it were introduced as part of a comprehensive restructuring of the health care system that incorporated strong incentives to choose efficient arrangements. Universal managed care in such a setting could facilitate greater control over the adoption of new technologies (for example, better guideline development or cost-saving strategies). However, under the current system, with its recent high rates of cost increase, even a 10 percent reduction in national health spending would be offset by approximately one year s growth in health care spending. Thus, universal adoption of some forms of managed care could yield substantial one-time savings but might not address the longer-term cost issue.
D ATAWATCH 257 The Author acknowledges research assistance provided by Julia Jacobsen and valuable comments from Kathryn Langwell. NOTES 1. E.W. Hoy, R.E. Curtis, and T. Rice, Change and Growth in Managed Care, Health Affairs (Winter 1991): 18-36. 2. Congressional Budget Office, The Potential Impact of Certain Forms of Managed Care on Health Care Expenditures (CBO Staff Memorandum, August 1992, revised). 3. W. Manning et al., A Controlled Trial of the Effect of a Prepaid Group Practice on Use of Services, The New England Journal of Medicine (7 June 1984): 1505-1510; R. Brown, Biased Selection in the Medicare Competition Demonstrations (Washington: Mathematics Policy Research, 1987); S. Greenfield et al., Variations in Resource Utilization among Medical Specialties and Systems of Care: Results from the Medical Outcomes Study, Journal of the American Medical Association (25 March 1992): 1624-1630; R. Khandker and W. Manning, The Impact of Utilization Review on Costs and Utilization, in Health Economics Worldwide, ed. P. Zweifel and H.E. Frech III (The Netherlands: Kluwer Academic Publishers, 1992), 47-62; P. Feldstein et al., Private Cost Containment: The Effects of Utilization Review Programs on Health Care Use and Expenditures, The New England Journal of Medicine (19 May 1988): 1310-1314; T. Wickizeretal., Does Utilization Review Reduce Unnecessary Care and Contain Costs? Medical Care (June 1989): 632-646; T. Wickizer, The Effect of Utilization Review on Hospital Use and Expenditures: A Review of the Literature and an Update on Recent Findings, Medical Care Review (Fall 1990): 327-363; R. Scheffler et al., The Impact of Blue Cross and Blue Shield Plan Utilization Management Programs, 1980-1988, Inquiry (Fall 1991): 263-275; SRI International, Evaluation of the Arizona Health Care Cost Containment System: Final Report (Report prepared for the Health Care Financing Administration, January 1989); Laguna Research Associates, Evaluation of the Arizona Health Care Cost Containment System Demonstration: First Outcome Report (Report prepared for HCFA, July 1991); and D.A. Freund et al., Evaluation of the Medicaid Competition Demonstrations, Health Care Financing Review (Winter 1989): 81-97. 4. Congressional Budget Office, The Effects of Managed Care on Use and Costs of Health Services (CBO Staff Memorandum, June 1992). 5. A detailed description of the simulation methodology is available from the author. 6. Within the analysis, public programs other than Medicare and Medicaid are grouped with private coverage as a primary source of insurance. 7. The Group Health Association of America has estimated administrative costs for HMOs. See J.F. Doherty, statement on health care administrative costs and the Canadian health care system before the Joint Economic Subcommittee on Education and Health, 16 October 1991. 8. Manning et al., A Controlled Trial of the Effect of a Prepaid Group Practice. 9. Congressional Budget Office, Projections of National Health Expenditures (Washington: CBO, October 1992), 22, Table 3. 10. A.C. Enthoven, Multiple Choice Health Insurance: The Lessons and Challenge to Employers, Inquiry (Winter 1990): 368-375; and AC. Enthoven and R. Kronick, "Universal Health Insurance through Incentives Reform, Journal of the American Medical Association (15 May 1991): 2532-2536. 11. See, for example, J. Newhouse et al., Are Fee-for-Service Costs Increasing Faster than HMOs Costs? Medical Care (August 1985): 960-966.