Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
The Impact of Cost Containment Efforts on Racial and Ethnic Disparities in Healthcare: A Conceptualization Thomas Rice, Ph.D. Department of Health Services UCLA School of Public Health INTRODUCTION Like all developed countries, the United States continues to battle high and rising health care costs. In the mid to late 1990s, U.S. health expendi- tures as a percentage of gross domestic product had temporarily stabi- lized for the first time in decades, albeit at a level almost 30% higher than in any other country (Anderson and Hussey, 2001). More recently, how- ever, there has been a resurgence in health care cost inflation, with some indications (unproven, as yet) that we are entering another era of double- digit annual increases.1 This resurgence appears to have been caused by a number of factors, including a spike in demand for pharmaceuticals, in- creasing consumer dissatisfaction with heavy-handed cost containment techniques used by managed care organizations, and the inability of pay- ers to squeeze additional savings from provider payments. This paper argues that the prevailing cost containment methods have the tendency to cause more harm to racial and ethnic minorities than to others. It might be argued, then, that one way to avoid problems result- ing from cost containment would be to eschew it as a policy goal. The point is not a trivial one. If individuals or their third-party payers wish to 1 One indication is the rate of growth in costs of two prominent purchasing cooperatives. In 2001, premiums for the Federal Employees Health Benefits Plan (FEHBP) increased by 10.5%, and costs for the California Public Employees Retirement System (CalPERS) rose by 13% (which was divided between premium increases and higher levels of patient cost sharing). These figures were obtained from http://www.opm.gov/pressrel/2000/fehb%20open%20season%202000.htm (FEHBP) and http://webmd-practice.medcast.com/Z/Channels/39/article60467 (CalPERS). 699
700 UNEQUAL TREATMENT spend more on health care and consequently, less on other things, why should they be stoppedâparticularly when it seems increasingly clear that certain new medical devices, products, and procedures can improve the quality and length of life? There are several reasons why. First, additional health care spending has significant opportunity costs. A dollar spent on health cannot be spent on other things like education, housing, or consumer goods. Second, there are various ways in which the health care market is imperfect that may lead to more spending than is desirable. Unlike other goods and services, health care services are often well insured, which insulates consumers from facing their true cost. In addition, because consumer information is often poor, people may demand medical goods and services in part be- cause of strong advertising, or because they are âinducedâ to do so by providers who have a pecuniary incentive to increase demand. Third, government now pays for almost half of U.S. health care spending. Even though the United States is now going through a period of unprecedented budget surpluses the future of social programs is nevertheless worrisome, particularly for Medicare, which faces more recipients and fewer contribu- tors when the âbaby boomâ generation retires. Finally, one of the major reasons that the number of uninsured persons continues to rise in the United States is because of health care costs. The ability of employers to offer insurance coverage, as well as workersâ ability to enroll when it is offered, is dependent on how much each has to pay for coverage. Rising costs have been shown not only to reduce employersâ ability to offer cov- erage, but also to dampen demand by employees to enroll in such cover- age when it is offered (Cooper and Schone, 1997). Thus, there are strong reasons to believe that we as a society should attempt to control health care costs. One of the major challenges is to design ways of doing so that not only preserve quality care, but also do not aggravateâand perhaps can even reduceâexisting racial and ethnic disparities in care. The purpose of this paper is to analyze how existing cost containment mechanisms may have a differential negative impact on racial and ethnic minorities. The next section provides a simple framework for categorizing cost containment strategies. The following section examines how various cost containment efforts may negatively affect racial and ethnic minorities as compared with other groups. The conclusion section discusses ways in which some of the problems raised here can perhaps be ameliorated. COST CONTAINMENT STRATEGIES There are numerous ways in which one can classify strategies aimed at controlling a nationâs health care expenses. The one adopted here is
701 IMPACT OF COST CONTAINMENT EFFORTS from Ellis and McGuire (1993), who distinguish between âsupply-sideâ and âdemand-sideâ cost sharing. According to these authors, the major feature of demand-side approaches is that âpatients must pay more in co- payments and deductibles,â whereas supply-side methods âseek to alter the incentives of health care workers to provide certain servicesâ (Ellis and McGuire, 1993, p. 135). A thirdâmanaged competitionâis also dis- cussed. Managed competition, more so than the other strategies, empha- sizes both demand- and supply-side measures for containing costs. Demand-Side Approaches In the traditional economic model, demand is paramount. Of course, to obtain a market equilibrium of price and quantity, it is necessary to consider both demand and supply. Beyond that, however, the role of supply is rather passive. If, for example, demand increases, resulting in a higher market clearing prices and higher profits, firms will increase sup- ply to meet this demand and reap these profits. In contrast, changes in supply are not supposed to influence peopleâs demand. The passive role of demand stems in large part from the economic modelâs reliance on consumer sovereigntyâthe assumption that people make better choices for themselves than others, such as government, can make for them. But many observers doubt that health care meets the necessary requirements for the proper functioning of a market. If this is the case, then relying on consumer sovereignty may not result in the best outcomes for society. Especially noteworthy are strong externalities,2 poor consumer information, the influence and market power of physicians, and the belief by many that people deserve health care irrespective of their ability to pay.3 As a result, more policy tools in health care have focused on the supply side, as described below. There are two major tools available for containing costs through de- mand. One mentioned by Ellis and McGuire (1993) is patient cost shar- ing. If people have to pay more, it is generally assumed that they use 2 Externalities exist when one person or organizationâs production or consumption has effects on others. Positive externalities imply that these activities help others; negative ex- ternalities lead to harm. Immunizations are a classic example of a positive externality. They help not only the recipient, but also others who are less likely to get a disease if more people are immunized. Free markets under-provide positive externalities because the recipient must bear the full cost even though others are benefiting. Industrial pollution is an example of a negative externality because pollution reduces the quality of life for others. Free mar- kets over-provide pollution because the producers of it do not bear its associated costs. 3 For a discussion of 15 assumptions necessary for a free market to result in the best health care system, see Rice (1998).
702 UNEQUAL TREATMENT fewer services, a finding almost always supported in the health care lit- erature. There are, in turn, two ways in which cost sharing is traditionally applied to consumers: by paying for insurance premiums, and by paying coinsurance, co-payments, and/or deductibles when they receive services. During the 1990s, the consumer expenditure towards cost sharing in job- based coverage was fairly steady in real dollars, although there was a shifting in these expenses to higher premiums and lower co-payments (Gabel et al., 2001). An instance where there is less support for the hypothesis that cost sharing reduces utilization is in the case of services provided by physi- cians after a patient has commenced a new episode of care. One of the more overlooked but intriguing findings of the RAND Health Insurance Experiment, the pre-eminent study of consumer demand for health care services, is that co-payments had a substantial impact on whether or not patients sought care for an illness, but little discernible effect on how much care they received once they sought medical attention (Manning et al., 1987; Newhouse, 1993). Apparently, it is the physician who controls re- source usage once an episode of care commences. This means that in con- sidering the role of patient cost sharing, it is important to realize that its major impact is on reducing the number of episodes of care for which medical care is sought rather than the cost of care per episode. The other tool available for containing costs through demand is giv- ing people better information. This information can pertain to particular services (e.g., informing people what services are medically appropriate) or to insurance itself (e.g., letting them know the price and quality of al- ternative insurance choices that may reflect different benefits, provider networks, etc.). The idea is to facilitate consumer sovereignty so that peoplesâ demand is informed, and therefore, most optimal for meeting their intended purposes. Providing consumers with more information has been one of the ma- jor developments in the health care services market in recent years. The main avenue has been by supplying consumers with information about the quality of alternative health plan choices, although there has also been some movement towards reporting on the quality of hospital and physi- cian groups as well. Parallel to that, there has been a vast expansion of medical information available to the lay public through the Internet, even though the accuracy of this information is, by its nature, often suspect. Whether the availability of such information does in fact lead to better consumer choices is a hotly debated topic, and the resolution is still up in the air. Some published studies have found little impact of quality infor- mation on bettering consumer choice (Chernew and Scanlon, 1998), al- though others have found a positive impact (Mukamel and Mushlin, 1998).
703 IMPACT OF COST CONTAINMENT EFFORTS Supply-Side Approaches Most tools that have been used to control health care costs focus on the supply side. These approaches, in general, are aimed at getting pro- viders such as hospitals and physicians to change their behavior, rather than focusing on the patientâs behavior. To give a few examples: â¢ Public programs, notably Medicaid, traditionally have paid physi- cians very low fees to treat patients. The resultâand in some instances, perhaps the intentâis to dissuade physicians from providing more ser- vices to program beneficiaries. â¢ Utilization review and practice guidelines are aimed at ensuring that physicians provide services that are seen as medically appropriate. â¢ Diagnosis-related groups (DRGs) and capitation try to instill in hospitals and physicians, respectively, an incentive not to over-provide services. They do this by paying a fixed sum of money for care regardless of how many services are actually performed. â¢ Supply and technology controls attempt to limit the number of hos- pitals, doctors, or capital equipment in the system as a means of control- ling overall usage. Space does not permit a detailed description of the prevalence of each of these, although it should be noted all except the last (supply and tech- nology controls) are commonly employed in the United States. To illus- trate a single example, we focus here on one of the more controversial ones: the use of capitation to pay physicians. Under capitation, the physi- cian receives a fixed amount of money per patient over a period of time such as a year, irrespective of how many services are provided. These capitation payments often include not only the services provided by pri- mary care physicians, but their referrals to hospitals and specialists as well. A positive way to view the consequences of capitation is that there is an incentive for physicians to provide preventive care but not to over- utilize marginally useful services. On the negative side, capitation may lead physicians to stint on providing useful services. Recent data on its prevalence is provided by a 1999 national survey conducted by Mathematica Policy Research, Inc. for the Medicare Payment Advisory Commission (2000). Table 1 shows how primary care physicians are paid by health plans, and Table 2 shows the same information for spe- cialists. Within the table, âwithholdsâ refers to the situation whereby some of the physicianâs remuneration is held back and paid only if certain cost containment goals are met, such as keeping down hospitalization and re- ferrals. âBonusesâ are extra payments that can be based on the meeting of individual or group utilization goals, high patient satisfaction, etc.
704 UNEQUAL TREATMENT TABLE 1. How Health Plans Pay Primary Care Physicians Northeast/ Southeast/ Mid- South All California Midwest Atlantic Central Plans Markets Markets Markets Markets Predominant payment method for primary care physicians Fee-for-service 24.7% 1.2% 23.2% 29.6% 50.9% Without witholds or bonuses 15.1% 1.2% 2.3% 10.5% 50.9% With witholds or bonuses 9.7 0.0 20.9 19.1 0.0 Capitation 61.2% 59.5% 74.3% 69.2% 41.0% Without witholds or bonuses 29.2% 30.7% 36.5% 31.6% 17.8% With witholds or bonuses 32.0 28.8 37.8 37.6 23.2 Salary 14.1% 39.4% 2.5% 1.2% 8.0% Without witholds or bonuses 13.3% 39.4% 0.0% 0.0% 8.0% With witholds or bonuses 0.8 0.0 2.5 1.2 0.0 Capitation is by far the most common method used in paying pri- mary care physicians (Table 1), with an estimated share of 61% of HMOs relying primarily on it. Fee-for-service is second with 25%, and salary is third with 14%. Withholds and bonuses are used about half of the time in capitation and about one-third of the time in fee-for-service, but are rarely used in salary arrangements. The results differ a great deal by geographic TABLE 2. How Health Plans Pay Specialists Northeast/ Southeast/ Mid- South All California Midwest Atlantic Central Plans Markets Markets Markets Markets Predominant payment method for specialists Fee-for-service (total) 75.3% 35.9% 100.0% 94.7% 80.2% Without withholding or bonuses 52.2% 23.8% 57.2% 73.0% 56.6% With witholding or bonuses 23.1 12.1 42.8 21.6 23.6 Capitation (total) 13.3% 25.1% 0.0% 5.3% 19.8% Without withholding or bonuses 7.1% 12.0% 0.0% 4.1% 10.7% With witholding or bonuses 6.2 13.1 0.0% 1.2 9.0 Salary (total) 11.4% 39.0% 0.0% 0.0% 0.0% Without withholding or bonuses 11.4% 39.0% 0.0% 0.0% 0.0% With witholding or bonuses 0.0 0.0 0.0 0.0 0.0
705 IMPACT OF COST CONTAINMENT EFFORTS area. California, for example, is far more likely to use salary to pay pri- mary care physicians, mainly due to the presence of Kaiser Permanente, a large group model HMO. Fee-for-service arrangements are almost un- heard of in California, but make up half of the arrangements in southern markets and about one-fourth in the other regions. In contrast, except in California, fee-for-service is the most common method of paying specialists (Table 2). This method accounts for 75% of the market nationally, and over 90% in much of the country. Capitation of specialists accounts for 13%, and salary, 11%. As before, salary is far more common in California than elsewhere. The presence of withholds and bo- nuses is comparable to those reported for primary care physicians. Strong financial methods for controlling costs in managed care are far more prevalent in the payment of primary care physicians. Only 15% of primary care physicians were paid on a fee-for-service without any bo- nuses or withholds, compared with 52% of specialists. The study also examined the types of performance measures used by health plans to adjust payments to primary care physicians. It found that quality measures were used most often (68%), followed by consumer sur- veys (48%), utilization and cost measures (46%), patient complaints (42%), and enrollee turnover rates (23%). On average, between 6%-10% of com- pensation was affected by physicianâs performance on these measures (Medicare Payment Advisory Commission, 2000). Managed Competition Managed competition combines both demand and supply-side incen- tives to control costs (Enthoven, 1978; Enthoven and Kronick, 1989). Un- der managed competition, health plans compete with each other for en- rollees. Payers such as employers or government provide a fixed amount of money to the enrollees to purchase insurance. The payers also provide information on alternative plan costs and measures of quality and en- rollee satisfaction. If enrollees choose a more expensive plan, they have to pay additional premiums out of pocket. These are the parts of managed competition aimed at the demand side. Health plans, in turn, need to keep their costs down to remain competitive, and one way they can do so is to pay providers in a manner that induces them to control costs. They may also use techniques such as utilization review. These are the aspects of managed competition aimed at controlling costs through the supply side. Managed competition was part of the failed Health Security Act pro- posed by the Clinton Administration, but many aspects of it have been adopted by some private payers. Most notable is the practice of offering employees a fixed contribution towards a menu of health plans. Re- search has shown that this can save payers considerable amounts of
706 UNEQUAL TREATMENT money. One problem, however, is that health plans that rely on fee-for- service medicine may obtain a sicker group of enrollees, which eventu- ally may lead to these types of plans being priced out of the market (Buchmueller, 1998). COST CONTAINMENT AND RACIAL AND ETHNIC DISPARITIES This section is divided into two parts: demand-side approaches for containing costs and supply-side approaches. Within each, I will indicate how the use of these techniques is likely to affect racial and ethnic dispari- ties in health care. Before proceeding, it is important to discuss a potential limitation of this analysis. Although I make an effort, when possible, to isolate (both conceptually and empirically) the impact of race and ethnicity from other sociodemographic characteristics, sometimes this is not possible or even desirable. Race and ethnicity are often correlated with socioeconomic sta- tus, and both of these âvariablesâ are related to various measures of health care outcomes. To anticipate an example discussed in more detail below, higher cost- sharing requirements are more of a financial burden on those with low incomes. They either result in more income being spent on services, or fewer services being purchased. Racial and ethnic minorities have, on average, lower incomes than whites, so they tend to be more adversely affected by cost sharing. Thus, race/ethnicity is not the cause of the prob- lemâlow income isâ but those in these subgroups of the population nev- ertheless bear a disproportionate burden. In other instances described below, however, race appears to be the true cause of disparities. One obvious example is the discussion of racial stereotyping on the part of some physicians. DEMAND-SIDE APPROACHES The two main demand-side approaches to containing costsâpatient cost sharing and consumer informationâare discussed in turn. Patient Cost Sharing Patient cost sharing, in the form of coinsurance, deductibles, and co- payments applied at the time of service usage, is more common in the United States than in the rest of the world. It also seems to be the cost- containment method of choice among a disproportionate number of health economists in the United States. The genesis of this belief may lie,
707 IMPACT OF COST CONTAINMENT EFFORTS in part, in the RAND Health Insurance Study, a vast social science experi- ment conducted between 1974 to 1982, and described in Newhouse (1993). One of the main findings of the study was that consumer demand was indeed sensitive to out-of-pocket costs in that people who had to pay more for services at point-of-service were considerably less likely to seek medi- cal care. Furthermore, in general, the researchers found little evidence that lower cost sharing resulted in better health outcomes (although there were some positive effects, particularly for some procedures undergone primarily by blacks, as discussed below). These results have been used in conjunction with traditional economic theory to demonstrate that higher cost sharing improves a countriesâ so- cial welfare (Feldman and Dowd, 1991). The point was made forcefully by Manning and colleagues (1987), who asked whether the cost of the RAND Health Insurance Study (over $200 million in todayâs dollars) was worth it. [W]e believe that the benefits of this particular experiment greatly exceeded the costs. . . . Between 1982 and 1984, there was a remarkable increase in initial cost sharing in the United States, at least for hospital services. For example, the number of major companies with first-dollar charges for hospital care rose from 30 to 63 percent in those two years, and the number of such firms with an annual deductible of $200 per person or more rose from 4 to 21 percent. Al- though it is impossible to know how much of this change can be attributed to the experimental results, the initial findings of the experiment were published . . . and given wide publicity in both the general and trade press. In certain instances a direct link between changes in cost sharing and the experimental results can be made (Manning et al., 1987, p. 272). Because the experiment showed that increased patient cost-sharing reduced medical expenditures, the researchers estimated that under the most optimistic scenario, the eight-year experiment could have paid for itself in a week. But why is the lower utilization that results from cost sharing sup- posed to make society better off? It is because the extra services that people use when they have full insurance are assumed to bring about less in the way of benefits. Economic theory posits that people will buy some- thing up until the point that the benefit of the last unit purchased equals the cost. With full insurance, the money price of services is zero; it is therefore assumed that the last service consumed has almost zero value. When this low benefit is compared with the cost of production, there is a âwelfare lossâ associated with the production and consumption of the serviceâit costs more to produce than the person (and therefore, society) gains.
708 UNEQUAL TREATMENT This theory has two key implications: society will be better off if people pay higher cost sharing amounts4 and cost sharing requirements should be highest for services that are more price sensitive. The latter argument is a bit more complicated. If the possession of insurance leads to a large increase in utilization, then welfare loss will be larger because more services will be purchased where the additional costs exceed extra benefits. Thus, one would improve social welfare by assessing higher patient coinsurance rates for such services, thereby reducing usage. In contrast, if utilization rates are not very sensitive to the possession of in- surance, then there is little welfare loss, and less need to charge high co- insurance rates. These are strong policy recommendations, and it is important to un- derstand their basis. The traditional economic model assumes that people make well-informed choices that maximize their own utility. Thus, in making the decision to buy or not buy a service, they are implicitly evalu- ating the utility or gain they would receive from the service against its cost or co-payment. It should be stressed that not all economists âbuy intoâ the applica- tion of this theory to health care. For example, Ellis and McGuire (1993) write, â[We] are skeptical that the observed demand can be interpreted as reflecting âsocially efficientâ consumption, [so] we interpret the demand curve in a more limited way, as an empirical relationship between the degree of cost sharing and quantity of use demanded by the patientâ (p. 142). Evans (1984) notes: âThe welfare burden is minimized when there is no insurance at allâ (p. 49). And if one takes this reasoning very far, Reinhardt (1992) points out that this logic will always find that the coun- try with higher patient cost-sharing requirements will have the more effi- cient health system. Thus, the U.S. system would be deemed more effi- cient than the Canadian system or any of a number of European systems, not because of a comparison of outcomes to costs, but rather simply from the fact that the U.S. imposes higher patient cost sharing, which in turn reduces utilization. The implications of relying on patient cost sharing on racial and eth- nic minorities are extremely important. Simply put, cost sharing results in de facto discrimination, for several reasons. First, in cases where racial and ethnic minorities are not deterred by the requirements, cost sharing 4 It is actually a bit more complicated. Excess insurance is assumed to lower social wel- fare, but the existence of insurance also raises welfare because ârisk averseâ individuals want protection against having to face catastrophically high medical expenses. The Feld- man/Dowd study concludes, however, that the welfare loss from excess insurance far ex- ceeds the additional utility conveyed by owning insurance.
709 IMPACT OF COST CONTAINMENT EFFORTS constitutes a much higher average burden. In 1998, median household income for whites was a about $41,000, compared with $25,000 for blacks and $28,000 for Hispanics (U.S. Census Bureau, 2000, p. 466). Second, certain racial and ethnic minority groups are in poorer health. As a result, they have a greater need for services. If they use them, then the problems associated with high cost sharing and low incomes are mag- nified for minorities compared with whites, who are both healthier and wealthier. To give some examples among the total population, 8% of whites rate their health as fair or poor, compared with 16% of blacks and 13% of Hispanics. Age-adjusted hypertension rates are about 50% higher among blacks than whites. Both blacks and Hispanics have about twice the rate of untreated dental caries as whites (U.S. National Center for Health Statistics, 2000, pp. 232, 234, 270). Among Medicare beneficiaries, 42% of black beneficiaries, and 44% of Hispanics rate their health as fair or poor, compared with 25% of whites. Similarly, members of both minority groups are twice as likely to have diabetes. Two-thirds of black Medicare beneficiaries have hypertension, compared with half of whites (Gornick, 2000). Third, in many cases cost sharing is a deterrent to necessary service usage. Thus, racial and ethnic minorities do not receive the care they need, in part because they simply canât afford the costs. The RAND Health Insurance Experiment did find some instances in which lower cost shar- ing improved health status. Some of these included: â¢ Low-income families at elevated risk benefited the most from free care. The reduction in diastolic blood pressure among lower-income per- sons who were judged to be at an elevated risk for hypertension was 3.3 mm Hg, compared with only 0.4 mm Hg for similar people with higher incomes (Brook et al., 1983). â¢ Low-income persons in poor health who were given free care had the largest reduction in serious symptoms (Shapiro et al., 1986). â¢ Among children of poor families who were at the highest risk, those with free care were less likely to have anemia than those in the cost-shar- ing plans (Valdez, 1986). Gornick (2000) has also shown the large disparities in service usage between whites and racial/ethnic Medicare beneficiaries, which are dou- bly of concern because these numbers do not adjust for the poorer health status of the latter. Blacks use 82% as many office visits and 77% as many specialist services as whites, and are only half as likely to get flu shots. In contrast, they are far more likely to get services that tend to result from seeking care too late. For example, amputations of lower limbs are more than triple among blacks compared with whites.
710 UNEQUAL TREATMENT It is not possible to know how many of these disparities are due to price; indeed, there are racial and ethnic disparities in health care utili- zation, as well as differences by social class in countries that have com- prehensive health insurance. In the case of Medicare, though, the evi- dence of a price effect is strong. We know from the RAND Health Insurance Experiment that cost sharing matters, and other studies have also shown that having supplemental insurance is a major determinant of service utilization as well (McCall et al., 1991; Ettner, 1997). Minori- ties are far less likely to have supplemental insurance. Whereas only 9% of whites on Medicare lack any form of supplementation, the figures for blacks and Latinos are 27% and 16%, respectively (Pourat et al., 2000). Finally, racial and ethnic minorities are hurt by cost sharing because they can benefit most from preventive careâdue to their worse health status and lower use of such care. As noted above, the traditional theory posits that cost sharing should be highest for services that are most price sensitive, and indeed, the RAND Health Insurance Experiment found pre- vention and dentistry to be among the most price-sensitive services (Newhouse, 1993).5 Application of this theory would therefore discour- age usage of preventive care, an area in which racial and ethnic minorities have both the greatest need and ability to benefit. Of all of the cost-containment methods reviewed in this section, pa- tient cost sharing is the one I believe is most problematic for racial and ethnic minorities. The reasons are best summarized by Evans and col- leagues (1993): The primary effect of substituting user fees for tax finance is cost shiftingâ the transfer of the burden of paying for health care from taxpayers to users of care. . . . [P]eople pay taxes in rough proportion to their incomes, and use health care in rough proportion to their health status or need for care. The relationships are not exact, but in general sicker people use more health care, and richer people pay more taxes. It follows that when health care is paid for from taxes, people with higher incomes pay a larger share of the total cost; when it is paid for by the users, sick people pay a larger share. . . . Whether one is a gainer or loser, then, depends upon where one is located in the distribution of both income . . . and health. . . . In general, a shift to more user fee financing redistributes net income . . . from lower to higher income people, and from sicker to healthier people. The wealthy and healthy gain, the poor and sick lose (Evans et al., 1993, p. 4). The other form of patient cost-sharing is the premiums paid by indi- viduals and families for health insurance. There has been a great deal of research indicating that premiums affect consumers. They do so in two ways. First, as premiums rise, individuals are less likely to purchase 5 This was true in the 25-95% patient coinsurance range but not the 0-25% range.
711 IMPACT OF COST CONTAINMENT EFFORTS health insurance policies offered by their employers (Cooper and Schone, 1997; Kronick and Gilmer, 1999). Second, there is much research to indi- cate that choice of health plan is extremely responsive to premiums (Cut- ler and Reber, 1998; Buchmueller, 1998). Because racial and ethnic mi- norities have, on average, lower incomes than whites, we would expect that their enrollment in insurance plans would also be lower. In 1999, 75% of whites had job-based health insurance coverage, while this was true of only 58% of blacks and 47% of Latinos (Brown et al., 2001). The second issue is whether higher premiums may force a dispropor- tionate number of racial and ethnic minorities to choose health plans that are of lower quality. This is an intriguing issue and will be discussed next. Consumer Information The second demand-side approach to cost containment is relying on consumer information. As noted earlier, this is manifested mainly by com- parisons of the benefits, costs, and quality of health plans, as well as con- sumer satisfaction. Indeed, the major assumption of most âpro-competi- tiveâ health initiatives in both the private and public sectors is that consumers are capable of using this information to make good plan choices. Whether this strategy works well is a subject of much debate. The fed- eral government has invested considerable funding and research firms have invested considerable energy in formulating ways of presenting and dis- seminating such information through such initiatives as the Consumer As- sessment of Health Plans (CAHPS, see http://www.ahcpr.gov/qual/ cahpfact.htm) sponsored by the Agency for Healthcare Research and Qual- ity). Furthermore, there have been major private initiatives such as the Health Plan Employer Data and Information Set (HEDIS) developed by the National Committee for Quality Assurance (see http://www. ncqa.org/ Programs/HEDIS). In addition, organizations such as the Pacific Business Group on Health have published extensive data on HEDIS and other mea- sures of quality and satisfaction (see http://www.pbgh. org). Although these efforts are admirable, the reliance on consumer infor- mation to make health plan choices disproportionately harms certain ra- cial and ethnic minorities in relation to whites. The groups that are disad- vantaged are those who have lower levels of education, and especially individuals whose primary language is not English. In 1999, 26% of whites age 25 had a four-year college degree, compared with 16% of blacks and 11% of Hispanics (U.S. Census Bureau, 2000, pp. 43, 46). Although it is difficult to find comparable data for those whose primary language is not English, the rates are undoubtedly lower. Research by Hibbard and col-
712 UNEQUAL TREATMENT leagues (2001) shows a strong relationship between more education and understanding comparative information about health plans up until age 80. One might argue that even though racial and ethnic minorities are disadvantaged by strategies that rely on good consumer information, there is little to be done except to improve education (both in general as well as education targeted to understanding things like report cards). This is too narrow a viewpoint. Few other countries rely on consumers to make choices among competing health plans. It is quite possible to organize a health care system whereby everyone has the same health plans and cost containment efforts are focused on things other than demand-side strate- gies. We evaluate some of these in the next section. SUPPLY-SIDE APPROACHES This section is divided into four parts, each reflecting a different set of supply-side approaches for containing costs: low physician fees; capita- tion and DRGs; utilization review and practice guidelines; and supply, technology, and expenditures controls. Low Physician Fees Traditionally, states have paid physicians very poorly for treating Medicaid patients. Economists have constructed conceptual models to help explain how physicians would be expected to respond. One such model is that of a âprice discriminating monopolist,â in which physicians can receive different amounts of revenue from different groups of patients (Sloan et al., 1978). One implication of the model is that physicians will treat the most lucrative patients first, and once that market is exhausted, treat others, as well. They will not tend to treat patients whose costs ex- ceed revenues. Thus, we would expect that Medicaid patients might have trouble finding a doctor willing to treat them. In addition, different phy- sicians have different costs and face different levels of demand. Those physicians who are better trained, more specialized, etc. tend to have higher costs, especially when including the value of their time. The model would predict that these more costly physicians would also tend to avoid low-revenue patients such as those covered by Medicaid. Most studies have shown just such effects. Medicaid patients are less likely to have access to physicians in their outpatient practices, as well as to specialists and others who may be perceived as more qualified (Mitchell, 1991). This, in turn, results in differential access problems for minorities. Whereas 6% of whites have Medicaid coverage, 19% of blacks and 14% of Hispanics do. Blacks are more than 30% more likely to use an
713 IMPACT OF COST CONTAINMENT EFFORTS emergency room than whites because it is less likely that their regular place of care is a physician office (U.S. National Center for Health Statis- tics, 2000, pp. 340, 267). A way to avoid this problem is to have all insurers pay the same amount to physicians, which is known as an âall-payer system.â Such a system exists in several European countries, as well as in Japan. When each patient has equal value to a physician, the latter no longer has a financial incentive to choose one type of patient over another. However, he or she may still prefer certain patients based on non-financial criteria, an issue explored next. Capitation and DRGs These strategies are combined because the incentives are similar. Under DRGs, hospitals are paid an amount of money for an in-patient stay that is, in most cases, completely unrelated to how many resources are used to treat the patient. Under capitation, physicians receive a fixed amount of money per patient per year, again largely unrelated to subse- quent resource usage. It might be argued that DRGs do not affect physi- cian decision making because they apply to how the hospitalârather than how the doctorâis paid. This ignores the fact that hospitals use a number of strategies to make physicians cognizant of the hospitalâs financial in- centive. These range from informing the doctor how long the patient has stayed in relation to the average for that DRGs, all the way to withdraw- ing privileges to practice in the hospital. In and of themselves, capitation and DRGs should not favor one ra- cial or ethnic group over another. Hospitals receive the same DRG pay- ment for a white and a Latino patient; doctors get the same capitation fee for whites and blacks (although different insurers may pay differing amounts, and have different racial/ethnic mixes of enrollees). Rather, the issue is more subtle. It is possible that the financial pressure exerted by these payment methods will result in physicians cutting back their ser- vices differentiallyâand that this differentiation is related to race and ethnicity. Consider the case of capitation. A physician who is capitated has an incentive to enroll more patients in his or her practice, and under certain schemes, may also have an incentive to control the number of hospital and specialist referrals. One scarce resource is the physicianâs time. Thus, there is a potential incentive to do less for the patient. It is important to recognize that research on this topic has yet to reach any consensus. Reviews of the literature by Miller and Luft (1997, 2001), one of which includes literature up to the year 2000, lists just as many studies finding that HMOs provide better quality care and worse quality
714 UNEQUAL TREATMENT care. However, this does not provide direct evidence, because here we are specifically considering the impact of capitating the physician. Unfor- tunately, to my knowledge there have been no studies that have directly assessed the impact of physician financial incentives on the actual quality of care delivered. Let us suppose, though, that a physician does feel the need to control how much time he or she spends with the patient and/or number of refer- rals. How could this manifest itself into racial and ethnic disparities? There are at least three possibilities. First, there could be overt discrimi- nation, with the physician willfully favoring his or her racial/ethnic group over others. Although certainly possible, this has not been considered the major driving factor by most researchers. A second and somewhat related explanation is that physicians stereo- type minority patients. One particularly interesting study of this possibil- ity was conducted by van Ryn and Burke (2000), who surveyed physi- ciansâ attitudes after patients received an angiogram in 10 New York state hospitals. They were asked questions about perceptions on such things as the patientâs intelligence and education, pleasantness, self-control, and rationality. They also were asked to rate patients on likely compliance with medical care, drug and alcohol use and lifestyle, as well as the likeli- hood that they would sue. In general, physicians rated whites higher than blacks on most di- mensionsâeven after the researchers controlled for the appropriate vari- ables. Overall, they found that black patients: âwere more likely to be seen as at risk for noncompliance with cardiac rehabili- tation, substance abuse, and having inadequate social support. In additional, physicians rated Black patients as less intelligent than white patients, even when patient sex, age, income, and education were controlled. Physicians also report less affiliative feelings towards black patientsâ (van Ryn and Burke, 2000, p. 821). The authors posit an explanation for these results. It may be that physicians have stereotypes about racial and ethnic minorities, and apply these generalizations to individuals in the group. They write: âPhysicians may fail to correctly incorporate individual diagnostic data, in- stead being swayed by their beliefs regarding the probabilities of individuals in a socio-demographic category having a given characteristic. In this way, physi- ciansâ understanding of epidemiological evidence regarding population-based likelihoods may function as stereotypes, and be applied to assessments and perceptions of individuals regardless of actual individual characteristics. It is possible that this is especially likely when population-based statistics are con-
715 IMPACT OF COST CONTAINMENT EFFORTS sistent with dominant biases. . . . This suggests that physicians are applying general race differences to their impressions of individuals patients and failing to incorporate disconfirming individual informationâ (van Ryn and Burke, 2000, pp. 822-823). Interestingly, this could result in a circle. Because physicians are pressed for time under capitation, stereotyping leads to less care to racial and ethnic minorities. In this regard, the study found physicians spent more time with white than black patients. These patients, in turn, may not be receiving as good advice on care, and furthermore, may feel alienated from the physician. As a result, they may reveal less information to the physician or seek care less often. A third and related reason that racial and ethnic minorities might fare worse in a DRG or capitated environment is through âstatistical discrimination.â This concept has been applied to health care by Balsa and McGuire (2001). In essence, the authors argue that physiciansâ deci- sions result from the inability to interpret information about the patient. If they feel they know less about a patientâs symptoms or needs, they will be less certain that a particular course of treatment is appropriate. Under strong financial pressures such as those generated by DRGs and capitation, physicians may therefore favor the patients for whom infor- mation is less ambiguous. To illustrate, they note that âa white male doctor might have an easier time interpreting the signal, âdoc, it really hurtsâ from a white male patient than from a black woman patient, or from a Latino woman patientâ (Balsa and McGuire, 2001, p. 1). Indeed one would expect particularly large effects among patients whose na- tive tongue is not English. I have argued that there are several reasons to believe that the incen- tives of managed care could differentially harm the care provided to ra- cial and ethnic minorities. An interesting test of this hypothesis was con- ducted by Tai-Seale and colleagues (2001), who focused on a group of Medicaid beneficiaries who were forced into HMOs. From this natural experiment, they found that compared with whites who were forced into HMOs, blacks âexperienced declines in relative use of physician services (among both adults and children) and an increase in relative use of the ER among childrenâ (Tai-Seale et al., 2001, p. 56). They further note that, âwhile the reduction in service use in itself is not necessarily a sign of poor access or discrimination, it is the difference in the decline of relative service use between African-American and white beneficiaries that war- rants further investigationâ (p. 57). The findings therefore are consistent with the belief that financial incentives that encourage physicians to re- duce resource usage under managed care differentially harm racial and ethnic minorities.
716 UNEQUAL TREATMENT Utilization Review and Practice Guidelines Utilization review is a practice engaged in, usually but not always by health care payers, to monitor whether a particular service is appropriate for a patient in a specific instance. It can be done prospectively (requiring that a hospital admission be approved in advance), concurrently (during a hospital stay), or retrospectively (reviewing services already rendered by a physician as part of a âpractice profileâ). Traditionally, utilization review has been thought of as a cost containment method because it was originally applied to fee-for-service medicine, in which there often is a financial incentive to over-provide. But it can also be used in a capitated environment to ensure that enough services are being delivered. In contrast, practice guidelines are designed to reduce inappropriate variation in the provision of medical services (Wennberg and Gittelsohn, 1982; Rutledge, 1998). Their implementation can lead to the provision or more or fewer services, depending on how prevailing practice patterns compare to the norms recommended in the guidelines. Practice guidelines should result in a reduction of racial and ethnic disparities. Since the guidelines are agnostic with regard to race and ethnicity, following those guidelines should result in standard care across these groups. A potential problem arises when guidelines are used not as a way of enhancing quality, but as a way to reduce costs. In such in- stances, they begin to resemble utilization review. Many of the problems described in the previous section also apply here. Suppose that a multispecialty group practice has contracts with a number of network-model HMOs and wishes to monitor the resource us- age of its physicians. Further suppose that, as a way of controlling costs, it uses practice guidelines but alerts physicians that they are deviating from the guidelines only when they are performing more services than recommended in those guidelines. It does not let them know when their utilization is lower than the specified levels (unless they are so deviant that there is a risk of malpractice). Under this scenario, physicians are likely to feel pressured to control their provision of and/or recommendations for additional services. If this is the case, then for the reasons discussed under the capitation/DRG section, they are likely on average to provide relatively fewer services or recommendations for services to racial and ethnic minorities. It is worth noting a particular study on physiciansâ recommendations for managing chest pain conducted by Schulman and colleagues (1999). Physicians at national meetings were recruited to participate in a study in which they viewed videos of patients and were then asked to assess whether they would recommend cardiac catheterization. The patients on the videos were actually actors, all of whom were directed to follow scripts
717 IMPACT OF COST CONTAINMENT EFFORTS in identical manners. In addition, each of these âpatientsâ was identified as having the same occupation and health history. Each physician partici- pant was randomly assigned to see and hear one such video. The study found that both women and blacks were much less likely to be recom- mended for cardiac catheterization than men and whites. The odds ratios for both were 0.6, and the odds ratio for black women (compared with white men) was 0.4. This study seems especially relevant because it shows how physicians are likely to ration when they are under pressureâas a result of utiliza- tion review or the inappropriate application of practice guidelinesâto keep costs and referrals down. These supply-side techniques therefore also have the potential to aggravate racial and ethnic disparities. Supply, Technology, and Expenditure Controls It is difficult to generalize about other countries, all of which have different health care systems and have relied on different methods of controlling costs. Nevertheless, if one were to risk doing so, it might be concluded that they rely much more heavily than the United States on âmacro-levelâ supply-side strategies. Rather than looking at the particu- lar services delivered,6 they tend to stress (in varying degrees) system- wide policies such as regulating the supply of hospital beds, physicians, specialists, and medical technologies. One of the most common methods of cost containment, especially for hospital and physician services, is the use of some kind of global budgets. These âtend to be prospectively set caps on spending for some portion of the health care industryâ (Wolfe and Moran, 1993, p. 55). The exact mean- ing, however, varies from country to country. In some countries, such as Canada, hospitals receive an annual global budget to cover their entire operating budget. In Germany, there are regional budgets for different types of physician services. A survey of nine European countries found that all used some form of global budgeting. Most studies of global bud- geting have found that global budgets do help control spending (Wolfe and Moran, 1993; U.S. General Accounting Office, 1991; Abel-Smith, 1992). It would seem that the potential for racial and ethnic disparities could still exist under these macro-level policies, just as they did under the more micro-level supply-side strategies employed in the United States. Indeed, even countries with universal coverage and low patient cost sharing re- 6 This is not the case everywhere. Germany, for example, historically has compared indi- vidual physician utilization profiles to those of other physicians and, when there is a large deviation, informed the physician and sometimes even withheld reimbursement.
718 UNEQUAL TREATMENT quirements have been unable to equalize access to care. To give some hypothetical examples consistent with the literature cited earlier: â¢ If the number of specialists are controlled, as is the case in many countries, then those that are in practice will experience excess demand for their services and may ration according to race and ethnicity. â¢ If the number of hospital beds or medical technologies are con- trolled, then these scarce resources may be rationed similarly. â¢ If there is a global budget on various sectors of the health system, each of these sectors will have to make its own allocation decisions. Again, there is little assurance that racial and ethnic disparities will be avoided. Thus, whatever the merits or demerits of the cost-containment sys- tems used in other countries, there is no assurance that importing them to the United States would be any less discriminatory than other supply- side strategies currently being used in this country. Perhaps the major lesson from other countries is that most do not rely on demand-side poli- cies, which were earlier shown to have the potential to create large dis- parities. This theme is further examined below. Conclusions Unfortunately, there are no easy answers to questions concerning how we can reduce racial and ethnic disparities in health care. This paper has shown that both demand- and supply-side cost containment methods have strong potential for aggravating existing inequities. In some ways, the demand-side problem is more vexing. The core notion is that goods should be allotted according to ability to pay. Those who lack that ability therefore, will either use fewer services or spend much more of their income using them. Since racial and ethnic minorities have lower average incomes and, for most indicators, worse health status, demand-side cost containment policies tend to hit particularly hard. There are no easy answers on the supply side either. Most papers on racial and ethnic disparities that show some form of discrimination on the part of physiciansâhowever unintentionalâsuggest that they be told of current disparities in treatment as part of a broader effort to make them more âculturally competentâ [see, for example, Brach and Fraser (2000) and van Ryn and Burke (2000)]. Needless to say, changes in deep-seated behavior of this kind will not take place overnight. I would posit that the reliance on both demand- and supply-side cost containment strategies, particularly through the application of managed competition, makes the problems more acute in the United States than in
719 IMPACT OF COST CONTAINMENT EFFORTS other countries. Managed competition relies on nearly all of the problem- atic cost-containment strategies outlined in this paper: â¢ Allocating services on the basis of ability to pay. â¢ Relying on consumerâs understanding of complicated comparative information when choosing health plans. â¢ Pressuring providers through payment mechanisms such as DRGs and capitation. â¢ Monitoring the provision of services through a variety of utiliza- tion review mechanisms. As noted, cost-containment methods used in others countries, which focus almost entirely on the supply side, do not offer a panacea for ending racial and ethnic disparities in health care. But they do shun the demand- side policies that have been embraced in the United States. Moving away from such policies is one tangible thing that the United States could do. It really comes down to an issue of fairness. To ensure that individu- als who are at a disadvantage have an equal probability of attaining good health, it is necessary to redistribute resources from those who have been more fortunate. Relying on ability-to-pay to allocate health care services, as noted earlier by Evans and colleagues (1993), does the opposite. Thus, while striving to increase the sensitivity of health care providers to exist- ing inequities, we must not put further barriers in the way of racial and ethnic minorities receiving needed health care services. REFERENCES Abel-Smith, B. 1992. âCost Containment and New Priorities in the European Community.â Milbank Quarterly 70(3):393â422. Anderson, G., Hussey, P.S. 2001. Comparing health system performance in OECD coun- tries. Health Affairs 20(3):219-232. Balsa, A.I., and McGuire, T.C. 2001. Statistical discrimination in health care. Journal of Health Economics 20:881-907. Brach, C., Fraser, I. 2000. Can cultural competency reduce racial and ethnic health dispari- ties? A review and conceptual model. Medical Care Research and Review 57 (Supplement 1):181-217. Brook, R.H., et al. 1983. Does free care improve adultsâ health? New England Journal of Medicine 309(23):1426â34. Brown, E.R., Ponce, N., Rice, T. 2001. The state of health insurance in California: Re- cent trends, future prospects. Los Angeles, CA: UCLA Center for Health Policy Research. Buchmueller, T.C. 1998. Does a fixed-dollar premium contribution lower spending? Health Affairs 17(6):228-235. Chernew, M., Scanlon, D.P. 1998. Health plan report cards and insurance choice. Inquiry 35(1):9-22.
720 UNEQUAL TREATMENT Cooper, B.F., Schone, B.S. 1997. More offers, fewer takers for employment-based health insurance: 1987 and 1996. Health Affairs 16(6):142-149. Cutler, D.M., Reber, S.J. 1998. Paying for health insurance: The trade-off between competi- tion and adverse selection. Quarterly Journal of Economics 433-466. Ellis, R.P., McGuire, T.G. 1993. Supply-side and demand-side cost sharing in health care. Journal of Economic Perspectives 4:135-151. Enthoven, A. 1978. Consumer choice health plan. New England Journal of Medicine 298 (12): 650â58 & 298(13):709-720. Enthoven, A., Kronick, R. 1989. A consumer-choice health plan for the 1990s. New England Journal of Medicine 320(1):29â37. Ettner, S.L. 1997. Adverse selection and the purchase of Medigap insurance. Journal of Health Economics 16(5):499-624. Evans, R. G. 1984. Strained Mercy. Toronto, Ontario: Butterworth. Feldman, R., Dowd, B. 1991. A new estimate of the welfare loss of excess health insurance. American Economic Review 81(1):297-301. Gabel, J.R., Ginsburg, P.B., Pickreign, J.D., et al. 2001. Trends in out-of-pocket spending by insured American workers, 1990-1997. Health Affairs 20(2):47-56. Gornick, M.E. 2000. Vulnerable populations and Medicare services. New York: Century Founda- tion Press. Hibbard, J.H., Slovic, F., Peters, E., et al. 2001. Is the informed-choice policy approach appropriate for Medicare beneficiaries? Health Affairs 20(3):199-203. Kronick, R., Gilmer, T. 1999. Explaining the decline in health insurance coverage, 1979- 1995. Health Affairs 18(2):30-47. Manning, W., Newhouse, J.P., Duan, N. et al. 1987. Health insurance and the demand for medical care: Evidence from a randomized experiment. American Economic Review 77(3): 251-277. McCall, N., Rice, T., Boismier, J. et al. 2001. Private health insurance and medical care utili- zation: evidence from the Medicare population. Inquiry 28:276-287. Medicare Payment Advisory Commission. 2000. Health plansâ selection and payment of health care providers, 1999. Washington, DC: Medicare Payment Advisory Commis- sion. Miller, R.H., Luft, H.S. 1987. Does managed care lead to better or worse quality of care? Health Affairs 16(5):7-25. Miller, R.H., Luft, H.S. 2001. HMO plan performance update: An analysis of recently published literature (1997-2000). Paper presented at the Eighth Princeton Conference on The Future of Managed Care, Princeton, NJ. Mitchell, J.B. 1991. Physician participation in Medicaid revisited. Medical Care 29(7):645- 653. Mukamel, D.B., Mushlin, A.I. 1998. Quality of care information makes a difference. Medical Care 36(7):945-954. Newhouse, J.P. and the Insurance Experiment Group. 1993. Free for all? Lessons from the RAND health insurance experiment. Cambridge, MA: Harvard University Press. Pourat, N., Rice, T., Kominski, G., et al. 2000. Socioeconomic differences in Medicare supple- mental coverage. Health Affairs 19(5):186-196. Reinhardt, U.E. 1992. Reflections on the meaning of efficiency: Can efficiency be separated from equity? Yale Law & Policy Review 10:302â15. Rice, T. 1998. The Eonomics of Health Reconsidered. Chicago: Health Administration Press. Rutledge, R. 1998. An analysis of 25 Milliman & Robertson guidelines for surgery. Annals of Surgery 228(4):579-587.
721 IMPACT OF COST CONTAINMENT EFFORTS Schulman, K.A., Berlin, J.A., Harless, W., et al. 1999. The effect of race and sex on physiciansâ recommendations for cardiac catheterization. New England Journal of Medicine 340(8): 618-626. Shapiro, M. F., et al. 1986. Effects of cost sharing on seeking care for serious and minor symptoms. Annals of Internal Medicine 104(2):246â51. Sloan, F.J., Cromwell, J, Mitchell, J. 1978. Private physicians and public program. Lexington, MA: D.C. Heath. Tai-Seale, M., Freund, D., and LoSasso, A. 2001. Racial disparities in service use among Medicaid beneficiaries after mandatory enrollment in managed care: A difference-in- differences approach. Inquiry 38:49-59. U.S. Census Bureau. 2000. Statistical Abstract of the United States: 2000. http://www. census.gov/prod/2001pubs/statab/sec14.pdf. U.S. General Accounting Office. 1991. Canadian Health Insurance: Lessons for the United States. Washington, DC: General Accounting Office. U.S. National Center or Health Statistics. 2000. Health, United States, 2000. http:// www.cdc.gov/nchs/data/hus00.pdf. Valdez, R. O. 1986. The Effects of Cost Sharing on the Health of Children. Santa Monica, CA: RAND Corporation. van Ryn, M., Burke, J. 2000. The effect of patient race and socio-economic status on physi- ciansâ perceptions of patients. Social Science & Medicine 50:813-828. Wennberg, J., Gittelsohn, A. 1982. Variations in medical care among small areas. Scientific American (April):122-123. Wolfe, P. R., and D. W. Moran. 1993. Global Budgeting in the OECD Countries. Health Care Financing Review 14(3):55â76.