Medicare and Medicaid: Conflicting Incentives for Long-Term Care

Abstract

The structure of Medicare and Medicaid creates conflicting incentives regarding dually eligible beneficiaries without coordinating their care. Both Medicare and Medicaid have an interest in limiting their costs, and neither has an incentive to take responsibility for the management or quality of care. Examples of misaligned incentives are Medicare's cost-sharing rules, cost shifting within home health care and nursing homes, and cost shifting across chronic and acute care settings. Several policy initiatives—capitation, pay-for-performance, and the shift of the dually eligible population's Medicaid costs to the federal government—may address these conflicting incentives, but all have strengths and weaknesses. With the aging baby boom generation and projected federal and state budget shortfalls, this issue will be a continuing focus of policymakers in the coming decades.

The presence of multiple payers in health care is known to introduce conflicting incentives for providers, which may have negative implications for cost containment, service delivery, and the quality of care (Glazer and McGuire 2002). The fundamental issue is that the actions of one payer may affect the costs and outcomes of patients covered by other payers. These “external” costs and benefits can occur both within and across health care settings, and payers have little incentive to incorporate them into payment and coverage decisions. As a result, the behaviors of health care payers—even public payers—often deviate substantially from the social optimum.

This observation is particularly relevant to the coverage of acute and long-term care services. The federally run Medicare program provides insurance benefits for virtually all individuals aged sixty-five and older, regardless of income, and for younger people with disabilities two years after they qualify for Social Security's disability benefit. Medicaid, a state-run program jointly funded by the state and federal governments, offers coverage for its low-income enrollees that supplements Medicare coverage. Many persons who are dually eligible for both Medicare and Medicaid require both extensive acute and long-term care services. For example, 22 percent of the dually eligible population resides in nursing homes, compared with 2 percent of other Medicare beneficiaries (Kaiser Commission on Medicaid and the Uninsured 2004b). However, because Medicare covers relatively few long-term care services, Medicaid must cover the bulk of these expenditures for dual eligibles.

Given the bifurcated coverage of acute and long-term care under Medicare and Medicaid, neither program has an incentive to internalize the risks and benefits of those of its actions that pertain to the other program. Each program has a narrow interest in limiting its share of costs, and neither program has an incentive to take responsibility for the management or quality of care. This article analyzes the tension between Medicaid and Medicare in the coverage of acute and long-term care services. After a brief review of the Medicaid and Medicare programs, I discuss the lack of coordination between Medicaid and Medicare in regard to cost sharing, cost shifting within health care settings, and cost shifting across health care settings. I then look at ways of addressing the conflicts between Medicaid and Medicare. Policy options include capitation, pay-for-performance, and “federalization,” in which the federal government would assume Medicaid's costs for the dually eligible population. I conclude with a discussion of the next steps for the policy and research communities.

Medicare and Medicaid Coverage

Eligibility for Medicare is based on working for at least forty quarters (ten years) in Medicare-qualifying employment. Once a person has met this requirement, both the individual and spouse are eligible for Medicare at age sixty-five. Younger workers and their dependents also qualify if they have been receiving federal disability insurance for two years or have end-stage renal disease. Individuals with work histories of less than forty quarters can buy into Medicare Part A (hospital insurance) by paying a monthly premium. Medicaid can buy Part A coverage for Medicaid beneficiaries who do not meet the forty-quarters test, as well as Part B coverage for physicians' services. Medicare's benefits include inpatient and outpatient hospital stays, physicians' fees, prescription drugs, diagnostic laboratory fees, and other professional medical services. Medicare, however, covers only limited long-term care services, such as skilled nursing facility (SNF) care and skilled home health care for enrollees who meet various conditions. Although Medicare provides health insurance for elderly and disabled individuals, it was never intended to be a comprehensive benefit package. On average, Medicare pays just over half of each enrollee's health care costs (excluding long-term care) (Centers for Medicare and Medicaid Services 2002). These uncovered expenses must be paid out-of-pocket by the enrollee or by Medicaid, supplemental insurance, or other sources.

Medicare beneficiaries who meet Medicaid's (low) income and resource eligibility standards may become dually eligible. Under federal rules, most states are required to offer Medicaid coverage to recipients of the Supplemental Security Income (SSI) program (Bruen, Wiener, and Thomas 2003). In 2005, the SSI income limit for an unmarried individual was $564, and the asset limit was $2,000. But many states' Medicaid programs cover elderly people who have incomes up to 100 percent of the federal poverty level and assets that do not exceed the SSI threshold. In regard to income, the states have adopted two broad sets of rules that expand eligibility: “medically needy” programs and special income rules. If an individual's income exceeds the state's income test, medically needy programs permit that person to subtract medical and long-term care expenses from his or her income in calculating Medicaid eligibility. Other states have enacted special income rules for people in nursing homes and in home- and community-based services (HCBS) waiver programs, which extend eligibility up to 300 percent of the SSI income limit.

Another important feature of Medicaid's eligibility rules is that a person's home is excluded from his or her total assets when calculating Medicaid nursing home eligibility, although the value of this housing exemption was recently capped as part of the 2005 Deficit Reduction Act. The determination of countable assets for Medicaid nursing home coverage is quite complex, but in addition to the house, the first $2,000 of household goods or personal effects, a car used to obtain medical treatment, and certain burial funds are excluded from consideration as well.

For dual eligibles with full Medicaid benefits, Medicaid typically pays for services that Medicare does not cover, such as transportation, dental and vision, and wraparound services, such as cost-sharing requirements for services covered by Medicare as well as acute care services (inpatient hospital, SNF, and home health care) that are delivered after the Medicare benefit has been exhausted. The principal uncovered Medicare service for dual eligibles is long-term care. For example, Komisar, Feder, and Gilden (2000) found that 78 percent of Medicaid's expenses for dual eligibles in 1995 was for long-term care. Home care accounted for 61 percent of Medicaid's spending for community-dwelling dual eligibles.

Medicare beneficiaries with somewhat higher incomes and greater assets are eligible for more limited Medicaid benefits, such as payment for the beneficiaries' Medicare premiums and some cost sharing, or assistance with only the premiums. These programs include qualified Medicare beneficiaries (QMBs), specified low-income Medicare beneficiaries (SLMBs), qualifying individuals (QIs), and qualified disabled working individuals (QDWIs).

Persons who are dually eligible for both Medicare and Medicaid have recently received considerable policy attention owing to their high cost and complex health needs. Although this population is relatively small in number, consisting of about 7.5 million individuals, spending on dual eligibles accounts for roughly 24 percent of Medicare's total spending and 42 percent of Medicaid's total spending (Kaiser Commission on Medicaid and the Uninsured 2004a). Mainly because of their poor health status, the Medicare costs of dually eligible beneficiaries are 1.5 times those of other Medicare beneficiaries (Medicare Payment Advisory Commission 2004).

Quality of care is also an important concern for the care of the dually eligible population (Haber and Mitchell 1999). In regard to those beneficiaries covered only by Medicare, dually eligible beneficiaries are much less likely to receive specific types of preventive care, follow-up care, or testing (Merrell, Colby, and Hogan 1997). For example, roughly 25 percent of dually eligible women received a mammography every two years, compared with 40 percent of Medicare-only beneficiaries. Of those Medicare beneficiaries with diabetes, dually eligible beneficiaries were less likely to receive an annual A1c test, a biennial ophthalmologic examination, and biennial lipid testing (McCall et al. 2004). In regard to long-term care needs, Komisar, Feder, and Kasper (2005) report that more than half (58 percent) of elderly community-based dual eligibles requiring assistance with activities of daily living have some unmet need.

Conflicting Incentives

Medicare and Medicaid interact in several ways that may have adverse implications for costs, access to services, and quality of care.1 First, Medicaid pays Medicare's cost-sharing requirements for the dually eligible population. Second, both Medicare and Medicaid cover certain home health care and nursing home services. Finally, Medicare pays for acute care services for dually eligible individuals, and Medicaid pays for their long-term care services.

Cost-Sharing Rules

Because Medicare is the primary payer for dual eligibles' hospital, physicians', and other acute medical care, Medicare policy influences not only Medicare's payments but also Medicaid's expenditures as a secondary payer. The states can decide whether they will pay the full copayments for Medicare services, the Medicaid rate for the same service, or an amount somewhere in between.2 Given the large difference in Medicare and Medicaid payment rates for certain services, the states may choose not to reimburse providers for the full 20 percent amount of Medicare coinsurance. Following the Balanced Budget Act of 1997, which explicitly upheld the rights of states not to reimburse providers for the full coinsurance amount, the number of states restricting their Medicare cost-sharing payments more than doubled to at least thirty (Nemore 1999). As expected, researchers have found that dually eligible beneficiaries in states with restrictive cost-sharing policies make fewer outpatient physician visits than do beneficiaries in states with more generous policies (Mitchell and Haber 2004).

From the perspective of long-term care, an important implication of restrictive cost-sharing rules is that Medicaid bears less of the cost of moving dually eligible long-term care recipients into acute care. Cost shifting across chronic and acute settings is explored more fully later, but the degree of cost sharing may affect this practice at the margin.3 In those states with more restrictive cost sharing, Medicaid has less incentive to enact policies discouraging transfers from long-term to acute care settings. As Medicaid's cost-sharing rules become less restrictive and it assumes a higher proportion of acute care costs, the program will be more likely to adopt policies to prevent transitions to acute care settings.

Cost Shifting within Home Health Care

Because dual eligibles can obtain similar home health services under both Medicare and Medicaid, cost shifting across the programs is possible.4 Medicare's home health care coverage guidelines were liberalized in 1989 in response to court decisions, which expanded the benefit from one focusing on the coverage of skilled home health care following a hospitalization to one also covering the chronic, long-term care of home health care patients. State Medicaid programs have taken advantage of these more liberal guidelines by employing a “Medicare maximization” strategy whenever possible to ensure that Medicare is the primary payer of home health care services (Wiener and Stevenson 1998b). For example, in 1996 Minnesota specifically enacted its Medicare Maximization Initiative, a program designed to teach home health care providers how to bill Medicare for dually eligible patients (U.S. General Accounting Office 1997). In a survey of states' units on aging and Medicaid departments conducted in 1998, three-quarters of them reported that they were maximizing Medicare's funding of home health care (Murtaugh et al. 1999).

The incentive to shift Medicaid's home health care costs to Medicare has been observed in the negative relationship (at the state level) between the utilization of Medicare and Medicaid home health care services (Cohen and Tumlinson 1997; Kenney and Rajan 2000; Kenney, Rajan, and Soscia 1998; Liu, Wissoker, and Rimes 1998). That is, beneficiaries in states with relatively low Medicaid home health spending more often used Medicare home health care services. Research also suggests that states with less generous Medicaid home- and community-based benefits have fewer people enrolled in Medicaid, implying that these states have an incentive to limit their Medicaid benefits (Pezzin and Kasper 2002).

The term Medicare maximization has also been applied to states' aggressive retrospective billing practices during the late 1980s in an attempt to shift home health care costs from Medicaid to Medicare. But those states that did this were not found to have an increase in Medicare home health care utilization or expenditures (Anderson, Norton, and Dow 2003). Moreover, recent cuts in Medicare home health care payments have not been shown to raise Medicaid's spending (Cheh 2001; Grabowski et al. 2006; Laguna Research Associates 2002). Nevertheless, the current system offers states an incentive to minimize Medicaid's home health benefits in order to maximize Medicare's payment of home health care services. Unless Medicaid's costs are fully shifted to Medicare, those states with less generous Medicaid benefits may reduce their access to services.

Cost Shifting within Nursing Homes

Medicare and Medicaid may also shift costs in the nursing home market. In the 1970s and early 1980s, nursing homes provided mainly custodial care to long-stay residents. The postacute, rehabilitative side of the nursing home market was negligible, with Medicare, the primary payer for these services, accounting for only 1.6 percent of total nursing home expenditures in 1980 (National Center for Health Statistics 2006). During this period, Medicare's coverage of nursing home care was often an “underused benefit,” with some states pursuing “Medicare maximization” policies to require nursing homes to bill Medicare for all potentially covered patients before billing Medicaid (Feder and Scanlon 1982).

Then a series of policy and market changes expanded the postacute side of the nursing home sector. By 2004, Medicare paid for 13.9 percent of all nursing homes' expenditures, with the majority of Medicare residents (87 percent) receiving care in a facility that also cared for Medicaid residents, and similarly, the majority of Medicaid residents (84 percent) were receiving care alongside Medicare residents.5 The general consensus of the nursing home industry (e.g., Floyd 2004), financial analysts (e.g., Feinstein and Fischbeck 2005), government (e.g., Medicare Payment Advisory Commission 2005), and researchers (e.g., Troyer 2002) is that Medicare residents are associated with higher profit margins compared with those of Medicaid residents. Although “Medicare maximization” policies for nursing home care are not as relevant today, given the growth of the postacute market, recent research has shown an important relationship between Medicare payment policies and the care of Medicaid nursing home residents. Specifically, Konetzka and colleagues (2004, 2006) found that the adoption of Medicare prospective payment for skilled nursing home care was associated with a lower quality of care for long-stay (i.e., predominantly Medicaid) residents. Presumably, the generosity of Medicaid payments may also have implications for the quality of care received by Medicare patients. Neither program, however, has an incentive to enact payment policies recognizing the welfare of residents covered by the other program. This disconnect between the two programs was encapsulated in a report by the Medicare Payment Advisory Commission (2005, p. 92) to Congress regarding nursing home payment policy: “If Medicare were to pay still higher rates to subsidize low Medicaid payments, states might be encouraged to reduce Medicaid payments even further.”

Cost Shifting from Nursing Homes to Hospitals

Because Medicaid and Medicare each cover certain aspects of care, they have little incentive to substitute care in low-cost settings for care in high-cost settings. An important example of this tension is Medicaid's coverage of nursing home care and Medicare's coverage of hospitalization for the dually eligible population. Residents of nursing homes are often hospitalized, with roughly one in four long-stay nursing home residents hospitalized annually (Grabowski, O'Malley, and Barhydt 2007). Many of these residents could be treated instead in the nursing home, and because hospitals are much more expensive, the total cost would be lower if unnecessary hospitalizations were avoided. Two examples from the literature support the idea that services could be delivered less expensively in the nursing home compared with the hospital. In the EverCare demonstration, Kane and colleagues (2004) reported that the per-diem cost of the hospital was $1,000, whereas an intensive service day (ISD) in the nursing home cost $425. An ISD refers to the specific EverCare payment associated with care that is more advanced than regular skilled nursing care. Similarly, Kruse and colleagues (2004) examined how hospitalizations for lower respiratory infections (LRI), primarily pneumonia and bronchitis, affected Missouri's expenses for nursing home residents. They found that the mean daily cost was $138.24 for the nursing home's treatment of LRI and was $419.75 for the hospital's treatment.

When considering the results of these two studies, it is important to acknowledge the possibility of a “moral hazard” problem, in which nursing homes may respond to the higher payments for particular conditions by enrolling in the EverCare (or Missouri LRI) program those residents who otherwise would not have been hospitalized. If program officials were able to select only those nursing home residents who otherwise would have been hospitalized, then to save money the nursing home treatment model would need to be only marginally less expensive than care in the hospital. However, as targeting becomes less exact, the aggregate savings from fewer hospitalizations would need to rise in order to cover the higher costs associated with the moral hazard problem. Nevertheless, both the EverCare and Missouri LRI studies suggest the potential for significant cost savings associated with treating residents in the nursing home, especially when the absolute number of hospitalizations is considered. An analysis of New York State's nursing home population, using merged hospital and nursing home administrative data, found 82,230 hospitalizations of long-stay nursing home residents in 2004. These hospitalizations accounted for $972 million of inpatient expenditures, of which 23 percent was for potentially avoidable hospitalizations (Grabowski, O'Malley, and Barhydt 2007). Moreover, inflation-adjusted expenditures for the hospitalization of nursing home residents increased 29 percent from 1999 to 2004.

Nursing homes that invest in the clinical services necessary to reduce the likelihood of hospitalization generate savings for Medicare, but at the same time, Medicaid often must pay for the higher cost of care in nursing homes. Thus, state Medicaid programs have little incentive to adopt policies to discourage hospitalizations. For example, the states have broad discretion to set the daily Medicaid nursing home payment rate (Wiener and Stevenson 1998a). Accordingly, in response to less generous Medicaid payments, nursing homes appear to reduce staffing (Cohen and Spector 1996; Grabowski 2001a, 2001b; Grabowski and Castle 2004; Intrator et al. 2005) and the use of nurse practitioners or physician assistants (NP/PAs) (Intrator et al. 2005). In facilities with inadequate staffing, temporary acute conditions are more likely to occur and are less likely to be well managed, typically necessitating hospitalization (Ackermann and Kemle 1998; Intrator, Castle, and Mor 1999; Intrator and Mor 2004; Intrator, Zinn, and Mor 2004; Reuben et al. 1999). Recent research found that a $10 increase in the Medicaid payment rate for nursing home care was significantly associated with 5 to 9 percent lower odds of hospitalization (Intrator et al. 2007; Intrator and Mor 2004).

Another Medicaid policy with potential implications for the hospitalization of nursing home residents are “bed-hold” policies. Bed-hold policies pay nursing homes to reserve the beds of those residents in the hospital for acute care. States vary in the proportion of the average Medicaid daily rate paid for bed-hold and the number of days covered. Some states also require a minimal occupancy rate before allowing bed-hold payments. The goal of bed-hold is to provide a continuous place of residence for the nursing home resident. In the absence of a bed-hold policy, some residents may refuse a needed hospitalization to avoid losing their bed (Nohlgren 2004). Conversely, if the marginal profit associated with the Medicaid bed-hold payment is greater than the marginal profit associated with the nursing home's Medicaid payment for continued care in the nursing home, then bed-hold introduces a financial incentive to hospitalize nursing home residents. Once again, recent research shows that the odds of hospitalization were 36 percent higher in states with bed-hold policies (Intrator et al. 2007). Thus, although many hospitalized nursing home residents could be effectively cared for in nursing homes that were given additional resources, state Medicaid programs have less incentive to increase Medicaid payment rates or repeal bed-hold policies because the savings associated with fewer hospitalizations will largely accrue to Medicare.

In addition to increasing overall costs, the hospitalization of nursing home patients may also have negative implications for quality. In a review of the literature on hospitalizations from nursing homes, Castle and Mor (1996) concluded that admission to an acute-care hospital can be traumatic for a nursing home resident. Frail older patients can suffer a number of iatrogenic problems while in the hospital for acute care. These include delirium, falls, incontinence, dehydration, adverse drug events, and nosocomial infections. Indeed, after being hospitalized, many nursing home residents return to the nursing home more functionally and cognitively impaired (Ouslander, Weinberg, and Phillips 2000). Research has suggested that certain conditions—such as pneumonia (Fried, Gillick, and Lipsitz 1997; Kruse et al. 2004; Naughton, Mylotte, and Tayara 2000; Thompson, Hall, and Szpiech 1999) and infections (Boockvar et al. 2005)—can be treated at least as well (if not better) in the nursing home compared with a hospital. Accordingly, the planned Medicare Nursing Home Value-Based Purchasing Demonstration recently proposed avoidable hospitalizations as a (negative) performance measure.

Cost Shifting from Home- and Community-Based Programs to Hospitals

The hospitalization of community-dwelling dual eligibles also introduces potentially conflicting incentives. The proportion of total Medicaid long-term care expenditures directed to home- and community-based services (HCBS) grew from 11 percent in 1988 to 27 percent in 2000, although some of this change can be attributed to spending for persons with mental retardation and persons with developmental disabilities (Wiener, Tilly, and Alecxih 2002). One of the key sources of growth has been the Medicaid 1915(c) HCBS waiver program, which was authorized by Congress in 1981 to give states matching federal dollars to expand HCBS (Miller, Ramsland, and Harrington 1999). Medicare is the payer of hospital services for dually eligible enrollees in these programs. Medicaid's HCBS programs that invest in the clinical services necessary to reduce the likelihood of hospitalization generate savings for Medicare. Again, therefore, state Medicaid programs have little incentive to adopt policies to discourage hospitalizations. Miller and Weissert (2003, p. 153) related the response of one state official to a proposed program to reduce hospitalizations of Medicaid HCBS clients: “Why would we want to do that that? Those are Medicare dollars. For us that's development money. We don't want to reduce Medicare expenditures in our state.” In support of this statement, a recent evaluation of Florida HCBS programs found evidence of cost shifting to Medicare in the higher number of inpatient hospital days for dually eligible enrollees (Mitchell et al. 2006). As the study notes, the Medicaid HCBS contractor is responsible for paying only the Medicare deductibles and copayments when a dually eligible enrollee is hospitalized, with Medicare paying the bulk of the hospital expenditures. If the dually eligible enrollee is not hospitalized, the HCBS contractor must pay the full cost of services, such as respite care. Thus, the program has an incentive to shift costs to Medicare by hospitalizing clients, which also introduces a number of potential health complications associated with hospitalization.

Another implication of the bifurcated coverage of HCBS and hospital care is a systematic lack of coordination among primary, acute, and long-term care providers (Peters 2005). Medicare and Medicaid focus on meeting specific covered service needs rather than addressing the interaction of acute and chronic needs. Even though case management is often an important component of Medicaid HCBS, the case manager is typically not responsible for enrollees when an acute care episode results in hospitalization or entry into a skilled nursing facility. This lack of coordination may increase the likelihood that the individual will not return to the community following an acute care episode. If Medicaid's HCBS programs had to pay the costs of their actions directly, this would produce a greater incentive to coordinate or manage services across acute and chronic care settings.

Cost Shifting from Hospitals to Nursing Homes

The previous two examples focused on cost shifting from Medicaid's chronic care settings to Medicare's acute care settings, but costs also may shift in the opposite direction. Although this type of cost shifting has not received much attention from the research community, it stands to reason that Medicare's investment in acute care services will affect Medicaid's spending on long-term care services. Because the use of long-term care services is often far downstream, it is hard to tie Medicare's investment in acute or primary care services directly to Medicaid's spending on custodial services. One example is Medicare's 1983 adoption of a prospective payment system (PPS) for hospital care, which led to patients being discharged “sicker and quicker” (Kosecoff et al. 1990). This change in payment contributed to the growth in Medicare-covered postacute nursing home care in the years following PPS (Dalton and Howard 2002). Given that more than half of all Medicaid nursing home stays began as a Medicare-covered postacute care admission following a hospital stay, it seems clear that Medicare's hospital PPS ultimately had downstream implications for the custodial Medicaid nursing home population. For example, there is evidence that custodial nursing home residents were more disabled after PPS in 1986 compared with those before PPS in 1982 (Shaughnessy and Kramer 1990).

The transfer of patients from the hospital to the nursing home also raises issues related to the coordination of care and the beneficiaries' health. Under the Medicare hospital PPS, discharge planners have more incentive to discharge patients as soon as (safely) possible but less incentive to consider the long-term cost and health implications of the initial discharge placement. For example, given the high number of Medicare nursing home stays that ultimately become Medicaid nursing home stays, it is desirable that the nursing home to which a hospitalized patient is discharged participate in Medicaid, even if the initial nursing home stay is financed by Medicare. Such placements would remove the need to transfer the patient when his or her Medicare coverage ends, thereby avoiding the adverse health consequences of transfers. Similarly, hospital discharge planners ideally should avoid transfers to nursing homes when adequate home care is available to support a community-based placement. This could improve the patient's welfare and lower Medicaid's spending, but under the current Medicare hospital payment system, discharge planners are not rewarded for placing patients in the most appropriate setting. They have little incentive to consider the long-term implications of the discharge placement for either the beneficiary's long-term health or Medicaid's budget.

Potential Solutions

Several policy options are available to address the conflicting incentives outlined in the previous section. The key objective of such policies is to have Medicare and Medicaid internalize each other's costs while also sharing any potential savings. They could do this by means of either broad or focused policy measures. The broader policy measures include capitation, which could blend the financing of the two programs, and the federalization of the Medicaid program. A more focused approach, such as pay-for-performance, could address the misalignment of particular incentives, such as the hospitalization of nursing home residents.

Capitation

One mechanism that has been proposed to address the bifurcation of Medicaid and Medicare is capitated managed care (Rudolph and Lubitz 1999). Demonstration programs have waived certain provisions of the Medicare and Medicaid programs, thereby allowing payment for services that would otherwise not be covered and the use of different methods to pay for these services. Some programs combine postacute and long-term care services through managed care. Although the nature and scope of the demonstration programs are quite diverse, the use of capitated payments may encourage a more efficient production of health care services.

Comprehensive reviews of these capitated managed care plans have been conducted elsewhere (Grabowski 2006; Miller and Weissert 2003; Saucier, Burwell, and Gerst 2005). The focus here is on programs that combine Medicaid and Medicare financing and address the two programs' conflicting incentives (Tritz 2006). Federal managed care initiatives include the Program of All-Inclusive Care for the Elderly (PACE), the Social/Health Maintenance Organization (S/HMO) demonstration, and the EverCare program. PACE is the only federal program that necessarily combines Medicaid and Medicare financing. State managed care initiatives that combine Medicaid and Medicare financing include Minnesota Senior Health Options (MSHO), Massachusetts Health Senior Care Options (SCO), New York Medicaid Advantage Program, Washington Medicare/Medicaid Integration Program, and Wisconsin Partnership Program. Two other programs, Arizona Long-Term Care System (ALTCS) and Texas STAR+PLUS, capitate only Medicaid but offer beneficiaries incentives to join optional companion Medicare managed care plans.

Overall, relatively few long-term care recipients are covered under capitated arrangements. In 2004, less than 3 percent of the publicly funded, long-term care population received their long-term care benefits through a managed care program (Saucier, Burwell, and Gerst 2005), and this number included programs that did not combine Medicare financing. Programs combining Medicare and Medicaid financing are not currently available in most parts of the country, and even in those areas with such programs, enrollment typically is voluntary. Medicare's freedom of choice gives beneficiaries the right to choose between a managed Medicare program and Medicare's fee-for-service, without being locked in to a particular choice over time. Thus, of the managed care programs just mentioned, only the ALTCS and Texas STAR+PLUS have required enrollment, and these are Medicaid long-term care plans, which are not necessarily combined with Medicare. A lack of enrollment can partially be explained by dual eligibles' concerns that the managed care programs mean that they must change doctors, go to locations for care, and have fewer choices (Peters 2005). Those programs that combine acute and long-term care have other concerns with managed long-term care, such as the overly “medicalized” model (entailing less consumer direction and little substitution, in favor of lower costs and fewer technical services) and the fear that acute care will absorb all the money (Wiener 1996). Because of these “demand-side” concerns with managed care, the PACE program attracted a disproportionate number of healthy enrollees (Irvin, Massey, and Dorsey 1997). A “supply-side” factor that might have contributed to this finding was the plans' niche marketing (or “cream skimming”) in an effort to attract the most profitable patients (Branch, Coulam, and Zimmerman 1995).

As noted in the previous section, neither Medicare nor Medicaid has a strong incentive to offer case management under fee-for-service payment, because neither program would fully internalize the potential savings. Managed care programs have used a number of service delivery models to coordinate the care for dually eligible beneficiaries enrolled in integrated plans (Bishop et al. 2007). For example, MSHO uses a single nurse or social worker to coordinate services; the Wisconsin Partnership Program employs a multidisciplinary team that includes a nurse practitioner; and the Massachusetts SCO program utilizes a team of nurses and social workers. In Wisconsin, nurse practitioners accompany patients on their visits to physicians as a means of more closely connecting community and medical care.

The PACE and MSHO programs are the only two models integrating Medicaid and Medicare that have been rigorously evaluated.6 These evaluations found that these programs cost more than those of the comparison groups. Specifically, the total capitated payment to PACE enrollees was 9.7 percent higher in the first year of enrollment than the projected Medicare and Medicaid cost if the enrollees had continued to receive care in a fee-for-service (FFS) program (White, Abel, and Kidder 2000).7 Interestingly, the PACE program was associated with 42 percent lower Medicare spending, but 86 percent higher Medicaid spending. The costs for both Medicaid and Medicare were higher for MSHO enrollees compared with the FFS costs for a control group of both community-dwelling individuals and nursing home residents after adjusting for demographic factors and prior health care utilization (Kane and Homyak 2003).8 Specifically, Medicare costs were 51 percent higher for community MSHO enrollees and 44 percent higher for nursing home residents, whereas Medicaid MSHO costs were 31 percent greater in the community and 7 percent greater in the nursing home. Thus, even though a primary goal of capitation is to lower spending, the two most rigorous evaluations of this model actually indicated higher costs. The reasons for this result could be the failure to target services to enrollees through a stringent preadmission process and the inability to contain spending on particular services.

Quality of care and enrollees' access to services were found to improve under PACE (Chatterji et al. 1998) and to remain relatively stable under MSHO (Kane et al. 2005; Kane et al. 2003). Specifically, PACE was associated with the following statistically significant outcomes for enrollees: greater use of adult day health care, fewer home visits by nurses, fewer hospitalizations, fewer nursing home admissions, a higher probability of receiving ambulatory care, greater survival, an increased number of days in the community, better health, better quality of life, greater satisfaction with overall care arrangements, and better functional status. The PACE enrollees with the most severely limiting conditions at baseline had the largest gain. A multivariate analysis of the MSHO did not show substantial differences across a number of outcomes, including mortality, nursing home admissions, functioning, satisfaction, and caregiver burden across the treatment and control groups. Thus, the best evidence to date from the PACE and MSHO programs does not suggest that integrated capitation models are cost-effective relative to fee-for-service comparison groups.

One recent policy innovation toward coordinating Medicare and Medicaid is the establishment of Medicare Advantage special needs plans (SNPs). SNPs were authorized under the Medicare Modernization Act (MMA) of 2003 with the idea of attracting a different type of beneficiary to Medicare Advantage (i.e., a means of receiving health care and Medicare coverage through private health plans, which are, most commonly, health maintenance organizations). From the perspective of program coordination, SNPs allow states the opportunity to combine Medicare's and Medicaid's managed care contracting for dually eligible beneficiaries without having to secure special Medicare demonstration authority from the CMS. By July 2006, the CMS reported that 273 SNPs had been approved, with the majority (226) being dual-eligibility SNPs (Saucier and Burwell 2007). However, the MMA does not require that these dual-eligibility SNPs coordinate benefits with Medicaid. SNPs are Medicare Advantage plans, and the vast majority of SNPs for dual eligibles provide Medicare-covered benefits at the higher capitation rate that CMS pays for dual eligibles. However, the MMA gave Medicaid plans a one-time opportunity to seek SNP designation and to automatically (unless the beneficiary notified the plan otherwise) enroll dually eligible members into their companion Medicare plans as part of the initial Medicare Part D enrollment process. For states with mature Medicaid managed care programs that included dual eligibles, automatic (or “passive”) enrollment provided an opportunity to greatly expand the number of dually eligible beneficiaries in integrated products. Specifically, an estimated 47,000 Medicaid beneficiaries in managed long-term care plans became dually enrolled through passive enrollment in Arizona (via ALTCS), Minnesota (via MSHO), and Texas (via STAR+PLUS) (Saucier and Burwell 2007).9

It remains to be seen whether SNPs will ultimately become a vehicle for significantly increasing the integration of Medicare and Medicaid. SNPs do not, however, address the underlying conflict between federal and state approaches to managed care. The federal approach to Medicare Advantage plans is based on consumer choice, with a strong preference for variation across plans in the marketplace. Medicare freedom of choice is an important beneficiary right, but it can confound coordination if dually eligibles enroll in different Medicare and Medicaid plans. In contrast to Medicare, state Medicaid plans typically emphasize long-term investments with a limited number of plans and uniform benefits to promote equity in a publicly funded program. Moreover, a potential issue for the states is the alignment of incentives under a combined Medicare-Medicaid product. Although CMS and the SNPs share in any savings from lower Medicare hospital costs, the states do not directly benefit (Saucier and Burwell 2007).

Pay-for-Performance

Capitated managed care is a global mechanism for aligning Medicare and Medicaid financing. A more focused measure to address the high rate hospitalizations of nursing home residents—an important area of disconnect between Medicare and Medicaid policy—is pay-for-performance. Although the existing empirical literature offers little evidence to support the effectiveness of paying for quality in the health care sector (Rosenthal and Frank 2006), a controlled experiment in San Diego found that nursing homes were responsive to monetary incentives for reducing the residents' level of dependency and the need for special nursing services (Norton 1992; Weissert et al. 1983). Thus, there is some previous support for paying for quality in nursing homes.

The federal nursing home pay-for-performance demonstration is currently in the planning stages. The demonstration will focus on four performance measures: avoidable hospitalizations, quality indicators (e.g., pressure ulcers), staffing levels and stability, and survey deficiencies. Given the requirement that the demonstration must be budget neutral, the plan is for the Medicare savings from fewer avoidable hospitalizations to fund the incentive-based payments to nursing homes. It is unclear whether the level of reward will be sufficient to encourage fewer hospitalizations. Moreover, a pay-for-performance type demonstration will not address many of the other inconsistencies between Medicaid and Medicare policy, although it certainly could with a broader design.

Pay-for-performance has some appeal at the federal level, but there is less incentive for state Medicaid programs to incorporate avoidable hospitalizations into similar payment systems. This issue is partly due to the lack of data sharing across Medicare and Medicaid, with many states unable to identify the rate of hospitalization from nursing homes without access to Medicare claims data. More broadly however, this issue arises because Medicaid would incur the full costs of rewarding nursing homes with incentive-based payments and enjoy little of the savings from decreased hospitalizations. Data from New York State in 2004 indicate Medicaid was the primary payer for $26.5 million (or 12 percent) of avoidable hospitalization costs, while Medicare was the primary payer for $188.5 million (or 84 percent) (Grabowski, O'Malley, and Barhydt 2007). Four states have Medicaid nursing home pay-for-performance systems either currently in place (Iowa and Kansas) or slated to come online in fiscal year 2007 (Minnesota and Ohio). These states reward nursing homes for good performance along a number of dimensions including staffing, clinical quality indicators, occupancy rates, resident/family satisfaction, and survey deficiencies, but none have incorporated hospitalizations as a performance measure. Until Medicaid and Medicare can identify a mechanism to share in the benefits of fewer hospitalizations, this will likely not be a primary area of emphasis for state policymakers.

Federalizing Medicaid for Dual Eligibles

Although capitation and pay-for-performance may help align incentives for the Medicare and Medicaid programs, a more dramatic proposal to eliminate the conflicting incentives altogether is to shift financial responsibility for the care of the dually eligible population, including long-term care, to the federal government (Holahan and Weil 2007; U.S. General Accounting Office 1995). The idea is that this shift—to either Medicare or some federal program—would improve the coordination of care for dually eligible enrollees and also offer substantial fiscal relief to the states. During the recent budget shortfalls, many states have struggled to maintain coverage and benefits for Medicaid beneficiaries. Because the federal government has broader taxing and borrowing authority and state revenues can often be quite volatile, the federalization of care for dual eligibles would place more of the costs (and risks for future cost growth) on the federal government. Bruen and Holahan (2003) estimated that in 2002 this change would have shifted $25.8 billion in long-term care spending from the states to the federal government. Nonetheless, this estimate is simply an accounting figure and does not take into account any beneficial “behavioral” effects that might result from the integration of Medicare and Medicaid. For example, with the federalization of care for the dual eligibles, Medicare would have a greater incentive to increase the subsidy of acute care in nursing homes in order to reduce the number of avoidable hospitalizations.

The idea of federalizing care for dually eligible enrollees dates back at least to the early 1980s (U.S. General Accounting Office 1995), with a recent endorsement by the state governors (National Governors Association 2005). There is not, however, a groundswell of political support for the proposal in the current U.S. Congress. The current fiscal constraints at both the federal and state levels have led to tremendous pressures to control the growth of Medicaid and Medicare dollars allocated to long-term care. It is unclear whether the federal government would be willing to assume these additional expenditures, even if it resulted in a comparable decrease in state spending. Moreover, even though shifting responsibilities from Medicaid to Medicare would resolve the two programs' conflicting financial incentives, it would not (necessarily) address the system's lack of case management or other problems.

Discussion

Bifurcated coverage introduces a number of conflicting incentives in Medicare and Medicaid regarding the delivery of long-term care services. This article reviewed the implications of the lack of coordination across the two programs in cost-sharing arrangements, cost shifting within settings such as home health care and nursing homes, and cost shifting across health care settings (see table 1 for a summary). This review offers a number of lessons for the research and policy communities.

Table 1.

Examples of Conflicting Incentives across Medicare and Medicaid

Medicare-Medicaid Interaction	Medicare's Role	Medicaid's Role	Incentives	Potential Implications
Cost sharing	Primary payer for dual eligibles' hospital, physician, and other acute medical care	States have discretion as to whether they pay the full Medicare copayments.	States have the incentive to restrict their Medicare cost-sharing payments.	• Less access for dually eligible beneficiaries in states with more restrictive policies
				• Less cost shifting to Medicare when states bear the full cost-sharing amount
Cost shifting within home health care	Pays for qualified home health care stays	Offers coverage as a “wrap-around” payer of home health care	States have the incentive to restrict Medicaid coverage and benefits in order to shift costs to Medicare.	• Unless costs are fully shifted to Medicare, Medicaid recipients have less coverage.
Cost shifting within nursing homes	Pays for postacute skilled nursing home care	Pays for custodial nursing home care	Given spillovers in costs and outcomes across postacute and custodial patients, both payers lack the full incentive to increase payment rates.	• Lower quality for both custodial and postacute nursing home residents
Hospitalizations of nursing home residents	Covers dual eligibles' hospital care	Covers long-stay custodial nursing home care	Medicaid programs have less incentive to adopt policies to discourage hospitalizations of nursing home residents.	• Greater costs associated with hospitalizations
				• Hospitalizations often associated with lower quality
				• Less primary care or case management in the nursing home setting
Hospitalizations for Medicaid home- and community-based services (HCBS) clients	Covers dual eligibles' hospital care	Covers Medicaid HCBS program, which often includes case management	Medicaid programs have less incentive to discourage hospitalizations or coordinate care across settings (e.g., HCBS case management does not extend to hospital setting).	• Greater costs associated with hospitalizations
				• Hospitalizations often associated with lower quality
Cost shifting from acute to chronic care settings	Covers dual eligibles' hospital care	Covers long-stay custodial nursing home care	Medicare has no incentive to discourage Medicaid long-term care costs or to coordinate care following hospital discharge.	• Higher Medicaid long-term care costs
				• Inappropriate placement of discharged hospital patients, which may introduce adverse quality implications

Lessons for Researchers

The evidence regarding capitated programs would greatly benefit from more sophisticated analyses of the issue of selection into a capitated program. That is, individuals enrolling in capitated programs (i.e., the treatment group) may differ in ways unobservable to the researcher compared with individuals not enrolling in these programs (i.e., the control group). These unobservable differences may accordingly bias comparisons of the two study groups' costs and health outcomes. Clearly, the ideal would be a randomized study design, although no evaluation of a capitated program meets this standard. The PACE, MSHO, Texas STAR+PLUS, and Wisconsin Family Care programs all would have been ideal candidates for the randomized assignment of individuals to either a treatment or a control group. When randomization is viewed as too costly or infeasible, an instrumental variables approach can also be used to address the issue of selection. By finding an instrument that predicts program enrollment but not the outcomes of interest such as costs and health outcomes, this approach can be used effectively to “randomize” individuals even in a voluntary program.

The results from the pay-for-performance demonstration will be important to establishing whether inappropriate hospitalizations of nursing home residents can be reduced by means of financial incentives. Once again, the use of a randomized controlled trial would be preferable, similar to the social experiment conducted in San Diego in the early 1980s (Norton 1992; Weissert et al. 1983). The federalization of Medicaid for dual eligibles may not have much current political support, but if the budget deficits in Medicaid become worse in the coming years, this option may become more palatable. The current evidence regarding the federalization of Medicaid is based on simulations that assume no “behavioral” effects, such as a more efficient use of resources or a better quality of care following the federalization of care. As such, it will be important to consider these issues when evaluating this option.

Lessons for Policymakers

In light of the current political climate, the most likely approach for addressing the problems that accompany dual eligibility is a further expansion of capitated managed care, perhaps with some combination of pay-for-performance type incentives. Given the recent emergence of Medicare Advantage SNPs, a key question is whether these plans can be a vehicle to dramatically increase enrollment in integrated managed care products. Although SNP enrollment has been modest thus far, it has been largely a function of the one-time opportunity that Medicare offered to plans to seek SNP designation and automatically enroll dually eligible members into their companion Medicare plans as part of the initial Medicare Part D enrollment process. It is unclear whether SNPs will offer integrated products and, more important, whether dually eligible beneficiaries will enroll in these products. From a policy perspective, a key issue is whether there are ways of expanding this market. The most obvious one is to make enrollment mandatory, similar to Medicaid managed care plans such as ALTCS and Texas STAR+PLUS. This option may be unpopular with beneficiaries, however. Another way of expanding capitation would be to increase the financial incentives for beneficiaries to enroll in these programs, although this may make it more difficult to offer beneficiaries a cost-effective product.

If the pay-for-performance demonstration is successful, it may be worthwhile to offer incentives to other areas of care for the dually eligible population. For example, providers of Medicaid and home- and community-based services (HCBS) could also be rewarded for preventing inappropriate hospitalizations. The key issue will be sharing the Medicare savings from these nonhospitalizations with home- and community-based providers typically paid by Medicaid, perhaps as a separate payment (as in the nursing home pay-for-performance demonstration) or as a higher Medicaid rate funded by Medicare. Moreover, pay-for-performance and capitation need not be thought of as mutually exclusive. For example, Weissert and colleagues (2003) suggested an intervention in which HCBS case managers in the ALTCS program would be responsible for the outcomes of their patients, rewarding them if their resource allocation decisions lowered costs and improved patient outcomes.

Finally, this article argued that Medicare policies may affect the financing and quality of care for Medicaid recipients. These interdependencies are, of course, reciprocal: Medicaid policies also affect Medicare's expenditures and patient outcomes. There is a critical distinction, however. Medicare is a national program administered by the federal government. In a federal system, the role of the central government is to set the rules, including the delegation of certain authorities (e.g., the Medicaid program) to the states, so as to promote joint welfare (Oates 1972). We cannot expect the states' Medicaid programs to set their policies to recognize externalities to Medicare. But we can expect (in a normative sense) the federal government to look beyond narrow program interests and set its policies to recognize their effects on other groups.

This view suggests that the federal government must take the lead to produce significant change. As pointed out earlier, it is unclear whether the political support can be found to shift full financial support for care of the dually eligible population to the federal government. Moreover, it also is unclear whether a federally run program would do a better job than the current bifurcated system is doing. The problems with Medicare go beyond the tensions with Medicaid reviewed here (Angell 1997; Vaughan 2003). The Medicare program began with the mandate to “not interfere with the practice of medicine,” and the program's uncoordinated fee-for-service approach fits this proviso. Moreover, Medicare's program rules often limit its ability to direct innovative programs to particularly costly subpopulations. In regard to cost, state Medicaid programs have often been much more aggressive than the federal government in controlling nursing home and home health care expenditures. It is unclear whether the federal government would be able to put the necessary constraints in place to limit the growth of long-term care expenditures. Thus, even under a single federal program that internalized all the risks and rewards associated with the care of the dually eligible population, there still might be barriers to case management, primary care, better quality, and lower costs. Addressing the conflicting incentives across Medicare and Medicaid would be an important step toward improving the long-term care system, but it may be only one part of a larger set of needed reforms.

In sum, the current structure of Medicare and Medicaid does not offer a coordinated system of care for the majority of dually eligible beneficiaries, thereby creating a number of conflicting incentives across the two programs. In turn, these conflicting incentives often lead to higher program costs for Medicare and Medicaid, a lack of care management, and poor quality of care. Although several mechanisms such as capitation, pay-for-performance, and the federalization of Medicaid for dual eligibles might resolve these conflicting incentives, all have drawbacks. With the aging baby boom generation and projected budget shortfalls at both the federal and state levels, this issue will be a continued area of interest to policymakers in the coming decades.