The Hidden Costs of Rebalancing Long-Term Care

Expanding the funding of home- and community-based services (HCBS) has become a key policy goal of many state Medicaid programs in recent decades. Whereas 87 percent of state Medicaid spending on long-term care went to institutional care (nursing homes) as of 1990, now almost half goes to HCBS (The Henry J. Kaiser Family Foundation 2012). This shift has been achieved in two primary ways—expanded provision of HCBS services under Medicaid state plan options, and the use of 1915(c) waiver programs allowing states to use Medicaid funds to provide long-term care services in noninstitutional settings. HCBS includes such services as home care, adult day care, respite care, and adult foster care, with home care being the most common.

Several provisions of the Affordable Care Act reinforce and encourage this trend. The Act extends the Money Follows the Person demonstration, in which states may use enhanced federal matching funds to transition Medicaid beneficiaries in nursing homes back to the community. Under the Act's New Balancing Incentive Program, states also receive additional matching funds to make structural changes to their Medicaid programs to increase availability of HCBS. Finally, some of the waiver requirements were relaxed to decrease the administrative burden associated with state expansions of HCBS.

Shifting from funding of institutional care to HCBS is often referred to as “rebalancing” or “balancing” (Harrington et al. 2009; Grabowski et al. 2010), under the implicit assumption that historical allocation of resources to the two types of care was out of balance. In a sense, this assumption was undoubtedly true in that regulatory policy was very restrictive about the care settings in which Medicaid recipients could receive long-term care. Individuals in need of long-term care services who relied on Medicaid generally only had the option of nursing home care. In large part, this was an attempt to contain costs by ensuring that only those with the most serious long-term care needs would obtain benefits. The unintended consequence was that many individuals with less severe long-term care needs (though severe enough to qualify) were living in nursing homes when a HCBS setting would potentially be both cheaper and preferred. From an economist's perspective, these were distortionary policies that impeded states' and consumers' ability to allocate resources efficiently.

To the extent that the push to “balance” long-term care removes distortionary barriers, both long-term care recipients and the Medicaid program should be better off. In addition, removing barriers to funding HCBS should increase the supply of providers of HCBS such that there are true options for more consumers. However, beyond removing overly restrictive regulations, the intended and unintended consequences of balancing are much more complicated than is often assumed, for three key reasons.

First, although a commonly used metric, a comparison of state Medicaid dollars spent on the two types of care has little meaning. States with older and sicker populations will have a greater need for nursing home care. At the individual level, nursing home residents are in substantially worse health on average. Even under some economies of scale in an institutional setting, one should expect nursing home residents to cost more than HCBS recipients due to their high health care needs, and this becomes increasingly true as less severely impaired residents substitute HCBS for nursing home care. Optimal allocation of resources across HCBS and nursing home care should depend on the needs of the population.

Second, several of the key assumptions driving the expansion of HCBS, while they may seem self-evident at first glance, are not based on solid evidence. One is that long-term care recipients have strong preferences to remain in their homes. This may be true on average and for low levels of functional impairment, but several recent studies reveal that preferences depend on health state, with a preference for institutional care emerging once cognitive impairment sets in (Wolff, Kasper, and Shore 2008; Guo, Konetzka, and Dale 2014). It seems clear that, in addition to some heterogeneity in preferences across individuals, there may be key “tipping points” where strong preferences to stay at home diminish or even reverse. The other key assumption is that providing services in the community is more cost-effective than institutional care. A review in 2006 found that the cost-effectiveness of HCBS was inconclusive (Grabowski 2006). While several state-level studies since then have provided support for some cost offsets (Muramatsu et al. 2007; Kaye, LaPlante, and Harrington 2009), these studies are inherently difficult to interpret due to potential selection bias and ecological fallacy. A new study at the individual level indicates that spending on home care may offset some spending on other types of care but the offset is not complete (Guo, Konetzka, and Manning in press). In other words, HCBS may be cost-effective if people are better off and it is worth the additional spending, but we should not think of HCBS as cost-saving. This is a key distinction in current times. These calculations also do not count the costs of informal caregiver burden, costs that are substantial in long-term care (Stone, Cafferata, and Sangl 1987; Arno, Levine, and Memmott 1999; Spillman and Pezzin 2000; Coe and Van Houtven 2009) and which are likely to be higher for HCBS, which often depends to a greater extent on synergies between formal and informal care.

The third key reason that the consequences of the drive to “balance” long-term care are complicated is that we know surprisingly little about the outcomes of HCBS, relative to the alternatives. Some evaluations of specific types of HCBS programs or demonstrations have evaluated an array of outcomes (Wieland et al. 2000; Carlson et al. 2007), but these results are not likely to be generalizable to expansion of HCBS more generally. Several recent studies examined potentially avoidable hospitalizations among HCBS users (Konetzka, Karon, and Potter 2012; Walsh et al. 2012), but they did not provide comparisons. Setting preferences aside, it is not clear that one should expect better health outcomes under HCBS, and in fact one might expect worse outcomes given the less intensive level of resources available. Home environments may not be safe or appropriately designed to accommodate long-term care needs, and informal caregivers may not be well trained to handle clinical issues.

In short, “balancing” long-term care is difficult because no one knows what the correct balance should be. From a societal perspective, the correct allocation should reflect direct and indirect costs of each potential setting of care, objective outcomes of care, and subjective assessments of quality of life or preferences for each type of care, all contingent on specific health states. Funding should go to services for which the incremental benefits outweigh the incremental costs. Although we may never have information that complete, we could get substantially closer to setting reasonable targets with an improvement in the evidence base. Furthermore, setting goals for balancing in the absence of strong evidence may have the unintended consequence of creating new and different distortionary policies. Some might argue that we are a long way from being at this point in the spectrum, but if individuals who are truly in need of nursing home care are instead pushed into HCBS, the result could be worse outcomes at higher cost.

In this issue of Health Services Research, Wysocki et al. (2014) provide a good example of potential “hidden costs” of expanding HCBS, costs that need to be taken into account in assessing how much to expand HCBS. They examine the probability of a potentially preventable hospital admission among elderly individuals dually enrolled in Medicare and Medicaid, comparing the probability among HCBS users to nursing home residents. At first glance, one might expect the hospitalization rate to be higher among the nursing home residents, due both to their worse health status and the well-known financial incentives nursing homes face that encourage transfers to the hospital (Grabowski 2007). Advocates of HCBS may expect lower hospitalization rates among HCBS if HCBS is a more appropriate setting. However, Wysocki and colleagues found that HCBS users had a one-percentage-point higher probability of a potentially preventable hospital admission, a large effect relative to the mean rate of 3.7 percent hospitalized for these conditions per quarter in the nursing home sample.

The data used for the analysis are excellent—3 years of merged Medicare and Medicaid claims from seven states. The analysis was carefully conducted, with inclusion of appropriate controls, multiple robustness checks, and diligent accounting for the possibility of competing risks of nonpreventable hospitalizations and death. The primary challenge, fully acknowledged by the authors, is addressing selection bias. The HCBS and nursing home populations are quite different, as is apparent in Table 1 of the Wysocki et al. paper. The nursing home population is older, more likely to be white, more likely to be diagnosed with dementia, depression, and stroke, and more likely to die. More important, individuals who use HCBS are also likely to be different from individuals who use nursing home care in unmeasured ways. Ideally, randomization or a valid instrumental variable would be used to eliminate this bias, but as is so often the case, neither was available. As the next best alternative, the authors used propensity score weighting to increase the extent of overlap in the distribution of characteristics in the two populations. The standard caveat with propensity score methods is that they do not address selection bias on unmeasured characteristics; thus, the authors cannot claim a causal connection between care setting and potentially avoidable hospitalizations. Arguably the most important source of bias would be unmeasured health status. In this case, however, we might assume that nursing home residents are sicker than HCBS recipients in unmeasured ways, implying that the increase in hospitalization attributable to HCBS is actually underestimated in the analysis. The finding of higher rates in HCBS is therefore quite plausible even allowing for some selection bias.

A key contribution of the paper by Wysocki et al. is that it begins to make inroads into exactly the type of evidence that is needed to make efficient choices about allocating scarce resources across long-term care settings. It makes a head-to-head comparison of a key outcome across care delivery modes and identifies a potentially hidden cost of HCBS. Higher rates of hospitalization are costly not only in terms of Medicare dollars but also in terms of the physical and psychological costs of these transfers for HCBS recipients. These are costs that need to be considered in expanding HCBS.

In their discussion, the authors conclude that higher hospitalization rates from HCBS point to the need for greater integration of long-term and acute care. Although greater integration is difficult to argue with, the need for it applies across settings. Furthermore, it is not clear that greater integration will compensate for the lower levels of technology and skill that individuals face in their homes, relative to institutional settings, such that hospitalizations could be reduced. Perhaps the more important point illuminated by this paper relates back to the broader context, that the advantages of expanding HCBS should not be taken as self-evident. For many people with long-term care needs, especially those at lower levels of need, HCBS will offer a higher quality of life and better outcomes at lower cost than care in a nursing home. For others, the benefits of HCBS may not outweigh the costs, costs which may include greater informal caregiver burden, higher hospitalization rates, or additional expenditures to avoid hospitalizations. Determining where the tipping point lies, relative to an individual's health state, should be a primary research aim.

Removing regulatory distortions to the choice of long-term care setting will facilitate the goal of making health care more person-centered, finding the right care at the right point in time for each individual. Pursuing a nebulous and arbitrary balance in the allocation of dollars for one setting versus another will not, and it may potentially create new distortions. To be able to make the difficult decisions about the most appropriate setting for each individual and to allocate funding efficiently, the evidence base is currently weak. Wysocki et al. identified exactly the right type of question to ask, but much more research of this type is needed.