|
Medicaid Payment And Risk-Adjusted Nursing Home Quality Measures
Various studies have observed low quality in the nursing home industry. Although Medicaid is the dominant payer of U.S. nursing home services, the association of Medicaid payment rates and quality is not entirely clear, in part because resident-level, risk-adjusted information on quality is lacking. This study examined the relationship between Medicaid payment rates and three risk-adjusted quality measures, controlling for market and facility characteristics. Higher payment was associated with lower incidence of pressure ulcers and physical restraints but not daily pain. Quality of nursing home care may suffer if budget shortfalls force state legislatures to freeze or reduce Medicaid rates.
Medicaid is the dominant payer of nursing home services in the United States, accounting for roughly half of all nursing home spending ($44 billion in 1998, the latest year for which data are available). The Balanced Budget Act (BBA) of 1997 repealed the Boren amendment, which had previously required that states’ Medicaid nursing home payment rates be adequate to cover the costs of care.1 Given current state budget shortfalls, there is concern that states will choose to slash Medicaid nursing home payment levels as a way of lowering state spending. A Henry J. Kaiser Family Foundation survey of state Medicaid directors found that forty-nine states planned to reduce the rate of growth in Medicaid spending, while nineteen planned actual cuts in their Medicaid spending for long-term care in fiscal year 2003.2 The quality of care in the U.S. nursing home industry is already an important policy issue.3 Consequently, the relationship of Medicaid payment levels and nursing home quality also is an important policy issue. Previous work on this relationship has been inconclusive. Studies from the 1980s found that an increase in Medicaid payments lowered quality in the presence of supply constraints such as certificate-of-need (CON) and bed construction moratoria.4 More recent work has found a small positive relationship between payment rates and quality.5 However, previous studies have been limited to facility-level measures of quality, which have not allowed for risk adjustment at the individual level. Furthermore, prior studies have often relied on proxies for quality such as staffing levels or deficiencies. In this study we used three recently developed nursing home quality measures—pain, high-risk pressure ulcers, and use of physical restraints—from the resident-level Minimum Data Set (MDS). Using a recent cross-section of all certified U.S. nursing homes, we examined the relationship of Medicaid payment and each of these quality measures, controlling for market and selected facility factors.
Data. We characterized nursing home quality by aggregating MDS resident-level assessments to the facility level. Individual MDS resident assessments are performed on admission, upon significant change in status, and at least quarterly.6 For this study we used the most recent (nonadmission) assessment for each nursing home resident from the second quarter of 1999. The quality measures we used were validated as part of the federal government’s recent effort to develop a new system of public reporting aimed at improving nursing home quality.7 Information about the facilities was obtained from the On-Line Survey, Certification, and Reporting (OSCAR) system for October 1998–December 1999. The OSCAR data elements we used are generally considered accurate and reliable.8 For the purposes of this study, all hospital-based facilities are eliminated from the data file, because Medicaid is an insignificant payer in such facilities. We used three other data sources: First, we merged the nursing home data with aggregate county-level data from the Bureau of Health Professions (BHPr) Area Resource File. Second, we obtained state-level Medicaid reimbursement methods and rates from the most recent edition of the State Data Book on Long-Term Care Program and Market Characteristics.9 Finally, we linked the area wage indexes from the Centers for Medicare and Medicaid Services (CMS) with the nursing home data. Measures of nursing home quality. This study used three separate measures of nursing home quality: pain, pressure ulcers, and use of physical restraints. It is important to approximate quality using multiple measures, given the lack of correlation across different measures.10 These three measures were selected because of their salience, recent evidence supporting their validity, and consensus regarding their importance (based on the number of papers on each of these topics and the extent to which the public is concerned about them).11 Finally, precisely because these measures are uncorrelated, if Medicaid payment rates are independently related to three orthogonal measures of quality, a case for the pervasiveness of the relationship of payment rates and quality would be established. We characterized "inadequate pain management" as the prevalence of "moderate pain at least daily" or "horrible/excruciating pain at any frequency in the last seven days" among long-stay residents. The measure was risk-adjusted to account for residents’ cognitive functioning as measured by daily decision making. For "high-risk" pressure ulcer prevalence, we created an indicator, whose denominator was limited to patients with any of the following situations: either bed mobility or transferring problems (that is, requiring extensive assistance or total dependence), secondary diseases related to malnutrition (International Classification of Diseases, Ninth Revision, or ICD-9 codes 260, 262, 263), comatose, or end-stage disease. These risk factors are largely independent of facilities’ treatment practices and therefore minimize the potential for overadjustment. For restraint use, the MDS-based measure excludes restraint use that is justified by the presence of a physician order supporting the need for restraints for the resident’s own safety. No risk adjustment of the restraint measure was applied because the procedure is totally at the discretion of the facility and practice expectations. Facility-, market-, and state-level explanatory variables. The key explanatory variable of interest in this study was the state’s Medicaid payment level in 1998.12 The MDS quality measures were collected during the second quarter of 1999, which provides a lag between the Medicaid payment and the development of potential quality problems. In addition to the Medicaid payment level, the analysis also controls for whether the state employs a case-mix Medicaid payment system.13 We included a series of other exogenous facility- and market-level variables as covariates in the regression analysis. The facility-level covariates were nonprofit status, government-owned status, chain membership, and total number of beds.14 The county-level variables were the median per capita income, CMS hospital wage index, population of people over age sixty-five per square mile, a dummy variable identifying the most restrictive quartile of markets (with fewer than 4.61 empty beds per 1,000 noninstitutionalized elderly people), and the Herfindahl-Hirschmann Index (HHI) of market concentration for private-pay residents.15 Statistical analysis. The unit of analysis in this study was the nursing home. For each quality measure, we performed regression analyses with the state-level Medicaid payment rate and a set of control variables. In all analyses, the control variables included the home, market, and state factors listed above. The dependent variable in each regression equation was treated as a continuous measure.
We initially identified 15,128 freestanding nursing facilities using the OSCAR system (see Exhibit 1
Variation in Medicaid payment rates. Medicaid payment rates vary considerably across states (Exhibit 2  ). The mean Medicaid payment rate was $93.44, with a standard deviation of $21.76. Alaska, the District of Columbia, Massachusetts, New York, and Maine had the highest payment rates; Arkansas, Oklahoma, Louisiana, Texas, and Iowa had the lowest.
Variation in quality. Quality also varied greatly across states (Exhibit 2  ). The national average pain rate was 13.08 percent, with Utah reporting the highest rate and Mississippi reporting the lowest. The national mean high-risk pressure ulcer rate was 15.27 percent, with New Jersey reporting the highest rate and North Dakota reporting the lowest. Finally, the national mean physical restraint rate was 9.66 percent, with Arkansas reporting the highest rate and Nebraska reporting the lowest. The three quality measures were not well correlated across facilities or states, which is consistent with other studies documenting the multidimensional nature of nursing home quality.16 For example, the correlation between the daily pain and high-risk pressure ulcer measures is 0.05, the correlation between daily pain and physical restraints is –0.004, and the correlation between high-risk pressure ulcers and physical restraints is 0.007. The multidimensional nature of nursing home quality makes the use of multiple quality measures in this study particularly important.
Association between payment and quality.
The results of the regression analysis analyzing the relationship between state-level Medicaid payment rates and nursing home quality are summarized in Exhibit 3
This study examined the relationship between states’ investment in nursing homes via the level of Medicaid payment and multiple resident-level measures of nursing home quality. We observed a positive relationship between the state Medicaid payment rate and two different quality indicators: high-risk pressure ulcers and physical restraints. These two quality measures are not well correlated with one another, which implies that greater Medicaid investment translates into better quality across two quite different domains of quality. Our counterintuitive finding concerning pain prevalence is perhaps attributable to known variations in pain ascertainment practices within and across states. Recent work has documented how facilities with an active hospice program are more likely to rate a higher proportion of their patients as being in pain (independent of individual hospice status), compared with facilities that do not operate such a program.18 The hospice program may be training staff to be more attuned to pain assessment, thereby making the facility look relatively worse on the pain prevalence indicator. Our finding of a positive relationship between payment rates and pain prevalence may be related to ascertainment bias, whereby facilities with more resources are using those resources to improve documentation of an often neglected domain of quality. Policy implications. The identification of clear empirical links between Medicaid payment and two important quality measures has timely policy implications. Recent state budget shortfalls will almost certainly give state policymakers the impetus to revisit Medicaid spending. States may look toward nursing home care, which accounts for a sizable portion of state Medicaid spending, as a potential area to cut. Based on our results, quality of care may suffer as a result of proposed efforts to reduce Medicaid nursing home payments. One of the long-standing concerns of advocates for nursing home consumers is the issue of profit taking. In a sector dominated by for-profit entities, the concern is that increased payments do not always yield improved quality, especially in areas with less-than-competitive markets or in cases where consumers have imperfect information on quality.19 However, this study found that increased payment was associated with higher quality for both for-profit and not-for-profit facilities. Our findings of an independent relationship between payment and quality, controlling for market characteristics, suggest a need to reexamine the "black box" that exists between payment and quality. How precisely do changes in Medicaid rates interact with staffing and infrastructure resources to yield changes in residents’ outcomes? Identifying these mechanisms is important if state funds are to be more effectively directed toward adequate nursing home care. That said, the results of this study indicate that the overall munificence of states’ Medicaid programs, not necessarily more staffing, better training, or improved processes of care, is at the root of some of the observed differences in outcomes. In the context of a recent call for minimum staffing standards in nursing homes as a means of improving quality, it should be noted that there may be multiple ways to achieve better nursing home quality, and forcing providers to invest more resources in a particular area may reduce innovation.20 This is not to say that "earmarking" Medicaid dollars through a minimum staffing standard is not good public policy, but rather that the federal and state governments need to strike the appropriate balance between market-based and regulatory activities in the nursing home sector. Instead of linking payment rates to staffing or processes of care, a better reimbursement system might provide direct incentives for higher quality and improved outcomes. A controlled experiment in San Diego found that the use of monetary incentives had beneficial effects on the health of nursing home residents.21 Moreover, nursing homes in the experimental group admitted patients with more severe disabilities, and the average length-of-stay was shortened. Although the cost of developing and administering such a program may be considerable, earmarking payments for better outcomes might be one innovative way to encourage nursing homes to compete on the basis of quality for the care of Medicaid residents. There is another black box worth exploring: how states channel funds to the various modes of long-term care offered in communities. As state legislatures attempt to manage Medicaid spending, decisions concerning payment for long-term care services often take place within a zero-sum framework.22 The finding that Medicaid nursing home payments cannot be reduced without affecting the quality of care must be considered in light of the demands to develop alternatives to that care. Any pledges of support for maintaining current payment levels to nursing homes—let alone increasing them—are likely to come at the expense of maintaining or expanding home and community-based alternatives. On the other hand, as legislative efforts to direct scarce resources into home and community-based programs are pursued, the frailest and neediest recipients of long-term care (for whom nursing home placement is often unavoidable) face greater risk of receiving poor-quality care in settings where Medicaid payments are inadequate. Medicaid nursing home payment policy is also related to other public and private funding sources. Although Medicare is not a substitute for Medicaid in funding long-stay nursing home care, Medicare complements state Medicaid programs by funding acute nursing home care. In 2001 Medicare accounted for 11.7 percent of total nursing home spending.23 A number of facilities care for both Medicaid and Medicare recipients, and cost-based Medicare payments have often subsidized prospective Medicaid payments.24 However, the adoption of a prospective payment system (PPS) for Medicare nursing home care in 1998 reduced homes’ ability to use Medicare dollars to subsidize the care of Medicaid recipients. During this same time period, we have seen growth in managed care coverage for nursing home care, which may also limit the degree of cross-subsidization available from private-pay residents. Thus, in an environment where states are considering decreasing Medicaid payments and Medicare has become less generous over time, the role of Medicaid as payer may be more important than ever. Study limitations. Our study was limited in several ways. Although our three quality indicators are distinct in nature, they cannot fully encompass the multidimensional nature of nursing home quality. Moreover, they focus on the technical aspects of care but do not capture the quality of life in the facility, an important dimension of quality. Second, when examining the effect of public policies on behavior, we often prefer a panel approach because other factors, unaccounted for by the analysis, might affect our relationship of interest. In this case, a state-specific factor could be correlated with both Medicaid reimbursement decisions and nursing home quality. Finally, although both the MDS and OSCAR data have been tested and used repeatedly for research purposes, both systems are collected for government oversight purposes, which may raise concerns about the accuracy and interfacility reliability and comparability of the measures. This paper has taken the novel approach of merging resident-level quality information and state-level Medicaid payment rates to examine whether greater state munificence was associated with better quality. We find evidence that higher payment rates were associated with higher quality for two of three quality indicators. Our findings document the need for states to consider quality trade-offs when setting Medicaid payment levels, especially in an era of diminished state resources.
David Grabowski (grabowski{at}hcp.med.harvard.edu) is an assistant professor in the Department of Health Care Policy, Harvard University. Joseph Angelelli is an assistant professor in the Department of Health Policy and Administration, Pennsylvania State University, in University Park. Vincent Mor is chair of the Department of Community Health, School of Medicine, Brown University, in Providence, Rhode Island. David Grabowski and Joseph Angelelli received funding for this project from the Agency for Healthcare Research and Quality (1 R03 HS11702-01). Vincent Mor received support from a National Institute on Aging MERIT award (AG-11624). The authors acknowledge the assistance of Chris Brostrup-Jensen and Yuwei Cang in preparing the data for analysis and Jeroan Allison for providing helpful comments on an earlier draft of this manuscript.
| |||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Results
