| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Address correspondence to Robert L. Kane, MD, University of Minnesota School of Public Health, D351 Mayo (MMC 197), 420 Delaware Street SE, Minneapolis, MN 55455. E-mail: kanex001{at}umn.edu
 |
Abstract |
|---|
 TOP Abstract BackgroundMinnesota's Quality-Based...DiscussionReferences |
|---|
Key Words: Nursing homes • Pay for performance • Quality • Payment • Accountability
Cries for better quality have failed to sufficiently motivate providers because there is no evidence that payers are willing to meet the higher costs often associated with better quality. Recently the press for better care has led to calls for pay for performance (Birkmeyer & Birkmeyer, 2006; Boyd et al., 2005; Forrest, Villagra, & Pope, 2006)
The belief in a market-driven approach fueled by informed consumers has encouraged many states to summarize quality measures in report cards (Castle & Lowe, 2005) and the Centers for Medicare and Medicaid Services (2004) to establish a Web site that reports on quality measures (Nursing Home Compare). Although consumers often fail to use such information when making utilization decisions (Castle & Lowe, 2005; Chernew & Scanlon, 1998; Hibbard, Slovic, & Jewett, 1997; Wicks & Meyer, 1999), consumers do appear to heed quality reports about nursing home care (Castle, 2003), the results of Health Plan Employer Data and Information Set reports (Beaulieu, 2002; Chernew, Gowrisankaran, McLaughlin, & Gibson, 2004; Scanlon, Chernew, McLaughlin, & Solon, 2002; Wedig & Tai-Seale, 2002), and satisfaction reports on Medicare health maintenance organizations (Dafney & Dranove, 2005). Another strategy is to create direct payment incentives that use quality performance as the basis for bonus payments to nursing homes (Geron, 1991) or to use quality as an intrinsic component in the calculation of payment rates.
Pay for performance is not necessarily incompatible with the other movements to improve nursing home quality, such as continuous quality improvement or culture change. Rather than addressing the root causes of quality problems, pay for performance provides an environment that rewards quality improvement. As a result, it can enlarge the gap between good and poor homes if good homes use additional revenue from pay for performance to further increase quality while poor homes lack the resources to improve their care. With very strong incentives, some pay for performance approaches might even provide motivation for poorer homes to leave the system and better ones to expand.
The case-mix reimbursement models adopted by the Medicare program and many state Medicaid programs may not actively promote good quality of care. On the one hand, if resident needs determine payment, then providers caring for higher acuity residents will have more resources at their disposal. On the other hand, there is no direct incentive to improve health or functional outcomes because payment typically is adjusted downward as acuity declines. The challenge, then, is to develop a payment approach that can deal with differences in resident acuity and also reward quality by paying nursing homes more if they produce better quality of care. Presuming problems in measuring quality, economic theory suggests that optimal payment schemes should combine a fixed payment with cost-based reimbursement (Ellis & McGuire, 1988). The size of the quality incentive must be sufficient to enable and encourage the nursing homes to undertake the improvements in care needed to achieve it; it must make good business sense as well as be socially desirable.
In response to a state legislative mandate, the State of Minnesota is currently developing a quality-based payment strategy that combines comprehensive measures of nursing home quality with an acuity-adjusted rate-setting approach that contains financial incentives for higher quality and greater efficiency. The new payment system is intended to (a) provide incentives to enhance quality of life and quality of care, while recognizing cost differences in the care of different types of populations, including subacute care and dementia care; (b) establish payment rates that are sufficient without being excessive; and (c) allow providers maximum flexibility in their business operations.
This article describes the Minnesota quality-based payment system, the research that went into its development, and plans for its implementation. Designing the system required confronting fundamental questions about nursing home quality and methods to achieve it. What does nursing home quality mean? How much importance should be placed on different dimensions of quality as viewed by different stakeholders? Once defined, are reliable and valid measures of quality available? Moreover, what does it take to produce better quality, and can it be done efficiently? Assuming that there are satisfactory answers to these questions, how should the Medicaid payment system be restructured to reward both quality and efficiency? Finally, what can be done to ensure providers respond positively to the payment incentives introduced?
The Minnesota Department of Human Services contracted with the University of Minnesota (with subcontracts to Myers and Stauffer LC and the University of Missouri at Kansas City) to provide the technical support for the project. The project represented a partnership between a state government known for its innovative programs, academic researchers in long-term care, and the nation's largest long-term-care accounting and consulting firm.
|
Background |
|---|
|
TOPAbstractBackgroundMinnesota's Quality-Based...DiscussionReferences |
|---|
The working definitions and technology for measuring nursing home quality have improved considerably since that time. Experts have developed nursing home quality indicators (QIs) and have applied them widely for regulation, quality improvement, and public reporting. The QIs rely on data elements from the Minimum Data Set (MDS), a standardized assessment instrument administered to residents at admission and at least every 90 days thereafter. Although the QIs have been criticized (Arling, Kane, Lewis, & Mueller, 2005; Mor et al., 2003; Zimmerman, 2003), they represent a comprehensive set of measures covering resident health and functioning that can be risk adjusted and aggregated to the facility level. One of the weaknesses of the MDS and the QIs is their absence of items that directly appraise resident quality of life. An instrument that assesses quality of life for nursing home residents, and effectively complements the clinically oriented MDS items, is now available (R. A. Kane et al., 2003).
Nursing home payment systems have also undergone change in the past decade with the widespread adoption of case-mix reimbursement based on the RUG-III (Resource Utilization Groups) case-mix classification system (Fries et al., 1994). The RUG-III system uses information from MDS assessments to classify residents into case-mix groups that are assigned an index score according to their relative acuity or costliness. Medicare and Medicaid programs in the majority of states, including Minnesota, have adopted a RUG-III-based nursing facility payment system. The relationship between case-mix payment and quality has been a subject of debate (Fries & Ikegami, 1994; R. A. Kane, 1994). Although case mix can potentially offer perverse incentives, off-setting factors minimize this effect. Providers face increased costs, regulatory sanctions, or other negative consequences if their poor care leads to a decline in resident health. Yet, there are no inherent incentives in case-mix payment systems or more traditional cost-based systems to offer good-quality care and better resident outcomes. Insufficient research has been conducted into nursing home case-mix reimbursement systems and care quality, at least in the 10 years since the introduction of RUG-based systems. Implementation of RUG-III-based Medicaid payment during the 1990s appeared to increase access for more dependent residents while leading to no significant change in care quality (Arling & Daneman, 2002; Grabowski, 2002).
A few states (e.g., Illinois and Florida) tried to introduce quality incentive payments during the 1980s; however, they abandoned them (Geron, 1991). More recently, the Iowa Medicaid program introduced a point system in which facilities could obtain a bonus (up to 3% of median cost) if they scored a sufficient number of points on a series of 10 accountability measures. Nursing homes receive points for high resident satisfaction, high staffing levels, high employee retention rates, deficiency-free surveys, low percentage of administrative costs, high Medicaid utilization, and other factors. During the 3-year period since the system was implemented (2001–2004), facilities have shown modest improvement in areas such as resident satisfaction, staffing, and employee retention (Arling et al., 2005).
As with many states, the payment for individual nursing homes in Minnesota reflects many historical artifacts. Current payment rates for the majority of facilities were set initially in 1994–1995 according to facility costs at that time, with an annual cost of living adjustment in most, but not all, years. The Minnesota case-mix system introduced in 1985, and more recently the RUG-III case-mix system begun in 2002, have recognized changes in acuity; however, very little allowance has been made for other operational changes in the past 10 years, and, as we will demonstrate here, the relationship between payment rate and quality is poor.
|
Minnesota's Quality-Based Payment Strategy |
|---|
|
TOPAbstractBackgroundMinnesota's Quality-Based...DiscussionReferences |
|---|
Minnesota has had a case-mix reimbursement system since 1985, although the current case-mix system adjusts only payments for nursing care. Under the quality-based payment approach, as with the current rate-setting system, nursing and other direct care costs are acuity adjusted with a RUG-III case-mix index. The facility's per diem payment rates are set separately for two cost components: nursing and direct care, and administration and operating. Rates are determined by the facility's quality score, ranging from 0 to 100, and its cost relative to a statewide efficiency target and the costs of other facilities in the state. Facilities with high quality and low costs (i.e., below the efficiency target) are paid a rate that exceeds their costs; facilities with low quality and/or high costs are paid a rate below their costs. Minnesota, like other states, has had considerable experience in collecting cost information and using it in the payment process. The state devoted much of its effort in designing the new payment system to defining and measuring quality of care. The Minnesota quality-based payment model applies to Medicaid nursing home rate setting; however, because Minnesota is one of only two states with rate equalization (whereby the privately paid rate cannot exceed that for Medicaid), the rate applies to a facility's private-pay residents as well as to its Medicaid residents.
Quality Score and Measures
The basic principle of the Minnesota model is that quality should affect the base price paid for care as well as the amount each facility ultimately receives. For reasons of political expediency, experts decided to develop the initial quality-scoring approach using readily extant data, with the expectation that the components could be amended as new data became available. Under the quality scoring system, each facility is assigned between 0 and 100 points according to six quality measures derived from state administrative data systems. (Table 1). Facilities must meet minimum thresholds to receive points on a given component, thus ensuring that poor quality is not rewarded.
|
Although the design of the quality elements was admittedly pragmatic, it can be cast in the familiar Donabedian (1988) framework of structure, process, and outcome. There was substantial disagreement among stakeholders about the relative value of these elements. Although one might argue that being able to assess outcomes should mitigate the need (even the advisability) of measuring structure and process, many stakeholders clung to the belief that staffing was a critical element and that current measures did not satisfactorily capture outcomes. Thus, the initial version heavily favored a structural component: staffing.
The weights assigned to each quality component and even the inclusion of components can be readily changed as more information becomes available or more attention is drawn to a given area. The current components include staff retention (25 points), staff turnover (15 points), use of pool staff (10 points), nursing home QIs (40 points), and survey deficiencies (10 points). The QI measure was based on facility rates for 25 QIs derived from the University of Wisconsin Center for Health Systems Research and Analysis QIs (Zimmerman et al., 1995), the Abt-Brown Mega QIs (Abt Associates, 2004), and the Centers for Medicare and Medicaid Services QIs (2004). Facilities are assigned points according to their ranking on risk-adjusted QI rates; however, they must reach a minimum quality threshold in order to receive points on a QI. Because stakeholders were generally dubious about survey deficiencies as reliable measures of quality, the score considers only very poor performance on selected care-related F-tags and has relatively little weight.
Data on two important measures, quality of life and resident satisfaction, have recently become available. A survey firm collected data on both topics with in-person interviews from a random sample of residents in all Minnesota nursing homes, using the instrument created for the Centers for Medicare and Medicaid Services (R. A. Kane et al., 2003). Subsequent versions will include that domain. The last column of Table 1 shows the proposed new weighting system.
Quality Score and Cost in the Current System
Figure 1 shows the distribution of facilities based on their quality scores. The facilities' quality scores were roughly bell shaped. Although some facilities had low quality scores (less than 50) and others had very high scores (90 and higher), most facilities fell in the range between 55 and 85. Facility per diem costs varied widely. Acuity and labor-market-adjusted per diem Medicaid rates were a median of $134 per day in 2004, and the interquartile range was nearly $20. Figure 2 shows a scatter diagram between quality score and per diem cost. There was virtually no relationship between the two measures; they had a nonsignificant negative correlation (–.068). Likewise, the correlation between facilities' case-mix adjusted Medicaid per diem payment rate and their quality scores was nonsignificant and negative (–.081).
|
|
The first step in rate setting was to divide operating costs into major categories: direct care costs (i.e., nursing and other care-related services) and support services costs (i.e., dietary, laundry, administrative, and other). We then adjusted costs for facility case mix (direct care costs only) and regional wage rates so that we could compare per diem figures across facilities with different resident acuity and regional labor markets. We then placed facilities into tiers from low to high cost depending on their rankings. We cross-tabulated the cost tiers with similarly constructed quality tiers in a large table or matrix. Table 2 shows a simplified version of the matrix. The number of tiers and rate-setting percentages are only an example; the design of the system and parameters are still under negotiation. The percentages in each cell represent the facility's Medicaid payment rate as a percentage of its costs. Facilities in the middle cells (average quality and cost) would have payment rates at about 100% of cost, facilities in the high-quality/low-cost quadrant would have rates at > 100% of cost, and facilities in the low-quality/high-cost quadrant would have rates at < 100% of cost. Rates in the other quadrants would be set with percentages biased toward higher quality (i.e., high-quality/high-cost facilities would get higher percentages than low-quality/low-cost facilities). One could expand the number of cost and quality cells to as many as 100 x 100 in order to be responsive to even small changes in quality or efficiency; however, this would make rate setting more complex and harder for providers to grasp.
|
|
Discussion |
|---|
|
TOPAbstractBackgroundMinnesota's Quality-Based...DiscussionReferences |
|---|
The success of this approach, and its eventual use as a basis for nursing home payment, will depend on all stakeholders accepting the idea that the score does reflect the important dimensions of nursing home quality. There will likely never be full consensus around the weights assigned to individual components, but we have pursued efforts to tap a range of opinions.
This approach relies on both empirical and political components. When we presented the stakeholders with a detailed analysis of the literature showing a weak relationship between staffing and quality, they nonetheless clung to the notion that staffing was a key element of quality. In the second iteration, staffing received slightly less emphasis (from 50 to 35 points)
The development of this approach did not directly include two groups of stakeholders: residents and their families. Instead, we relied on various consumer advocates, including the organization that represents nursing residents and sponsors resident councils, the ombudsmen, the Alzheimer's Association, and AARP.
We did not test the validity of the overall quality score, although the components represent measures that have been used before and were subjected to validity testing (e.g., QIs and quality of life/resident satisfaction) or were based on administrative data (e.g., staffing level and retention rate). The QIs were subjected to more risk adjustment than were those in other applications, such as Nursing Home Compare. The validity of the overall score and its components depends heavily on face validity, especially the values and opinions of key stakeholders.
Winners-and-losers analyses have shown that the new approach does not favor any one subgroup over another, but the new approach does mean a redistribution of resources across homes. The need for facilities to report annual costs and for the state to audit these costs also became a matter of contention in the development of this approach. Finally, the proposed model was difficult for many providers to understand. The nursing home trade associations have publicly expressed support for the underlying concept but have lobbied to slow its implementation, claiming a need to work out operational details. Legislators are loath to support any program that might mean fewer revenues to their constituent nursing homes.
Because of opposition to the proposed pay for performance model, the Minnesota state legislature enacted a simpler model involving a bonus payment of up to 3% of the daily per diem rate depending on a facility's quality score. This bonus approach has a less powerful incentive and does not reward greater efficiency; that is, the bonus depends only on a provider's quality score regardless of the expenditures required to produce higher quality. We view this as an interim step that the full pay for performance model will hopefully replace in time.
If providers are to endorse and use a pay for performance approach, they must understand it on at least three levels: (a) They must accept the measures and the calculations of quality as valid; (b) they must understand the way they are used to create a payment rate; and (c) they must be able to develop and implement strategies to improve quality, as measured through investments that will result in rate increases sufficient to justify those expenses. Gaining acceptance of the measures relies on using measures with a pedigree and generating weights after broad input from stakeholders. Information sessions about the quality scoring approach are already being held across Minnesota and have thus far been well received by the providers. Clarifying how the measures are translated into a payment system is more complicated. Applying the matrix system is somewhat complicated, although the concept is straightforward. Nursing facilities are paid a higher proportion of their costs (even exceeding 100%) if their quality is better. The proportion depends on their quality score and how their costs compare to those of other nursing homes. No facility with poor quality will receive its full costs.
In keeping with the idea that every facility should compete only with itself, the initial quality scores form a base year against which future scores are normed. Thus, facilities can improve their income by improving their quality. Because one of the principles underlying this approach is the recognition of absolute improvement as opposed to relative improvement based on comparing one home against another, all of the quality scores are based on fixed standards. For example, QI incidence or prevalence rate thresholds in 2005 will be used for point assignment in subsequent years. In theory, all facilities could improve their QI rates above the thresholds required for the highest points. The decision to allow every home to obtain a quality payment raises the possible costs. If every home achieved a strong score, the state's coffers could be threatened. Presumably homes doing an especially bad job would be at greater risk than good-quality homes of losing money.
We are currently engaged in a program to help providers understand the relative costs of various quality improvement strategies that target different quality components. In many cases, facilities can make great strides at low cost by addressing leadership issues. In other cases, they must make tangible investments. Although the evidence that links quality improvement with specific actions is often sparse, some approaches offer promise (Tsilimingras, Rosen, & Berlowitz, 2003).
The equally challenging element involves developing a method to actively engage and inform consumers about quality. The degree that consumers will ultimately respond to report cards remains to be determined.
Making quality information consumer friendly requires that it be meaningful and simple. We have approached the former by creating a computer-based system that allows consumers to indicate the quality elements of greatest salience to them. The computer can then prioritize nursing facilities on the basis of the most salient quality parameters, while allowing the consumer to obtain full information on each facility of interest as a second step. We are working on developing ways to present quality scores that recognize the variation in reporting and show only differences that are significant.
In order to decide which level and type of quality investment will yield the best return, facilities must understand what is required to improve quality. Facilities will likely vary in their risk-taking philosophies, but all will be served by better information on the likely return on investment in quality. Some investments appear to offer a more likely return than others, but they may require a substantial investment. For example, adding direct care staff or increasing wages and benefits may reduce turnover, increase retention, or yield better QI scores yet at a potentially high cost. By contrast, a continuous quality improvement project focused on a QI or a strategy to eliminate the use of nurse pools may provide improved quality at a lower cost.
The nursing home industry and other major stakeholders have agreed on the general principles of the new payment system as well as its basic design. However, they have had difficulty agreeing on many of the system's technical details, such as operational definitions, weights, and scoring thresholds for different measures. On a more fundamental level, the nursing home industry has taken the position that base Medicaid funding must increase to cover costs of producing higher quality and to cushion the effects of rate decreases on low-quality facilities. Moreover, the industry wants assurance that if it produces higher quality, in aggregate, the future Medicaid payment system will recognize this investment. With the vicissitudes of state tax revenue and pressure for spending in other areas, policy makers have been hesitant to increase current Medicaid nursing home funding or build in future budget increases. Nonetheless, the concept of linking payment to care quality may be taking hold.
|
Footnotes |
|---|
1 University of Minnesota School of Public Health, Minneapolis.
2 School of Public and Enviromental Affairs, Indiana University Center for Aging Research/Regenstrief Insitute, Indiana University, Purdue University, Indianapolis.
3 University of Minnesota School of Nursing, Minneapolis.
4 Nursing Facility Rates and Policy Division, Minnesota Department of Human Services, St. Paul.
Decision Editor: Nancy Morrow-Howell, PhD
Received for publication April 7, 2006. Accepted for publication July 3, 2006.
|
References |
|---|
|
TOPAbstractBackgroundMinnesota's Quality-Based...DiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  P. Kemper, B. Heier, T. Barry, D. Brannon, J. Angelelli, J. Vasey, and M. Anderson-Knott What Do Direct Care Workers Say Would Improve Their Jobs? Differences Across Settings Gerontologist, July 1, 2008; 48(suppl_1): 17 - 25. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
