| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Department of Public Health Sciences, Wichita State University, Kansas
b University of Kansas Medical Center, Kansas City, Kansas
c Social & Behavioral Sciences, University of California, San Francisco
Correspondence: James Swan, PhD, Department of Public Health Sciences, Box 152, Wichita State University, 1845 N. Fairmount, Wichita, KS 67260-0152. E-mail: swan{at}chp.twsu.edu.
Decision Editor: Laurence G. Branch, PhD
 |
Abstract |
|---|
 TOP Abstract BackgroundMethodsResultsDiscussionReferences |
|---|
Key Words: Medicaid • Nursing facility • Ancillary services
Medicaid nursing home expenditures have recently grown at a slower rate than in the past (Levit et al., 2000), and well below the average growth for national health expenditures; but expenditures are large enough to give policy makers ample reason for concern (
Coleman 1996
). Nursing home expenditures were $87.8 billion in 1998, and these were expected to grow to $148.3 billion in 2007 (Levit et al., 2000). Medicaid covered 46% of nursing facility payment in 1998, for 60 to 65% of residents (
HCFA 2000a; Levit et al., 2000;
Short, Feinleib, and Cunningham 1994).
Medicaid nursing home reimbursement methods and per diem reimbursement rates are of great importance in part because they influence the costs of providing care (
Harrington and Swan 1987). However, these policies serve goals beyond cost constraint: equitable payment to providers, access for Medicaid eligibles, and quality of care received. Concern is heightened because of the myriad of ways in which these goals may conflict (
Gertler 1991;
Holahan and Sulvetta 1989;
Swan, Harrington, and Grant 1988;
Swan, Harrington, Grant, Leuhrs, and Preston 1993). Of particular concern is the issue of adequate payment, that which is adequate to promote access and cover the costs of higher quality care. Medicaid reimbursement rates were on average substantially below Medicare nursing home reimbursement, which was $301 per day in 1997 (
American Health Care Association 1999) compared with only $95.09 for Medicaid (
Swan et al. 2000), although it should be noted that the Medicare data include much payment to hospital-based facilities, which are more expensive.
A new Institute of Medicine report (
Wunderlich and Kohler 2000) raised concerns about whether state Medicaid reimbursement rates are adequate to ensure high quality of nursing home care. The report pointed to the quality of care problems in many nursing homes and the inadequate staffing levels and urged new research on the relationship between reimbursement rates, access, and quality. The Health Care Financing Administration (
HCFA 2000b) also reported serious understaffing in nursing homes throughout the United States, which was found to be related to poor quality of care. State legislators have also been interested in comparing their Medicaid reimbursement rates with other states and some have made new efforts to increase rates in order to improve staffing levels. For example, a 1999 survey of states found that seven states implemented wage pass-throughs for nursing home workers in their 1999 Medicaid reimbursement rates (
HCFA 2000b).
Nursing facilities depend highly on Medicaid as a payer (Levit et al., 2000). They additionally rely on state Medicaid programs to pay rates adequate to cover costs of care (
Batavia, Ozminkowski, Gaumer and Gabay 1993;
Reid and Coburn 1996;
Stone and Reublinger 1995). Because of impacts on access and quality, Medicaid nursing facility reimbursement policies are of intense interest to the consumer and consumer advocacy groups (
Sparer 1993;
U.S. General Accounting Office 1995). Whatever their relative interest in these varied goals, policy makers are understandably concerned with the wide variation across states in per diem rates, particularly because the ratio of the highest average state rate has been four and six times higher than the lowest average state rate every year for over two decades (
Swan et al. 2000).
Comparison of rates is made complex, however, by such factors as interstate differences in what is included in daily rates. States vary widely in whether they include various ancillary services and commodities in daily rates versus covering the costs of such ancillaries outside of the rates (
Swan et al. 2000). This study focuses on the question of whether ancillaries are included in the Medicaid nursing facility per diem rate and how ancillaries influence the interstate comparison of such rates. It does not attempt to establish overall costs for a day of care, which would necessitate different estimation techniques and consideration of additional elements beyond the list of ancillaries considered here.
|
Background |
|---|
|
TOPAbstractBackgroundMethodsResultsDiscussionReferences |
|---|
;
Sutter and Keough 1999). Average costs of providing a given type of ancillary (e.g., physical therapy) may vary considerably across facilities and providers, depending particularly on both volume and intensity of use of the ancillary (
Sutter and Keough 1999). Thus, although some types of providers (e.g., hospitals) may fare quite well in being separately reimbursed for the provision of ancillaries (
Larkin 1999;
Pretzer 1999), nursing facilities may not do so well. For example, low reimbursement is a reason physicians are reluctant to provide care in nursing facilities to their patients who are residents, even though regular monitoring by physicians has the potential for improving the quality of required resident assessments (
McCartney and McCartney 1997).
Ancillary services are important to nursing facility care. "Ancillary" is defined as "being auxiliary or supplementary" (Merriam Webster Medical Dictionary, 1997). An ancillary service for nursing facility care is any service considered auxiliary or supplementary to nursing care. Such ancillaries include various therapies, drugs, supplies, and equipment (The Balanced Budget Act of 1997, 1999), and an array of other items. Each state can have its own list of ancillaries, and searches have not dislodged any HCFA designation. The list used here resulted from researchers' (
Swan et al. 1988) presenting a list of seven ancillaries found to be commonly referenced in state Medicaid reimbursement policies.
When ancillaries are included in rates, however, there can be considerably greater variation, combined with no assurance that average costs are even considered. This is particularly true in recent years in the Medicaid program, where concerns with constraining costs may outweigh concerns with adequately paying for each type of care (
Swan and Pickard in press).
Policies to include ancillaries in rates have effects other than cost. In particular, including ancillaries in rates may be a disincentive to provision of the ancillaries, because a facility receives no additional revenue from providing a given ancillary already included in its daily rate. This is of particular concern because of the relatively low rate of provision of such ancillaries as rehabilitative therapies in the United States compared to other countries (
Berg et al. 1997). Likewise, policies related to reimbursement of prescription drugs, and their influence on actual use of drugs in nursing facilities, are important. This is of concern especially in light of the estimates that the costs in morbidity and mortality associated with drug treatment in nursing facilities may outweigh the actual costs of drug treatment (
Bootman, Harrison, and Cox 1997). Few studies have examined ancillary rate policies and their effects on nursing homes.
Although there are wider implications of reimbursement policies for ancillaries, the focus here remains on the implications for reimbursement systems. The question is how inclusion of ancillaries in the per diem nursing facility rate influences the interstate comparison of such rates. This approach cannot hope to account for the costs of the ancillaries, instead focusing on evidence of presumed state adjustment of rates to cover such costs.
|
Methods |
|---|
|
TOPAbstractBackgroundMethodsResultsDiscussionReferences |
|---|
), the contents of cost reports vary widely and seldom contain the wealth of detail that would allow the costing of specific ancillaries. Consequently, the approach used here is to attempt to model the costs of ancillaries through the use of panel regression analysis over a multiyear period, with appropriate adjustment for correlated error over time.
Data for a 12-year period, 1987–1998, were drawn from a series of national surveys of Medicaid long-term care reimbursement. The reimbursement rate data were collected from a series of telephone surveys conducted between 1983 and 1999 to collect data from 1987–1998 on Medicaid long-term care payment methods. A combination of telephone interviewing and mail surveys elicited information to clarify responses. The data in this study are collected at the end of the state fiscal year to represent actual reimbursement rates for the year. In contrast, HCFA studies (
HCFA 1992; HCIA, 1992) report information derived from state Medicaid plans and amendments, estimating ahead of time what rates will be each year; they are not adjusted when states make changes in their rates during a year. Thus, the rates presented here are outputs—average rates determined to have been used during a given year. These may differ somewhat from but be more accurate than estimates based on state plans.
Employing a random-effects model, estimated using the PANEL option in LIMDEP (
Greene 1995), we separately estimated effects on state average daily Medicaid nursing facility rates in constant dollars and in actual dollars of the inclusion of a set of ancillary services in those rates. Ancillaries considered were: physical therapy, occupational therapy, prescription drugs, nonlegend drugs, durable medical equipment (DME), medical supplies, and physician services. Table 1 reports what states included each of these ancillaries in their rates in 1998.
|
Because state rate-setting policies (basic method and use of case-mix) may be endogenous with inclusion of ancillaries, equations were estimated both including and excluding such policy measures. No matter which measures were included in an equation, the adjustment of rates for the inclusion of ancillaries uses only the coefficient on the ancillaries.
By the logic of the model, all coefficients on ancillaries should be positive, representing additional per diem payment to cover the costs of providing these ancillaries. Some coefficients may be found to be negative, however, perhaps even significant. It does not seem logical that rates would have been reduced when an ancillary was included in a rate. It does seem likely, however, that states may sometimes be less generous, providing both lower rates and including more services to be covered by those rates; thus, a negative coefficient on an ancillary may be seen as spurious, resulting from an unmeasured tendency to be less generous. Thus, such a finding must be treated with extreme caution.
Comparison of ancillary-adjusted rates was done for only the latest year, 1998. Adjustment involved the subtracting of the coefficient for each ancillary from a state's 1998 rate if the state included that ancillary in the 1998 rate. This was done to attempt to compare rates after presumed state coverage of ancillary costs was subtracted out. The intent was neither to place a total value on a day of care (which in any case could not be done without considering other inclusions besides these seven ancillary services) nor to estimate actual costs of ancillary services. Therefore, the coefficients for excluded ancillaries were not added to the rate to produce a "total value" rate.
As noted, only the coefficients on ancillaries were used for this adjustment, even though additional variables were employed in two of the three estimation equations. Where the coefficient was negative, this resulted in an addition to the rate.
|
Results |
|---|
|
TOPAbstractBackgroundMethodsResultsDiscussionReferences |
|---|
Table 2 reports the three equations estimated to predict daily rates. Although negative coefficients on ancillaries were estimated in all equations, none was significant in the equation that included only the ancillary-inclusion measures. Two coefficients were significant, that on the inclusion of nonlegend drugs and that on the inclusion of DME, the inclusion of either estimated to increase 1998 rates by about $6.50. There were negative but insignificant coefficients on physical therapy and on medical supplies.
|
Table 3 provides rates adjusted by subtracting the coefficients on the ancillaries from the two models from the state average nursing facility per diem rates for 1998. The ancillary-only model yields adjusted rates that average about $7.00 less than actual rates, so about 7% of average rates across states (all figures unweighted for state size). By contrast, the model controlling for state characteristics and policies yields adjusted rates averaging almost exactly the same as actual rates; this suggests that the costs of ancillaries were not subtracted out of the rates on average, although differing state rates were adjusted for inclusion or exclusion of ancillaries. The states were ranked by their actual 1998 rates and by each of their ancillary-adjusted rates and are reported in Table 3 according to their rankings.
|
|
|
Discussion |
|---|
|
TOPAbstractBackgroundMethodsResultsDiscussionReferences |
|---|
We conclude that public and industry policy makers, as well as researchers, should consider the inclusion of ancillaries in rates when considering the relative adequacy of per diem rates across states. Ancillaries as a group, and some specific ancillaries, show significant influence on average state Medicaid nursing facility rates. More importantly, some states show considerable changes in their ranking vis à vis other states when their rates are adjusted for inclusion of ancillaries. Some or all ancillaries should be adjusted for when comparing rates across states.
Lower ranking should not in itself be taken as payment of an inferior rate. Some rankings are determined by only a few cents' difference in rates. However, after adjustment, 1998 rates still show from 5–1 to 4–1 ratios of highest to lowest adjusted average rates across states; and breaks between some ranks involve many dollars' difference. This extreme variation exists even after the list of ancillaries, and in some equations other factors, have been taken into account. This surely demonstrates great differences across states in the adequacy of their rates to cover good quality care, to provide adequate access to care, and to fairly reimburse providers for the care they give.
This study is only of rates and their predictors, not of such other issues as quality, access, or equity. However, average state rates, as well as rates applied to specific facilities, can be and are used in studies focusing on these topics (e.g., studies of the relative quality provided or of access to facilities). We suggest that rates used in studies of such topics be adjusted for such factors as the inclusion of ancillaries in rates, in order to better understand the effects of rates on these factors net the effects of differential inclusion in them of ancillaries.
There are limitations to this analysis. First, no adjustment was made for state fiscal year, so rates for some states are not adjusted for being up to 6 months older than rates for other states. Further, input factors that may be driving rates are not fully controlled. Thus, for example, New York will have higher rates than Kansas because land prices, labor, and so on are more expensive in New York. Average income per capita, percentage metropolitan, and nursing facility beds per aged should partially, but not fully, adjust for such factors. Better adjustment for costs of ancillaries would involve pricing of their costs and some determination of how those costs were, or were not, actually reflected in state rate setting.
|
Acknowledgments |
|---|
Received for publication January 26, 2001. Accepted for publication June 5, 2001.
|
References |
|---|
|
TOPAbstractBackgroundMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. Kitchener, J. H. Swan, and C. Harrington Medicaid Nursing Facility Utilization Research on Aging, July 1, 2006; 28(4): 493 - 514. [Abstract] [PDF]
|
|
|
|
 |
|
 J. M. Teno, V. Mor, D. DeSilva, G. Kabumoto, J. Roy, and T. Wetle Use of Feeding Tubes in Nursing Home Residents With Severe Cognitive Impairment JAMA, June 26, 2002; 287(24): 3211 - 3212. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
