Health Affairs, 22, no. 3 (2003): 214-223 doi: 10.1377/hlthaff.22.3.214 © 2003 by Project HOPE	New Online   New Issue: China & India   Obesity In China   Pay Cuts For Medicare Docs   Access To Care Woes

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (16) Citing Articles via Google Scholar Google Scholar Articles by White, C. Search for Related Content PubMed PubMed Citation Articles by White, C. Related Collections Long-Term Care Medicare

DataWatch

Rehabilitation Therapy In Skilled Nursing Facilities: Effects Of Medicare’s New Prospective Payment System

Abstract

 
In 1998 the Centers for Medicare and Medicaid Services (CMS) began phasing in a new prospective payment system (PPS) for Medicare payments to skilled nursing facilities (SNFs). I examine the effects of the new PPS on the level of rehabilitation therapy provided in SNFs. The percentage of residents of freestanding SNFs receiving extremely high levels of rehabilitation therapy dropped significantly, and the percentage receiving moderate levels increased. Freestanding SNFs, particularly for-profits, dramatically altered the services they provided in response to new financial incentives. This responsiveness underscores the importance of efforts now under way to refine the SNF PPS.

As hospital length-of-stay fell, the role of skilled nursing facilities (SNFs), home health agencies, and other postacute care providers expanded. A SNF is a nursing home that has been certified by the Centers for Medicare and Medicaid Services (CMS) to provide Medicare-reimbursable short-term skilled nursing services. Among SNF residents, the most common diagnoses are hip fracture, stroke, pneumonia, and heart failure.² The goal of SNF care is to restore recently hospitalized people to their prior level of functioning.

One implication of the shift toward postacute care is that a Medicare beneficiary requiring, for example, rehabilitation therapy following a hip fracture can choose from several treatment settings, including home, SNF, or outpatient clinic. Another implication is that SNFs and home health agencies now provide services formerly provided in the hospital. These providers differ fundamentally from hospitals in the scale and scope of services provided and in their ownership patterns (they are more likely than hospitals are to be for-profit).³

With the Balanced Budget Act (BBA) of 1997, Congress overhauled Medicare’s payment system for SNFs and home health agencies. The goal of this paper is to identify how SNFs responded to the BBA, specifically in terms of the level of rehabilitation therapy provided. An understanding of SNFs’ responses can help guide improvements in Medicare’s payment system for SNFs and postacute care providers more generally.

Provisions Of The Balanced Budget Act Of 1997

Top Provisions Of The Balanced... Study Methods Study Results Discussion NOTES

 
Until 1998 Medicare reimbursed SNFs on a retrospective, reasonable-cost basis. This encouraged a rapid increase in the use and cost of SNF care. Between 1991 and 1997 SNF care increased from 0.7 days per Medicare enrollee per year to 1.3 days, and Medicare’s average payment per SNF day increased (in 2001 dollars) from $160 to $289.⁴

In the 1990s the SNF industry expanded in two ways: First, freestanding nursing facilities obtained certification from Medicare to provide skilled nursing care (thereby gaining the designation "skilled nursing facility"), and second, hospitals established their own on-site SNFs. As reflected in the cost per SNF day, service intensity also increased. SNFs expanded their provision of rehabilitation therapy (physical, occupational, and speech), ventilator therapy, and "high-tech" nursing services such as tracheostomy care and administration of intravenous medications.

In the BBA, Congress responded to the growth in Medicare SNF spending by mandating that the CMS phase in a case-mix-adjusted PPS. Under the old cost-based payment system, SNFs were reimbursed for the actual costs of providing care (with some limits on routine operating costs). Under the new PPS, Medicare payments vary with the residents’ case-mix category (described below) but do not vary with the actual costs of providing care.⁵ Under the new PPS, therefore, SNFs have two potentially conflicting incentives: to move residents into higher-payment case-mix groups, and to limit the cost per day of providing care. In this paper I examine the extent to which SNFs responded to these new incentives.

The details of the SNF PPS are important for understanding the incentives it creates and its effects on rehabilitation therapy. The BBA mandated that the new SNF PPS use days as the unit of payment. The short-stay hospital PPS had used admissions; the home health PPS, home health episodes.⁶ The SNF PPS creates an incentive for SNFs to reduce the cost of providing a day of care, without necessarily creating an incentive to reduce lengths-of-stay.

The BBA required that payments to SNFs under the PPS be adjusted for case-mix. In a case-mix-adjusted payment system, people are assigned to case-mix groups based on selected individual characteristics (such as disease diagnoses or minutes of rehabilitation therapy received per week). Each case-mix group is assigned a payment rate based, roughly, on the average cost of providing care to people in that group.

For case-mix classification in SNFs, the CMS uses Resource Utilization Groups, Version III (RUG-III).⁷ RUG-III uses five rehabilitation categories and six nonrehabilitation categories.⁸ Among the five rehabilitation categories, classification is driven primarily by the minutes of rehabilitation therapy received per week (Exhibit 1). In 2000, 78 percent of Medicare-covered SNF stays were assigned to one of the five rehabilitation categories.⁹

View larger version (23K): [in this window] [in a new window]   EXHIBIT 2 Illustration Of Marginal And Average Payments To Skilled Nursing Facilities (SNFs) Per Minute Of Rehabilitation Therapy, Before And After The Prospective Payment System (PPS), 1998

The new case-mix-adjusted PPS differs substantially from the old system. Several points in particular are worthy of notice. First, the provision of rehabilitation therapy beyond a certain point (namely, 720 minutes per week) generates no additional payment, making the marginal payment zero beyond that point. Second, the average payment per minute for moderate levels of rehabilitation therapy is fairly generous. At 200 minutes per week, the average payment per minute under the new PPS actually exceeds the average payment under the old system. Third, marginal payment under PPS is quite "jagged," as a result of the discrete jumps in payment as residents move from one RUG-III rehabilitation category to the next.

Each SNF transitioned to the new PPS at the beginning of its first cost-reporting period on or after 1 July 1998. The BBA defines two types of SNFs, which I refer to as "new" and "old." Old SNFs are defined in the BBA as those that received their first payment from Medicare before 1 October 1995. The BBA stipulated that for the first three years after transitioning to the PPS, old SNFs would receive a blend of the case-mix-adjusted "federal" rates (analogous to diagnosis-related group, or DRG, prices for hospital stays) and a facility-specific rate based on each SNF’s historical costs. The blend in the first year was 75 percent facility-specific rate and 25 percent case-mix-adjusted; in the second year it was 50 percent facility-specific and 50 percent case-mix-adjusted; and so on. New SNFs, on the other hand, moved to a 100 percent case-mix-adjusted federal rate at the beginning of their first cost-reporting period on or after 1 July 1998. The Balanced Budget Refinement Act (BBRA) of 1999 altered these provisions by allowing old SNFs to skip the three-year transition and move to a 100 percent federal rate at the beginning of their first cost-reporting period on or after 1 January 2000. For old SNFs in the first year of the transition, marginal and average payments per minute of rehabilitation therapy were only 25 percent as large as payments to new SNFs were. For at least the first year of the SNF PPS, therefore, new SNFs had a much stronger incentive than old SNFs had to provide additional rehabilitation therapy to their residents.

The provisions of the SNF PPS suggest several research questions. First, did the number of days of SNF care change dramatically with the introduction of the PPS? A change in the level of SNF rehabilitation therapy could simply be attributable to reduction in the size of the industry. Second, did SNFs change the level of rehabilitation therapy provided in order to shift residents into more-profitable case-mix categories? Third, did different types of SNFs (for-profits versus nonprofits, hospital-based versus freestanding, new versus old) respond differently to the new PPS? I expect that for-profit SNFs altered their service provision more than nonprofits did and that, post-PPS, new SNFs provided higher levels of rehabilitation therapy than old SNFs did.

Study Methods

Top Provisions Of The Balanced... Study Methods Study Results Discussion NOTES

 
My main measure of the level of rehabilitation therapy is total SNF charges for physical, occupational, and speech therapy, discounted by a cost-to-charge ratio and inflated to July 2001 dollars using the hospital Producer Price Index.¹² I refer to this measure as "SNF rehabilitation charges." I created a hospital stay–level data set containing individual demographics, stay characteristics (such as length), and discounted SNF rehabilitation charges for all Medicare hospital stays and SNF stays occurring during 1997–2000. I excluded health maintenance organization (HMO) enrollees because hospital and SNF claims are generally not available for them. I combined the following Medicare administrative data sets: MEDPAR hospital and SNF stay claims, Medicare Denominator files, OSCAR data, Provider of Services extract data, and SNF cost reports.¹³

To generate Exhibit 3, I measured average SNF rehabilitation charges per hospital stay, where SNF rehabilitation charges are zero for hospital stays without a subsequent SNF stay. I then analyzed the three components that determine average SNF rehabilitation charges per hospital stay: the probability of being discharged to a SNF following a hospital stay, SNF length-of-stay conditional on a SNF stay occurring, and SNF rehabilitation charges per SNF day conditional on a SNF stay occurring.

View larger version (9K): [in this window] [in a new window]   EXHIBIT 3 Mean Medicare Skilled Nursing Facility (SNF) Rehabilitation Charges Per Hospital Stay, By Month And Year Of Hospital Admission, January 1997–October 2000

View larger version (15K): [in this window] [in a new window]   EXHIBIT 4 Percentage Of Skilled Nursing Facility (SNF) Residents Receiving Various Levels Of Rehabilitation Therapy Per Day, By SNF Type, 1997

View larger version (15K): [in this window] [in a new window]   EXHIBIT 5 Change In Percentage Of Skilled Nursing Facility (SNF) Residents Receiving Various Levels Of Rehabilitation Therapy Per Day, By SNF Type, 1997–2000

Study Results

Top Provisions Of The Balanced... Study Methods Study Results Discussion NOTES

 
The average SNF rehabilitation charge per hospital stay dropped by 44.6 percent after the new PPS was phased in, from $421 in 1997 to $233 in 2000 (Exhibit 3). The timing of this drop corresponds precisely with the phasing in of the new PPS. The drop from 1997 to 2000 implies that one or more of the following changed: the probability of being discharged to a SNF, SNF length-of-stay, or SNF rehabilitation charges per SNF day. I explore these possible explanations in turn.

Following a hospital stay, the probability of being discharged to a SNF dropped somewhat, from 16.3 percent in 1997 to 14.7 percent in 2000. Average SNF length-of-stay was relatively steady, changing from 23.8 in 1997 to 22.9 in 2000. The total number of SNF days dropped somewhat, from 42.0 million in 1997 to 36.9 million in 2000. These findings together imply that the large drop in average SNF rehabilitation charges per hospital stay is only partly explained by a drop in the probability of discharge to a SNF and is not explained by a change in SNF length-of-stay.

For-profit freestanding SNFs, which accounted for more than three-fifths of SNF days during 1997–2000, reduced average SNF rehabilitation charges per resident per day from $116 in 1997 to $62 in 2000, a 47 percent decline.¹⁵ Nonprofit freestanding SNFs, which accounted for one-fifth of SNF days, reduced average SNF rehabilitation charges by 23 percent, from $83 in 1997 to $64 in 2000. Hospital-based SNFs, on the other hand, slightly increased SNF rehabilitation charges per resident per day over this period.

In Exhibit 4, SNF residents in 1997 are categorized by the level of SNF rehabilitation charges per day and by SNF type. Exhibit 5 illustrates the change from 1997 to 2000 in the fraction of SNF residents in each category, again by SNF type. The most striking change occurred at for-profit freestanding SNFs. In 1997, 19.0 percent of residents at such facilities were receiving more than $200 per day in rehabilitation therapy; by 2000 this group had dropped to 1.6 percent. At nonprofit freestanding SNFs, the portion of residents receiving more than $200 per day also dropped, from 10.6 percent in 1997 to 1.8 percent in 2000. The portion of residents receiving little or no rehabilitation therapy (less than $10 per day) also dropped at all types of SNFs. These results show that following the implementation of the SNF PPS, SNF residents were more likely to receive moderate levels of rehabilitation and less likely to receive either no rehabilitation or an extremely high level of rehabilitation.

Among new freestanding SNFs, rehabilitation charges per SNF day declined from $112 in 1997 to $69 in 2000 (a 39 percent drop). At old freestanding SNFs, the decline in rehabilitation charges per SNF day was similar, from $108 in 1997 to $62 in 2000 (a 43 percent drop). Among new and old freestanding SNFs post-PPS, the distributions of residents across rehabilitation therapy categories were very similar (results not shown).

Discussion

Top Provisions Of The Balanced... Study Methods Study Results Discussion NOTES

 
In 1998 and 1999, as the new SNF PPS was being phased in, average SNF rehab therapy charges per hospital stay plummeted. Only a small part of the drop is attributable to a decline in the probability of being discharged to a SNF; the main explanation is a decline at freestanding SNFs in rehabilitation charges per SNF day. This drop reflects the incentives created by the new PPS. Under the old payment system, SNFs were reimbursed for the cost of providing therapy without regard to the minutes per week provided to each resident. Under the new PPS, SNFs receive fairly generous payments for providing moderate to high levels of rehabilitation therapy, but beyond a certain point (720 minutes a week), the provision of additional therapy generates no additional payments.

Providers and policymakers have raised concerns regarding the adequacy of payment rates and the accuracy of case-mix classification.¹⁶ The fact that there was only a moderate drop from 1997 to 2000 in total SNF days allays concerns somewhat regarding overall payment adequacy, but the adequacy of payments for hospital-based SNFs deserves further attention. My research suggests several new concerns: Are there SNF residents who could benefit greatly from receiving more than 720 minutes of rehabilitation therapy per week but who receive less because of the PPS? Are there SNF residents for whom rehabilitation therapy is inappropriate but who are receiving low or moderate levels of therapy because of the PPS? Freestanding SNFs’ responsiveness to payment incentives suggests that payment for rehabilitation therapy should be tied not just to the amount provided but also to some indication of clinical appropriateness.

Another obvious question raised by these findings is whether the extremely high levels of therapy provided pre-PPS provided any clinical benefit. The fact that the use of extremely high levels of rehabilitation therapy was so concentrated in for-profit freestanding SNFs suggests that this practice pattern was motivated more by profits than by clinical judgment. Regarding the sharp drop in the level of therapy at these SNFs, the charitable interpretation is that clinically superfluous rehabilitation was eliminated, but beneficial rehabilitation was maintained. Another possibility is that patients requiring intense rehabilitation therapy were shifted to other settings, such as rehabilitation hospitals.

The lack of a drop in rehabilitation therapy charges among hospital-based SNFs is interesting. Between 1997 and 2000 hospital-based SNFs faced a drop of $69 in average Medicare reimbursement per resident per day, while freestanding SNFs saw an increase of $20 per resident per day.¹⁷ We might, therefore, expect hospital-based SNFs to have reduced services more than freestanding SNFs did. On the other hand, hospital-based SNFs in 1997 had few residents receiving extremely high levels of rehabilitation therapy (4.2 percent of residents at nonprofits and 6.6 percent at for-profits) and also few residents receiving little or no therapy (15.6 percent at nonprofits and 15.5 percent at for-profits). Before the PPS was implemented, hospital-based SNFs may have already been providing something close to the "ideal" (from the viewpoint of generating reimbursement under the PPS) level of rehabilitation. Perhaps more importantly, the sheer size of a typical hospital, the diversity of its revenue sources, and the presence of physicians might shield a hospital-based SNF (particularly a nonprofit) from pressure to switch to more profitable levels of rehabilitation therapy.

Freestanding and hospital-based SNFs behaved differently before the PPS was implemented and responded to it differently as well. Within these two types of SNFs, though, for-profits and nonprofits behaved similarly. This does not necessarily imply that ownership form (for-profit versus nonprofit) is irrelevant; rather, the ownership form of the preponderant type of provider may be more important than the ownership form of an individual provider. Ownership form likely matters, perhaps a great deal, but at the industry or "peer group" level rather than the provider level. The fact that new and old SNFs behaved so similarly reinforces the importance of peer groups.

These findings suggest several valuable avenues for future research. The first question is the extent to which the drop in SNF rehabilitation therapy was made up for by increases in therapy from other sources such as outpatient or rehabilitation hospital–based services. The second question is whether the change in the distribution of SNF rehabilitation therapy (fewer residents receiving extremely high levels and fewer residents receiving no therapy) was associated with clinically meaningful changes in health outcomes.

The broader question is how to structure Medicare payments to reward clinically appropriate care in the appropriate setting. The assumption that providers will provide appropriate care regardless of payment incentives seems unfounded in the context of SNFs. The demonstrated responsiveness of freestanding SNFs to payment incentives underscores the importance and time urgency of efforts now under way to refine the SNF PPS.

Editor's Notes

 
Chapin White is a predoctoral fellow in the Economics of Aging Program at the National Bureau of Economic Research in Cambridge, Massachusetts.

The author acknowledges the National Institute on Aging (Grant no. T32 AG00186) and the Agency for Healthcare Research and Quality for providing predoctoral fellowship support during the preparation of this manuscript.

NOTES

Top Provisions Of The Balanced... Study Methods Study Results Discussion NOTES

 

1. National Center for Health Statistics, Health, United States, 2001 (Hyattsville, Md.: U.S. Public Health Service, 2001).
2. Author’s calculations from MEDPAR SNF claims matched to MEDPAR short-stay hospital claims for 1997 through 2000.
3. U.S. General Accounting Office, Medicare: Home Health Utilization Expands while Program Controls Deteriorate, Pub. no. GAO/HEHS-96-16 (Washington: GAO, March 1996).
4. Data on SNF use and costs are from Committee on Ways and Means, U.S. House of Representatives, Green Book:Background Material and Data on Programs within the Jurisdiction of the Committee on Ways and Means (Washington: House of Representatives, 6 October 2000). Data on the Consumer Price Index are from Bureau of Labor Statistics, "Consumer Price Index—All Urban Consumers," June 2002, www.bls.gov/data (5 August 2002).
5. U.S. Department of Health and Human Services, "Medicare Program: Prospective Payment System and Consolidated Billing for Skilled Nursing Facilities; Final Rule," FederalRegister (12 May 1998): 26098–26147.
6. Medicare Payment Advisory Commission, Report to the Congress: Medicare Payment Policy (Washington: MedPAC, March 2002).
7. B.E. Fries et al., "Refining a Case-Mix Measure for Nursing Homes: Resource Utilization Groups (RUG-III)," Medical Care 32, no. 7 (1994): 668–685.[Medline]
8. The nonrehabilitation categories are "Extensive Services," "Special Care," "Clinically Complex," "Impaired Cognition," "Behavior Only," and "Physical Function Reduced."
9. Author’s calculations based on the universe of MDS-2.0 Medicare five-day assessments from June 1998 through December 2000.
10. The cost per minute of rehab therapy is calculated by dividing total minutes of therapy (from MDS-2.0 five-day assessments) by total costs of therapy (equal to therapy charges discounted by a cost-to-charge ratio). Because MDS-2.0 assessments are only available for June 1998 through December 2000, the cost per minute as calculated reflects costs for the post-PPS period. For the purposes of the illustration, these costs are suitable (although pre-PPS costs per minute would be preferable).
11. Note that $4.19 =($238.87–$179.01/[(200–100)/7], and $2.15 =($238.87–$177.36)/(200/7).
12. Bureau of Labor Statistics, "Producer Price Index Revision—Current Series, General Medical and Surgical Hospitals," 2002, www.bls.gov/data (12 September 2002).
13. The methods used to create the data set are described fully in a Technical Appendix available at www.people.fas.harvard.edu/~cdwhite/Chapin_White.home.html.
14. To select the cutoffs, the minutes per day required to qualify for each rehabilitation group were multiplied by $1.80 per minute and rounded off. The charge per day categories correspond roughly to the RUG-III rehabilitation categories as follows: $0–$10 does not qualify for a rehabilitation category; $10–$80 qualifies for low or medium rehabilitation; $80–$200 qualifies for high or very high rehabilitation; $200+ qualifies for ultra high rehabilitation.
15. Based on the author’s calculations.
16. MedPAC, Report to the Congress: Medicare Payment Policy; and GAO, Skilled Nursing Facilities: Medicare Payments Need to Better Account for Nontherapy Ancillary Cost Variation, Pub. no. GAO/HEHS-99–185 (Washington: GAO, 30 September 1999).
17. Based on the author’s calculations.

                      What's this?

This article has been cited by other articles:

				A. Deutsch, C. V. Granger, A. W. Heinemann, R. C. Fiedler, G. DeJong, R. L. Kane, K. J. Ottenbacher, J. P. Naughton, and M. Trevisan Poststroke Rehabilitation: Outcomes and Reimbursement of Inpatient Rehabilitation Facilities and Subacute Rehabilitation Programs Stroke, June 1, 2006; 37(6): 1477 - 1482. [Abstract] [Full Text] [PDF]

				R. K. Bode, A. W. Heinemann, P. Semik, and T. Mallinson Relative Importance of Rehabilitation Therapy Characteristics on Functional Outcomes for Persons With Stroke Stroke, November 1, 2004; 35(11): 2537 - 2542. [Abstract] [Full Text] [PDF]