Trends in Inpatient Rehabilitation Stroke Outcomes Before and After Advent of the Prospective Payment System: A Systematic Review

Rinere O'Brien, Suzanne PT, NCS, MS

Journal of Neurologic Physical Therapy:
doi: 10.1097/NPT.0b013e3181cfd3ac

Background and Purpose: The purpose of this systematic review was to examine quality care indicators for inpatient stroke rehabilitation, trends for length of stay (LOS), functional outcomes, and discharge destination. In order to examine the influence of the prospective payment system (PPS), which was instituted in 2002, particular attention was paid to the pre-PPS to post-PPS period. This is the first review of literature to examine the quality of stroke care provided in inpatient rehabilitation facilities in the United States.

Methods: A search of Ovid Medline and Ovid Cumulative Index of Nursing and Allied Health databases was performed for articles published between 1990 and 2007. Search terms included treatment outcome, outcome assessment, activities of daily living, exercise, rehabilitation, cerebrovascular accident, LOS, and rehabilitation centers.

Results: Twelve articles met the criteria for review. A trend for shorter LOS was evident in the literature up until the time of implementation of PPS. An insufficient amount of literature was available to confirm whether this trend continued after the implantation of PPS. The most recent data indicated that average LOS in inpatient rehabilitation facilities for stroke was <20 days. Functional Independence Measure (FIM) discharge scores remained stable through the 1990s. After the implementation of PPS, discharge FIM scores may be decreasing, but revisions to the FIM tool may confound interpretation of post-PPS findings. Data for discharge to noninstitutional settings after stroke rehabilitation were inconclusive pre-PPS. There may be indications that discharges to institutional settings are increasing post-PPS.

Conclusions: The impact of PPS on quality care indicators for inpatient stroke rehabilitation, trends for LOS, and trends for functional outcomes are insufficiently documented in the medical literature. Further research is needed to understand the influence of LOS on functional outcomes and discharge destination. More information is needed on post-PPS outcomes to substantiate the benefit of inpatient rehabilitation for individuals with stroke.

Author Information

University of Rochester Medical Center, School of Nursing, Rochester, New York.

Address correspondence to: Suzanne Rinere O'Brien, E-mail:

Article Outline
Back to Top | Article Outline


Inpatient rehabilitation is often necessary after stroke for individuals to recover from and learn to compensate for the associated deficits resulting from stroke. Of the ∼780,000 people who have a first stroke or a recurrent stroke each year in the United States,1 ∼247, 000 require inpatient stroke rehabilitation, in either an acute inpatient rehabilitation facility (IRF) or skilled nursing facility. According to the Centers for Disease Control and Prevention, ∼140,000 people will receive care in an IRF each year.2 Inpatient rehabilitation is the setting of choice for those most at risk of chronic functional disability. Individuals who receive rehabilitation in an IRF experience a comprehensive, multidisciplinary approach to care.3 It is expensive to provide this level of care, but there is evidence that IRF stroke units provide better outcomes than noncomprehensive types of care, such as rehabilitation received in a skilled nursing facility.4–6

The advent of the prospective payment system (PPS), the system used by the Centers for Medicare and Medicaid7 to reimburse the healthcare providers, was a significant occurrence for IRF care. Passed in the Balanced Budget Act of 1997 and implemented in 2002, this change had two purposes: (1) to decrease the cost of rehabilitation services to Medicare, the largest payer for IRF rehabilitation8,9 and (2) to adjust reimbursement according to the presence of comorbidities and stroke severity, so that care for those with more comorbidities and more severe strokes would be reimbursed at higher rates.8–10 The reimbursement formula began with a fixed amount based on the diagnosis, and this rate was adjusted based on the qualifiers for age and previous chronic medical problems, if applicable.8–10

The standardized reimbursement rates in PPS created incentives to reduce the costs of care to maximize profit. During the period from 1990 to 2007, IRFs sought to cut costs, typically by decreasing length of stay (LOS).11 Little is known about the effect of reduced LOS on outcomes after inpatient rehabilitation. Because IRF care is needed by large numbers of individuals each year, it is important to examine the possibility of less favorable outcomes with shorter LOS. No reviews have been published on this subject.

The purposes of this systematic review were to examine the trend for LOS at IRFs for individuals with stroke, to determine whether there were trends for outcomes during this time period, and to compare the pre-PPS and post-PPS periods for effects on LOS and outcomes. At every level of healthcare, whether the insurance system, the delivery system, or the reimbursement system, the US healthcare system is unlike those in most other countries, and equitable comparisons cannot be made between the United States and most other countries.12 As a result, this review was limited to stroke rehabilitation provided by IRFs in the United States.

Back to Top | Article Outline


Search and Screening Strategy

To complete this review, the author searched three electronic databases: Ovid Medline, Ovid Cumulative Index of Nursing and Allied Health, and Physiotherapy Evidence Database. Search terms included treatment outcome, outcome assessment, activities of daily living, exercise, rehabilitation, cerebrovascular accident, LOS, and rehabilitation centers. The search strategy included the use of scope notes to ensure an inclusive list of terminology and key words. Only English-language articles published during the period of January1990 to July 2008 were included. The author examined a total of 242 abstracts for inclusion. An article was excluded if the focus of the article was limited to (1) treatment and not outcomes, (2) IRFs outside the United States, (3) a subtype of stroke, for example, posterior cerebral artery, or (4) Veterans Health Administration hospitals. Flowcharts of the search strategy and results for the Medline and Cumulative Index of Nursing and Allied Health searches are illustrated in Figure 1. A flowchart for the Physiotherapy Evidence Database search was not included because no new citations were recovered from that database.

Articles included in this review met the following criteria: (1) reported data of patients with stroke from IRFs in the United States, (2) measured outcomes using the Functional Independence Measure (FIM) or discharge destination, and (3) reported LOS data.21 Twelve articles describing 10 studies (one study was described in three articles) met the inclusion criteria. All 12 articles were original reports. Articles included in the review are listed in Table 1.

Back to Top | Article Outline
Data Collection

The following characteristics were abstracted from each article: sample size of patients and size of IRFs, study design, type and location of the facility, type of data, years data were collected, outcomes measured, type of analysis conducted by the authors, and which of the three objectives of this review was met by the article.

Back to Top | Article Outline
Data Sources and Measures

The Uniform Data System for Medical Rehabilitation (UDSmr), which receives reports from nearly 70% of the IRFs in the United States,20 is a major source of data on IRF care used in several of the analyses discussed in this review. The UDSmr database contains information related to the following variables: patient characteristics, LOS, total charges, and functional status. These data were designed to describe the realm of rehabilitation for clinical and research purposes.21

In this review, LOS was defined as the number of days in the IRF. LOS is considered an important indicator of the process of care, and for the potential amount of improvement a patient can achieve during the rehabilitation stay.22 The number of days of therapy actually received was not differentiated from total LOS in the studies in this review. The outcome of care was defined as the indicator of quality from the cumulative effects of interventions received by the patients.22 Outcomes after care in an IRF can be measured in two ways, each having the potential to reflect the quality of care. The first, mean discharge FIM scores or mean FIM change scores, denotes the functional progress made by patients from all the interventions received in the IRF. The second, discharge destination, identifies the setting where the patient lives after the IRF intervention. Discharge to the community or home, instead to an institution, is often viewed as an important quality benchmark for IRFs.

During the period of interest of this review, the FIM was the measurement tool used nearly universally by IRFs to assess functional status. The FIM is a scale that is scored by clinicians at admission, at regular intervals (typically weekly) during the rehabilitation stay, and at discharge. This tool was designed to measure basic global functional status, the burden of care, quality of care, and the effectiveness of care.21,23 The original version of the FIM included scores for 18 different motor and cognitive tasks, with each task having a lowest possible score of 1 (dependent) and a maximal score of 7 (independent without need for an assistive device). Total FIM scores range from 18 to 126 points.21,23

The FIM has been in existence for ∼20 years, and a large body of literature is available regarding the psychometric properties of this tool in patients, including those with stroke, who are undergoing physical rehabilitation. The FIM has two dimensions, motor and cognitive, with all 18 items functioning similarly across diagnoses, and across LOS.24–28 Test-retest reliability of the FIM has been shown to be high during both short durations (seven to 10 days; intraclass correlation coefficient [ICC] = 0.99) and long durations (four to six weeks; ICC = 0.92).29 Reliability of the FIM has been found to be greater when used by examiners trained in its use (ICCs from 0.97 to 0.99) in comparison with examiners at IRFs who did not train personnel in its use (ICCs from 0.89 to 0.96).30 Concurrent validity of the FIM has been supported during the course of the rehabilitation stay through correlations with Barthel Index.31 FIM scores have been found to predict the burden of care after stroke.25 Using hypothesis testing, construct validity of the FIM was affirmed by the finding that higher scores were achieved by younger patients, patients with fewer comorbidities, and patients who were discharged home.32

With the advent of PPS, the FIM underwent several revisions. One significant revision was to the FIM scoring system, to which a low score of 0 was added. Using the UDSmr database, one study examined the psychometric properties of the revised FIM.23 Item functioning was found to be very similar to the original FIM, but the revised FIM may provide lower scores compared with the original version. None of the articles included in the present review took into consideration how FIM revisions may have altered reported scores.

Back to Top | Article Outline


Trends for LOS

Studies reporting data from the 1990s showed a trend toward shorter LOS for IRF stroke rehabilitation.5,11,13–15,17–19 Data arranged by year of data collection, rather than by year of publication, are illustrated in Table 2 to present the data in chronologic order. This arrangement of studies suggested a descending trend for LOS from the early to late 1990s. This apparent trend was clearly demonstrated in a retrospective, longitudinal study of a nationally representative database of 744 IRFs using the UDSmr database from 1994 to 2001, which confirmed the downward trend during the 1990s and showed that it continued through 2001. During the course of this study, average LOS decreased significantly from 24 days to 16 days (P < 0.001).11 The study included patients from five diagnosis groups, but only data related to stroke have been reported in this review. A separate longitudinal study of a smaller sample of six IRFs in the United States used data from 2001 to 200317 and reported average LOS to be 18.6 days.

Back to Top | Article Outline
Trends in Outcomes: FIM Scores and Discharge Destination

Despite decreasing LOS, discharge FIM scores remained steady throughout the 1990s. As illustrated in Table 3, patients with stroke were consistently achieving discharge FIM scores between 83 and 87 out of 126 points.11,15,17,18 Bode and Heinemann13 stratified patients with stroke by the number of weeks (from two through six weeks) spent in rehabilitation and monitored FIM outcome by each stratum. They compared patients with stroke (n = 129), brain injury, and spinal cord injury, although only the data for patients with stroke were included in this review. Patients with more severe strokes remained in IRF care for a greater number of weeks and achieved lower FIM scores at discharge, but change in FIM scores from admission to discharge did not differ by stroke severity. In other words, patients with more severe strokes did not reach the same FIM scores as those with less severe strokes, but made the same amount of change as those with less severe strokes, although over a longer period of time.

In this literature, discharge destination was typically reported as rates of community discharge. There were no clear trends for this quality marker across studies in the review, possibly because of fewer studies reporting this variable. As can be seen in Table 4, rates of community discharges reported after IRF stroke rehabilitation ranged from 74% to 81%.11,12,17,18

Back to Top | Article Outline
Comparison of LOS and Outcomes Before and After PPS

Two studies specifically compared outcomes in the pre-PPS time period with those in the post-PPS time period.7,8 These studies can be instructive because longitudinal methodology provides a reliable approach to study outcomes across a particular time period. Each study was derived from a relatively small database compared with the studies based on the larger UDSmr database, and included three IRFs and one IRF, respectively. DeJong et al8 noted that although LOS did not change, a significant decrease of 2.9 points (P = 0.034) in average FIM change scores occurred during the period beginning one year before PPS and ending one year after PPS. According to the authors, decreased scores for the moderately impaired and severely impaired groups accounted for the overall decrease, despite an average increase in discharge FIM scores in the mildly impaired group during the pre-PPS period to post-PPS period. Over the course of the three-year study, the authors reported trends toward decreasing discharge to home across the mild, moderate, and severe stroke groups of 7.1%, 3.1% and 2.0%, respectively, but none of these trends were significant.

Gillen et al9 compared the impact of PPS over a longer period, from five years before to 3.5 years after PPS implementation. In this larger data set, these authors reported a 4.5-day decrease in LOS (P < 0.001), a decrease in mean discharge FIM score from 92.75 to 87.7 points (P < 0.001), and an accompanying 10% decrease in those discharged home after PPS (P < 0.001).

Back to Top | Article Outline


The purpose of this review was to identify the trends over time for stroke outcomes in IRFs in the United States, with special attention to the critical period when IRF reimbursement was altered in 2002. Three factors reflecting the process and outcomes of IRF care, such as LOS, FIM scores, and discharge destination, were examined for trends present from 1990 through 2008. The findings of this review provide mostly consistent data on the continuing trend toward shorter LOS for IRFs, which began well before the implementation of the PPS. Well before the implementation of the PPS, IRFs were decreasing patients' LOS. However, studies specifically examined the pre-PPS period to post-PPS period reported conflicting trends in LOS data post-PPS. Further research is necessary to determine whether LOS has continued to decrease since the advent of PPS. If such a trend has continued, then there is potential for a negative impact on outcomes, FIM scores, and discharge destination after stroke rehabilitation at IRFs.

FIM scores at discharge reflect functional outcomes after IRF care and can help clinicians ascertain the postdischarge needs of patients. Discharge FIM scores were not shown to decrease until after PPS was implemented, after which two studies7,8 identified significant decreases in discharge FIM scores. This finding is confounded by the fact that the FIM tool was revised with the advent of PPS. Neither study reported took into account revisions to the tool during analysis of FIM score data. Granger et al23 recently reported that the new post-PPS version of the FIM functions somewhat differently from the original version. Future investigations should control for the different FIM versions during analysis to ensure the accurate evaluation of outcomes.

The trends in FIM scores reported in this review give reason for a positive outlook. Interventions that patients with stroke receive from therapists in IRFs may have enabled FIM scores to remain stable despite shorter LOS. Across the studies included in this review, patients with differing levels of stroke severity, ages, and LOS who received care at IRFs all made significant functional gains with rehabilitation. Further studies to determine which interventions promote the greatest functional gain, and in which subgroups of individuals with stroke, should continue. Many possibilities for subgroup analyses exist, including severity of stroke, type of lesion or stroke syndrome, or specific demographic factors, such as sex, which may alter responses to rehabilitation interventions.

Another outcome, discharge destination, did not demonstrate consistent changes in IRF performance over time. Trends for discharge to home remained stable before PPS, but there may be worsening outcomes for post-PPS discharge destination. Although post-PPS discharge destination did not reach significance in the analysis by DeJong et al,8 detrimental trends were observed, most surprisingly in the mild stroke group, which was found to have the largest increase in discharges to institutional settings, with the moderate stroke and severe stroke groups also increasingly being discharged to institutional settings. Within a longer period of observation, Gillen et al9 found significantly more patients discharged to institutional settings. More study is needed to analyze these potential trends.

Back to Top | Article Outline

There are several limitations to this review. First, there is a paucity of research, and no randomized controlled trials, on the relationship between payment system and rehabilitation outcomes. Further studies are needed to elucidate the functioning of patients with stroke after IRF care and also to elucidate the functioning of IRFs that treat such patients. Second, FIM revisions have complicated comparisons of studies from the pre-PPS period with the post-PPS period. Any future study of this literature should describe which version of the FIM was used to measure function. Third, studies included a wide variety of sample sizes at the IRF level and at the patient level, resulting in potential problems with generalizability. Larger studies would improve generalizability, but authors should carefully describe the characteristics of the IRFs and the patients. The available literature was satisfactory for describing patient characteristics, but more detailed information regarding IRF factors is needed. Factors such as IRF size, location, and ownership could provide novel information about how these characteristics influence LOS and outcomes. Finally, factors that affect the decision-making process regarding discharge destination were not discussed in this literature. Decisions made at the time of discharge are not typically made simply based on the FIM score. Other factors must be considered by clinicians to ensure safety after discharge. For instance, in some cases, the home must be wheelchair accessible, and in other cases, supervision or assistance will be required for the patient. No study reviewed included controls for these factors.

Schlenker et al5 cautioned that PPS could provide incentives to shorten LOS to maximize IRF reimbursement. Evidence may be emerging that a threshold has been crossed and that rehabilitation stays are too short to allow patients to return to the home after care in the IRF setting. Further research is needed to improve the understanding of these phenomena and to protect the quality outcomes for which IRF stroke rehabilitation is known.

Back to Top | Article Outline


IRF care has undergone at least two major changes since 1990: decreasing LOS and altered reimbursement. The effects of LOS changes are unclear, but some research suggests cause for concern about the outcomes of care being provided under these conditions. If further research determines that outcomes are indeed deteriorating, action must be taken to prevent worsening function, increased disability, increased institutional discharges, and higher costs for the US healthcare system.

Back to Top | Article Outline


1. American Heart Association. Heart Disease and Stroke Statistics—2008 Update. Dallas, TX: American Heart Association; 2008.
2. Centers for Disease Control and Prevention. Public health and aging: hospitalizations for stroke among adults >65 years—United States, 2000. MMWR Morb Mortal Wkly Rep. 2003;52:586–589.
3. Centers for Medicare and Medicaid. IRF classification criteria. Accessed December 1, 2007.
4. Deutsch A, Granger CV, Heinemann AW, et al. Poststroke rehabilitation: outcomes and reimbursement of inpatient rehabilitation facilities and subacute rehabilitation programs. Stroke. 2006;37:1477–1482.
5. Schlenker RE, Kramer AM, Hrincevich CA, et al. Rehabilitation costs: implications for prospective payment. Health Services Res. 1997;32:651–668.
6. Langhorne P, Duncan P. Does the organization of postacute stroke care really matter? Stroke. 2001;32:268–274.
7. Centers for Medicare and Medicaid Services. Prospective payment system. Accessed December 1, 2007.
8. DeJong G, Horn SD, Smout RJ, et al. The early impact of the inpatient rehabilitation facility prospective payment system on stroke rehabilitation case mix, practice patterns, and outcomes. Arch Phys Med Rehabil. 2005;86(Suppl 2):93–100.
9. Gillen R, Tennen H, McKee T. The impact of the inpatient rehabilitation facility prospective payment system on stroke program outcomes. Am J Phys Med Rehabil. 2007;86:356–363.
10. McCue MJ, Thompson JM. Early effects of the prospective payment system on inpatient rehabilitation hospital performance. Arch Phys Med Rehabil. 2006;87:198–202.
11. Ottenbacher KJ, Smith PM, Illig SB, et al. Trends in length of stay, living setting, functional outcome, and mortality following medical rehabilitation. JAMA. 2004;292:1687–1695.
12. Shi L, Singh DA. Delivering Healthcare in America: A Systems Approach. 4th ed. Boston, MA: Jones and Bartlett; 2008.
13. Bode RK, Heinemann AW. Course of functional improvement after stroke, spinal cord injury, and traumatic brain injury. Arch Phys Med Rehabil. 2002;83:100–106.
14. Bode RK, Heinemann AW, Semik P, et al. Relative importance of rehabilitation therapy characteristics on functional outcomes for persons with stroke. Stroke. 2004;35:2537–2542.
15. Hamilton BB, Granger CV. Disability outcomes following inpatient rehabilitation for stroke. Phys Ther. 1994;74:494–503.
16. Horn SD, DeJong G, Smout RJ, et al. Stroke rehabilitation patients, practice and outcomes: is earlier and more aggressive therapy better? Arch Phys Med Rehabil. 2005;86(Suppl 2):101–114.
17. McNaughton H, DeJong G, Smout RJ, et al. A comparison of stroke rehabilitation practice and outcomes between New Zealand and United States facilities. Arch Phys Med Rehabil. 2005;86(Suppl 2):115–120.
18. Murakami M, Inouye M. Stroke rehabilitation outcome study: a comparison of Japan with the United States. Am J Phys Med Rehabil. 2002;81:279–282.
19. Ruff RM, Yarnell S, Marinos JM. Are stroke patients discharged sooner if in-patient rehabilitation services are provided seven v six days per week? Am J Phys Med Rehabil. 1999;78:143–146.
20. Uniform Data System for Medical Rehabilitation Web site. Accessed October 18, 2008.
21. Hamilton BB, Granger CV, Sherwin FS, et al. A uniform national data system for medical rehabilitation. In: Fuhrer MJ, ed. Rehabilitation Outcomes: Analysis and Measurement.. Baltimore, MD: Paul H. Brookes Publishing; 1987:137–147.
22. Donabedian A. The quality of medical care. Science. 1978;200:856–864.
23. Granger CV, Deutsch A, Linn RT. Modifications of the FIM instrument under the inpatient rehabilitation facility prospective payment system. Am J Phys Med Rehabil. 2007;86:883–892.
24. Chang WC, Chan C. Rasch analysis for outcomes measures: some methodological considerations. Arch Phys Med Rehabil. 1995;76:934–939.
25. Granger CV, Cotter AC, Hamilton BB, et al. Functional assessment scales: a study of persons after stroke. Arch Phys Med Rehabil. 1993;74:133–138.
26. Heinemann AW, Linacre JM, Wright BD, et al. Relationships between impairment and physical disability as measured by the functional independence measure. Arch Phys Med Rehabil. 1993;74:566–573.
27. Linacre JM, Heinemann AW, Wright BD, et al. The structure and stability of the functional independence measure. Arch Phys Med Rehabil. 1994;75:127–132.
28. Stineman MG, Shea JA, Jette A, et al. The functional independence measure: tests of scaling assumptions, structure, and reliability across 20 diverse impairment categories. Arch Phys Med Rehabil. 1996;77:1101–1108.
29. Ottenbacher KJ, Mann WC, Granger CV, et al. Inter-rater agreement and stability of functional assessment in the community-based elderly. Arch Phys Med Rehabil. 1994;75:1297–1301.
30. Hamilton BB, Laughlin JA, Fiedler RC, et al. Interrater reliability of the 7-level functional independence measure (FIM). Scand J Rehabil Med. 1994;26:115–119.
31. Roth E, Davidoff G, Haughton J, et al. Functional assessment in spinal cord injury: a comparison of the modified Barthel index and the ‘adapted’ functional independence measure. Clin Rehabil. 1990;4:277–285.
32. Dodds TA, Martin DP, Stolov WC, et al. A validation of the functional independence measure and its performance among rehabilitation patients. Arch Phys Med Rehabil. 1993;74:531–536.

stroke; rehabilitation; outcomes; Prospective Payment System; length of stay; Functional Independence Measure; discharge destination

© 2010 Neurology Section, APTA