Inpatient Rehabilitation Outcome Trends: Title and subTitle BreakImplications for the Future

The goal of inpatient rehabilitation is to provide efficient care to patients with potentially disabling conditions to optimize physical and cognitive function and to discharge to the least restrictive setting, ideally to home. To benefit from inpatient rehabilitation, patients must be medically stable so that ongoing medical problems do not interfere with participation in therapy. To qualify for inpatient rehabilitation, patients must be able to participate in at least 3 hours of rehabilitation therapies a day and demonstrate improving function.^{1 - 2} The US Centers for Medicare & Medicaid Services (CMS) recently estimated that 660 000 patients annually receive inpatient rehabilitation, with 70% funded by Medicare. For 2003, Medicare payments for inpatient rehabilitation were $5.9 billion.¹ Despite these large numbers, rigorous information regarding outcome trends and the impact of changes in CMS regulations for inpatient rehabilitation is limited.

In this issue of JAMA, Ottenbacher and colleagues³ document trends in inpatient medical rehabilitation length of stay (LOS), functional status, discharge setting, and mortality over the 8-year period (1994-2001) before the implementation of the inpatient rehabilitation prospective payment system (PPS) in 2002. The data were collected through the Uniform Data System for Medical Rehabilitation (UDSMR), a large, reliable database that included 148 807 patients from 744 inpatient rehabilitation facilities (IRFs) in the United States. The authors demonstrate that several significant changes occurred during the study period. Between 1994 and 2001, IRFs increased efficiency as measured by patient functional gain per day and decreased the median LOS by 8 days while maintaining a relatively stable gain in functional improvement as measured by the Functional Independence Measure (FIM). While there were statistically significant differences in FIM gain over time, a difference of 1 or 2 FIM points is not clinically important. The rate of discharge to home remained stable over time, but mortality after discharge increased after adjusting for age, comorbidities, and admission FIM scores.

Interpretation of these findings requires understanding of several developments that occurred during and since the study period. During this period, many rehabilitation programs expanded interdisciplinary rehabilitation services from 5 to 6 or 7 days per week, which may have contributed to more rapid functional gains.⁴ Managed care oversight increased in the 1990s with heightened scrutiny of rehabilitation admissions, treatment goals, and patient progress.^{5 - 6} This routine monitoring resulted in pressure to achieve functional goals faster with a shorter LOS. The use of alternative treatment options also increased, such as nursing home–based subacute rehabilitation and consolidation of IRFs by large for-profit corporations.⁷ Clinical pathways were designed to increase efficiency and decrease LOS but also changed patterns of rehabilitation admissions.^{8 - 9} In one study of patients who had undergone total knee arthroplasty, 39% of patients were discharged to a postacute facility (IRF or nursing home) in 1992 and this number increased to 99% in 1995 after implementation of a clinical pathway.¹⁰ All of these factors may have contributed to improved efficiency of inpatient rehabilitation during 1994-2001.

Another important factor is that a PPS for acute care hospitalization was enacted in 1983, but due to the complexity of defining “usual” LOS in rehabilitation patients, IRFs, along with psychiatric and long-term care facilities, were exempt from this diagnosis related group acute care PPS. In the 1990s the number of facilities exempt from the acute care PPS steadily increased. Payments to these facilities increased 17% per year in the 1990s, and payments to rehabilitation hospitals reportedly exceeded costs by 5% in 1996.² Due in part to concerns about these trends, Congress enacted the Balanced Budget Act of 1997¹¹ that authorized the implementation of a PPS for inpatient rehabilitation, which started in 2002. The PPS classifies patients into a case-mix group based on diagnosis, functional status measured by the FIM, age, and comorbidities. The case-mix group is then used to determine a payment level.^{12 - 13}

Future analysis will determine the impact of the PPS, but preliminary research reveals a potential negative financial impact. In a retrospective analysis of rehabilitation patients with traumatic brain injury treated prior to implementation of the PPS, median cost exceeded projected PPS payment by 16%.¹⁴ The PPS system currently includes only Medicare patients, who represent about 70% of all rehabilitation inpatients.¹⁵ Those IRFs with a higher proportion of patients covered by Medicare will have greater potential risk with PPS. Facilities with a higher proportion of revenue from Medicare are more likely to be nonprofit, offer fewer services, operate with lower staff levels, and have lower operating profits. The change in Medicare payment may have a significant impact on the financial performance of, and likely the clinical care at, these facilities.¹⁶

Another impact on IRFs is enforcement of the 75% rule. In 1984 the US Department of Health and Human Services defined an IRF as a facility providing intensive rehabilitation services in which at least 75% of the patients received services for one of the following conditions: stroke, spinal cord injury, congenital deformity, amputation, major multiple trauma, fracture of femur (hip fracture), brain injury, polyarthritis, neurologic disorders, and burns.¹ While these categories cover a majority of rehabilitation patients, a large number of patients benefit from inpatient rehabilitation after joint replacement surgery or severe medical or surgical conditions (cardiac, pulmonary, cancer).

Inconsistent methods were used to determine if a facility was in compliance and many rehabilitation centers interpreted the regulations such that patients with joint replacements were included in the polyarthritis classification. In UDSMR data from 1999, 23% of IRF discharges were for lower extremity joint replacement.¹⁷ CMS recently estimated that only 13% of IRFs were in compliance with the 75% rule and changed the list of approved conditions.¹ Polyarthritis was replaced with 3 new conditions: active, polyarticular rheumatoid arthritis, psoriatic arthritis, and seronegative arthropathies; systemic vasculidities with joint inflammation; and severe or advanced osteoarthritis involving 2 or more joints. CMS noted that an osteoarthritic joint replaced by a prosthesis is no longer considered to have osteoarthritis and cannot be counted as 1 of the 2 involved joints. CMS also included an additional covered condition to include patients with knee and/or hip replacement and who have bilateral joint replacement, are extremely obese, or are aged 85 years or older. To allow IRFs time to adapt to the new regulations and enforcement, CMS lowered the compliance threshold to 50% with yearly increases until January 2007, when it will again reach 75%.¹

The impact of enforcement of the 75% rule is not known. CMS estimates that IRFs will have less than 1% reduction in Medicare payments in the first year of implementation of the revised 75% rule, but these payment reductions will steadily increase as the compliance threshold increases.¹ Facilities that care for a large number of patients with single joint replacement without multijoint osteoarthritis may not be able to continue to provide this care and be in compliance with the 75% rule. In the report by Ottenbacher et al, orthopedic conditions were the largest group in the database, with 30% of the patients.³ It is not known how many of these patients would be included in the new 75% rule. These regulatory changes may be obscure to the primary care clinician, but they will likely affect many patients who undergo some form of rehabilitation. If rehabilitation centers that are primarily caring for hip or knee replacement patients are forced to close due to the changes in regulations, patients with other conditions previously served in those centers may have reduced access to inpatient rehabilitation treatment.

Perhaps the most provocative finding, as well as the outcome most difficult to explain, in the study by Ottenbacher et al is that mortality in rehabilitation patients increased uniformly from 1994-2001. Overall, mortality increased from less than 1% in 1994 to 4.7% in 2001, and it was not related to more comorbidities, older age, or lower functional status at time of admission.³ To understand this trend, it is useful to examine expected mortality. Reports of survival 1 month after stroke vary from 83% in data from 1985-1989¹⁸ to 95% in data collected from 1990-1997.¹⁹ There is little information regarding survival in patients treated in IRFs. Lai et al²⁰ followed up patients with stroke admitted in 1987-1989, a time when IRF admission criteria were similar to today but LOS was much longer. Of those transferred to an IRF, the survival was 97.7% at 3 months and 94.6% at 6 months, similar to those discharged to home. The survival rate for patients discharged to nursing homes was 84.4% at 3 months.²⁰ Ottenbacher and colleagues completed follow-up at an average of 3 months after discharge from inpatient rehabilitation or about 4 months after stroke event, for example, and they reported an average stroke mortality of 3.1%, increasing from about 0.5% in 1994 to almost 5% in 2001.³ While it is difficult to make direct comparisons, this mortality rate is similar to what might be expected 4 months after a stroke. Instead of questioning why stroke mortality increased to almost 5% in 2001, perhaps the question is why was it so low in 1994.

That mortality increased in all the impairment groups would indicate that some factor influencing mortality is common to different diagnostic groups. Potential factors related to increased mortality such as more comorbidities, older age, or lower admission FIM score were not significantly different. The increase in mortality is consistent with a previous study by Ottenbacher et al,²¹ which reported increased rates of hospital readmissions after inpatient rehabilitation. The medical stability of the patients admitted over this time may have changed due to the pressure to decrease the acute care LOS and to move patients to inpatient rehabilitation earlier in their recovery. While patients must be medically stable to participate in an inpatient rehabilitation program, there is no set definition of medical stability and many IRFs are capable of caring for medically complex patients. Patients admitted with significant medical problems may still participate in 3 hours of therapy a day at a low level as the medical problems improve. However, patients with greater medical problems would be expected to have lower admission FIM scores and longer LOS, and that was not seen in this study. In summary, the reason for increased mortality over time remains obscure but may reflect changes in admission criteria and medical stability of patients admitted to IRFs that were not measured in the study by Ottenbacher et al. The relationship between earlier acute care discharge, rehabilitation admission criteria, and LOS requires further study to understand changes in mortality.

The study by Ottenbacher et al³ in this issue of JAMA documents trends in inpatient rehabilitation from 1994-2001, a period of significant change in the health care industry with the increasing role of managed care, but with stable Medicare payment for inpatient rehabilitation. The data presented precede the potentially significant changes caused by the implementation of the inpatient rehabilitation PPS and enforcement of the 75% rule. Rehabilitation facilities have previously altered admission case mix and LOS to optimize reimbursement,^{22 - 23} and these new regulations will likely result in ongoing changes in admission criteria, therapy services provided, LOS, and discharge location. Facilities will potentially change several aspects of the rehabilitation program to maintain financial viability in response to the PPS and enforcement of the 75% rule, perhaps by avoiding admittance of patients with complex medical conditions or patients with an unclear discharge plan who will require an extended LOS, exceeding the PPS-projected LOS.

The impact of the 75% rule will reduce the number of patients with joint replacements admitted to inpatient rehabilitation, which may lead to increased discharges to nursing homes or discharges to home at a lower functional level, requiring more home care. Length of stay likely will continue to decline as IRFs determine the discharge date based on the PPS LOS instead of the patient’s rehabilitation goals. Earlier discharge will place increased burden on postacute care such as home health, nursing homes, and outpatient therapy programs. Earlier discharge and increased care needs at discharge also dramatically increase the family burden for patients discharged to home. In the past, IRFs have responded to pressure to decrease LOS by increasing efficiency, but further decreases in LOS may result in discharge of patients at lower functional levels without further gains in efficiency. Future research to document trends in inpatient rehabilitation after the implementation of PPS and the 75% rule is essential, and the UDSMR will be an important tool to document the impact of these changes.