The three types of orthopedic procedures most commonly requiring postacute rehabilitation services include total knee replacement (TKR), total hip replacement (THR), and hip fracture. The overall increase in procedures as well as the rapid discharge from acute care hospitals has resulted in an increased demand for postacute care services, particularly in the Medicare population, to inpatient rehabilitation facilities (IRF) and skilled nursing facilities (SNF). According to the 2005 Medicare Payment Advisory Commission's (MedPAC) assessment, one-third of beneficiaries discharged from acute care hospitals use postacute care (PAC) services. Following hip and knee replacement, 35% of patients use home care, 35% are admitted to IRFs, and 30% are admitted to SNFs.1 Although many patients recovering from joint replacement can go directly home, medical complexity and(or functional limitations often preclude this.2,3 Rehabilitation services are provided to a varying degree in each of these settings, with the goals of increasing function, preserving independence, reducing complications of disability, and enabling a transition back to residence at home or the least restrictive setting.
There is considerable variance in the distribution of these facilities across the United States, and there is concomitant variation in practice patterns for referral of patients. The recommendation of the care providers in the acute hospital setting is based on many poorly studied factors, such as the opinion of the referring physicians, the preferences of the patient and family, insurance benefits, and the availability of options in their community. A few studies have identified some of these factors. De Pablo et al4 reported that the most important predictor of discharge to an IRF following elective total hip replacement (THR) was poor functional status, especially walking before discharge. Advanced age, female sex, living alone, lower income, and obesity have also been identified as “risk factors” for discharge to IRFs. Insufficient social support by family and friends has been cited as promoting discharges to IRFs and SNFs compared to discharges home.5 Availability of PAC options and the effects of distance to providers are additional factors when choosing between IRF and SNF care. Buntin et al6 analyzed data on Medicare joint replacement patients discharged from hospitals and reported that availability of PAC is an important determinant of type of facility used; the further away the nearest IRF is and the closer the nearest SNF is, the less likely a patient will go to an IRF.
There is significant overlap in the services provided in SNF and IRF settings, most likely influenced by difference in reimbursement by Medicare. IRFs have a prospective payment system (PPS), and are paid on a per discharge basis, with the payment rate influenced by the patient's diagnosis, age, comorbid complications and functional burden of care assessed during an initial 3 day assessment period. SNFs are also on a Medicare prospective payment system, but are reimbursed on a per diem rather than discharge basis with the amount influenced by the Resource Utilization Groups (RUGs). The SNF prospective payment system has a strong financial incentive for providers to furnish rehabilitation services in the lower minute classification category of the RUGs system but for a longer length of stay whereas IRFs have a strong incentive to reduce the length of stay. This incentive is somewhat mitigated by a reduced payment to IRFs for short stay outliers.7
When IRF and SNF settings are compared, there are also differences in staffing and program design, as well as in the intensity of therapy services that are provided. In general, SNFs provide less physician contact, fewer hours of skilled therapy, and lower nurse-patient ratios than IRFs.8 In a recent study by DeJong et al,9 IRFs provided an average of 1 hour more of therapy per day (physical therapy and occupational therapy combined) compared to freestanding SNFs, and patients received more OT overall at IRFs than at SNFs. These researchers also found that therapy intensity, a practice feature strongly associated with IRFs, was positively correlated with functional outcome.10 Although therapy intensity in SNFs is not regulated, Jette et al11 demonstrated that, for orthopedic conditions, increased level of intensity was correlated with a reduced length of stay.
Studies comparing functional outcomes, complications, and reimbursement differences among PAC settings are limited, and no large scale randomized controlled trials have been conducted.12 For patients with hip fracture, Kramer et al13 found that patients admitted to IRFs were no more likely to return home than patients in SNFs. For the same patient population, Kane et al14 showed significantly more ADL independence at 6 months in patients with hip fracture treated in an IRF compared to those in SNF. Munin et al15 concluded that patients with hip fracture, treated in an IRF had superior 12 week functional outcomes compared with those in a SNF with significantly fewer days in the institution. Deutsch and colleagues16 analyzed data from a large rehabilitation outcome system, reporting no difference in patient discharge to the community between IRF and SNF rehabilitation settings in 3 of the 5 case-mix groups (CMG) after controlling for covariates. For the most disabled population (highest CMGs), a slightly higher discharge to community rate was reported for the SNF based programs. The outcome portion of the study reported that the mean Functional Independence Measure (FIM) score of patients receiving care in IRF was significantly higher at discharge.
Among older adults with joint replacement, Walsh and Herbold17 report that case matched patients had a shorter length of stay and superior functional outcomes when treated in IRFs compared to SNFs; those receiving care in the IRF experienced a larger net motor functional gain and achieved these gains in shorter period resulting in better outcomes. In an analysis comparison of Medicare spending and outcomes for lower extremity joint replacement, the working paper of Buntin et al18 to MedPAC reported that IRF patients began their stay at a lower level and achieved a higher level of function at discharge.
There is limited evidence published comparing reimbursement differences between IRFs and SNFs. Deutsch et al16 reported a mean unadjusted Medicare Part A payment for patients with hip fracture as $10,671 for IRFs, compared to the significantly lower payment of $7433 for SNFs, in a sample of patients from the mid 1990s; however, representation of SNF was limited to facilities who submitted data to Uniform Data System for Medical Rehabilitation. While their IRF sample represented 60% of all IRF settings, the SNF sample represented only 11% of those focused on subacute rehabilitation and thus may not be representative of the highly variable SNF setting. It is important to note that many of these studies were conducted prior to implementation of the IRF-PPS.
Among older adults with joint replacement, Buntin et al18 analyzed Medicare claims from 2002 to 2003 data claims, after the implementation of IRF-PPS. Medicare payments for episodes of care in IRFs ($8878.63) were much higher than those for care in SNFs ($4805.02). Although expenses were less in a SNF setting, patients receiving care in SNF had a higher risk of institutionalization than those in IRF settings.
As Medicare explores other provider reimbursement options (including episode payment which combines the entire episode of care into one payment bundle or establishing provider care partnerships), tracking patient outcomes and analyzing 30-day readmission rates will become critical. This proposed system of payment will foster collaboration between institutions in search of efficient and effective clinical models and interventions. Providers will partner and track cost and outcomes across settings and have the potential to discover best practice.19
In an effort to expand the knowledge of identifying the most efficient facilities and services for the treatment of patients with orthopedic diagnoses who are not discharged directly home, we replicated Walsh and Herbold's case matched design comparing outcomes of IRF and SNF total joint replacement patients, but with two key differences: (1) we added patients with hip fracture to the sample and (2) we examined reimbursement differences between settings. We chose to add patients with hip fracture because they are a vulnerable population that is treated in both types of facilities and, on some occasions, their injury results in a THR but continues to be classified as a hip fracture.
Our research questions were the following: (1) Do the functional outcomes of patients treated in the SNFs differ from those achieved by patients treated in the IRF programs? (2) Does the inpatient reimbursement for patients treated in these settings differ? (3) Does the rate of transfer to acute care differ between these settings?
The study used a quasi-experimental case control design to compare the outcomes of primary TKR, THR and patients with hip fracture treated at a 150 bed, freestanding, not-for-profit, university-affiliated IRF to those of patients treated from 5 freestanding SNF settings in the same geographical area. From a sample of 541 patients (409 from the IRF and 131 from the SNF), 102 matched IRF-SNF pairs were created for analysis. From the initial IRF sample, the average age was 72.7 years, with 65% female—12% having a hip fracture, 35% a THR, and 52% a TKR. For the SNF population, the average age was 73.7 years with 71% female—37% hip fractures, 26% hip replacement, and 37% knee replacement. Table 1 provides the clinical characteristics of all participants.
Cases were matched on diagnosis, age, gender, severity index, and initial ambulation FIM score. Because of the strict matching criteria and a large number of patients received care in the IRF, not all of the SNF patients were matched with an IRF counterpart. Physical and occupational therapy interventions were similar as part of the consistent treatment protocol established by the sites, though the intensity and range of services differed. The SNFs were not-for-profit institutions with a designated subacute unit and multidisciplinary staff. The SNF units treated all categories of patients, and patients were not in contiguous space by diagnosis. The staff of the referring acute care hospitals, along with the patient and family input, made the decision to discharge patients from the acute care facility to one of the SNFs or to the IRF. Acceptance by the IRF or SNF was based on medical necessity criteria as well as insurance coverage. This study was reviewed and approved by the institutional review board of the IRF. All patients who enrolled in the study provided written informed consent.
To account for potential confounding variables, the records of patients of the SNF pool were matched on the basis of sex, age (within 4 years), initial ambulation FIM on admission (within 1 point), and 3M severity index to records of patients discharged from the IRF following the same surgical procedures. In instances in which more than 1 IRF patient record was available with the same surgical procedure, age, sex, admission ambulation, FIM, and severity index as an SNF patient's record, the IRF patient's record was chosen at random from those available. The patients were discharged from the IRF or SNF between July 11, 2006, and August 11, 2008.
For each patient receiving care in the SNF setting, the physical and occupational therapist recorded the dates of admission and discharge, discharge setting, and use of home health services following discharge. The FIM, routinely used in IRF setting and used increasingly in SNF environments, is a standardized measure to estimate the amount of assistance needed by the amount of effort required from the patient to perform 18 functional tasks.20 The tool contains 18 items; 13 are related to functional activities of daily living and 5 pertain to cognitive and communicative skills. Admission and discharge motor FIM scores for ambulation, transfers, dressing, and stair negotiation were documented, cognitive FIM scores of problem solving and memory and the ambulation device at discharge were also recorded. The selected items from the FIM21 were chosen to best represent functional change in the orthopedic population and were scored easily by the therapy staff. Scores representing other activities of daily living (ADL), such as bladder/bowel, eating, grooming, bathing, and toileting, traditionally scored by nursing, were not collected by SNF therapists, as it would have been a burden on staff. The therapists also recorded patient age, sex, diagnosis, and date of surgery.
At the IRF, the therapists recorded the motor FIM scores for ambulation, stairs, transfers, and dressing; the cognitive FIM score for problem solving and memory; and the ambulation device used for mobility on admission and discharge in the computerized patient record. The other parameters such as discharge disposition, the need for homecare services, and the comorbid conditions were retrieved from the medical record as entered by other IRF staff such as the social work/case manager and medical record coding staff. All staff was trained on the entire FIM tool by the main campus FIM coordinator.
Reimbursement data was provided by the financial department of each facility for the inpatient episode of care. At the SNFs, this was determined by the RUGs or managed care daily rate. At the IRF, the CMG or managed care negotiated rate was used. These reimbursement rates were based only on payment rates keeping with RUGs and CMGs and did not include patient copays or deductibles.
The 3M All Patient Refined Diagnosis Related Groups (APR DRGs) (3M Center, St Paul, MN) severity of illness is a category value representing a 4-point ordinal scale identifying risk of mortality (1 = minor risk, 2 = moderate risk, 3 = major risk, 4 = extreme/severe risk).1 A UB-04 hospital billing form was retrieved that provided the necessary comorbidity codes for the 3M APR DRG severity and mortality index. This index was utilized to evaluate the interactions of multiple comorbidities, age, procedures, and principal diagnosis. Both SNF and IRF patients in each of the 4 severity levels experience similar resource utilization within the 3M APR DRG assignment. Unlike the CMS DRGs, which focus only on the Medicare population, the 3M APR DRGs reflect all patients seen in an acute care setting.
The main outcome variables of interest were the (1) percent of patients who achieved functional independence using the FIM tool for ambulation, stairs, lower body dressing, and toilet transfers, (2) length of stay, (3) discharge disposition, (4) device used for ambulation at discharge, (5) use of home health services following discharge, (6) frequency of transfers to acute care, and (7) reimbursement for the rehabilitation episode of care.
Differences between matched SNF and IRF patients on continuous measures were assessed using paired t tests. For several measures (total cost, length of stay, and discharge device), the sample size was reduced because of lack of discharge data due to a transfer of the patient to an acute care hospital for treatment of medical complications. Differences in the frequencies of noncontinuous measures were assessed by using McNemar's test. Statistical calculations were performed using Statistical Package for the Social Sciences (Release 11.5).
A total of 540 patients (IRF: n = 409, SNF: n = 131) consented to participate. Because the two populations were different in characteristics of interest at the start of the study, cases were matched using age, gender, diagnosis, severity index, and initial ambulation FIM score. This created 102 patient pairs. Clinical characteristics of the 102 patient pairs are presented in Table 2.
Matched pair analysis revealed no difference in race, days to postacute care, cognitive status (FIM memory and problem solving scores), or functional status (FIM ambulation and lower body dressing) on admission for rehabilitation. The differences found in stair negotiation and toilet transfer are likely attributed to differing practice patterns in IRFs versus SNFs. In many SNF settings, stairs management is not typically assessed until close to discharge. Similarly, the ability to perform toilet transfers is often not rated by OT at the initiation of care. When an activity is not attempted during initial examination in a SNF setting, it is coded as “0.”
Differences between the IRF and SNF settings in the variables of interest are presented in Table 3. For a variety of reasons, 29 of the SNF patients could not be matched. Six of the SNF patients were at a very high level as noted by the admission FIM ambulation scores of 5 or more, whereas none of the IRF patients had a score of more than 4. Seven SNF patients were younger than 60 years and 7 patients were older than 85 years.
Length of stay was significantly shorter in those receiving care in the IRF (P < .001), an expected result consistent with previous studies. Because FIM is an ordinal variable, the authors choose to represent data in terms of percentage of study patients who achieved independence in key functional areas. Ambulation (P = .02) and stairs (P < .001) were statistically different across settings, whereas lower body dressing (P = .15) and toilet transfers (P = .83) were not. The FIM change is also presented as a point of comparison as well but does not reveal information related to overall outcome. Two highly important findings were that those receiving rehabilitation care in the SNF setting were more frequently referred to home care (P < .001) and required the use of a walker at discharge (P < .001). There was no significant difference between settings in terms of transfer to acute care (P = .69), or likelihood of discharge to home (P = .21). Although payments for the inpatient PAC stay associated with an IRF setting are approximately $2000 higher than the SNFs, a high percentage of patients in the SNFs were insured through managed care payers, who typically reimburse at lower negotiated fee for service rate than Medicare (Table 4).
With several options available for PAC following an acute care stay, it is important to identify appropriate patients for each type of setting. Although overlap exits in the availability of therapy between the two primary inpatient settings for postacute care, IRF and SNF, the intensity and full spectrum of services differ between them.
The purpose of this study was to compare the clinical outcomes and reimbursement for care provided in an IRF and several SNFs for patients with THR, TKR, or hip fracture. Because the factors influencing site of postacute rehabilitation after discharge for patients with hip fractures or total joint replacement are not clearly defined, analysis of outcomes is difficult to interpret without a randomized controlled trial. We attempted to address this issue by matching patients treated at SNFs to patients treated at an IRF on the basis of 5 criteria including clinical severity. Despite a significantly shorter length of stay, patients receiving rehabilitation in the IRF had better functional outcomes in the locomotion measures of ambulation and stair negotiation, with fewer requiring a walker at discharge. The increased frequency of walker use suggests that those cared for in SNF settings were less stable and in need of further rehabilitation, primarily in a home-based setting. The shorter length of stay for those receiving care in the IRF may allow them to return to the community more rapidly, resuming their regular activities and lifestyles sooner than those receiving care in SNFs.
The goal of patients following orthopedic diagnosis such as TKR, THR, and hip fracture is to return home as quickly as possible at the highest functional level. Unnecessary institutionalization may have deleterious effects on the patients in the form of reduced muscle mass, reduced endurance, and decline in function. Alley et al22 reported changes in total mass, body composition, and knee extensor strength in community-dwelling older adults following hospitalization especially in those hospitalized for 8 or more days.22 In addition, a study of older men and women aged 70 to 79 years showed that lower leg muscle mass and lower muscle attenuation are associated with poorer lower extremity performance, which was measured by a 6-m walk and a repeated chair standing test.23
This study had several limitations that should be considered when interpreting results. Although the matched-pair analysis is statistically strong, it is not a randomized control trial, and the level of evidence provided is not as powerful. Because the IRF and SNFs used in the study were in the same geographic area, they may not be entirely representative of IRF and SNFs in other states; and generalizability of results may be limited. In contrast, previous studies used outcomes from a national benchmark system, representing a self-selecting subset of the national population of SNFs. The participating SNFs in this study do not contribute to a benchmark system and may be more representative of SNFs in this country. Our data set lack additional patient characteristics such as prior level of function, patient education level, caregiver availability, and prehospital living setting. In addition, issues related to bowel and bladder function, eating, grooming, and other functional skills available in the full FIM scale were absent because the SNF facilities do not routinely collect these items as part of the patient assessment.
During the analysis of reimbursement, we included patients with both Medicare and health maintenance organization–based insurance. This likely affected the length of stay and reimbursement for these patients. We captured the reimbursement for the inpatient stay, but this did not include physician and pharmacy payments or any of the additional services related to follow up care or readmission. These will be important issues when considering bundled payment models for the combined acute and postacute episode of care. Simply comparing payments for the stay in the SNF with the reimbursement of the stay in the IRF setting for the same diagnosis are important but not inclusive. Payments associated with transfers to acute care hospitals as well as the additional home care or outpatient services result in a substantial increase in payment for the total episode of care. When combined with meaningful patient outcome measures, these additional factors may better inform public policy.
The goal of a short length of stay with good functional outcomes is critical for persons recovering from hip fracture and total joint replacement. Determining the optimal setting for rehabilitation following an orthopedic surgery or injury is important for achieving the best possible outcome.
In addition to payment associated with the inpatient episode of illness, there is a need to examine components of care which drive costs, such as intensity of therapy services, frequency of physician interactions, nursing-patient ratios, and additional services of social work and counseling. As bundled payment options are being considered in the health care payment reform, it is also critical to look at all of the resources and services needed to successfully treat each patient across the continuum. Linking the services and costs of all the episodes of care associated the injury or illness (acute care stay, postacute services, home care services, outpatient and physician follow up care, and readmission rates due to complications) is essential for obtaining a complete picture and for maximizing cost-effective outcomes.
This study has policy implications concerning payment for postacute care services, both by Medicare and by private insurers. The Medicare payment difference of $2000 between IRF and SNF would likely be lower if only fee-for-service Medicare was compared. As the concept of bundling is explored as part of the Patient Protection and Affordable Care Act (HR 3590; Pub L No. 111–148), resulting in an episode-based payment, careful analysis of cost, outcomes, and readmission is needed.
A balance between access to services and cost-efficient treatment that results in the best long-term outcome is an important goal of the Medicare program. The results of the current study suggest that orthopedic patients in need of inpatient postacute rehabilitation services have superior functional outcomes in a shorter length of stay at an IRF, although treatment in an IRF was more costly than treatment in a SNF. Future studies should examine total costs of the episode of care and should attempt to identify more cost-effective treatment strategies in all settings.
The authors thank B. Timothy Walsh, MD, for his valuable contributions to the study design and the therapists at Burke Rehabilitation Hospital and the affiliating SNFs for their assistance with the FIM data collection.
1. Medicare Payment Advisory Commission. Report to the Congress: Issues in a Modernized Medicare Program. Washington, DC: MedPAC; 2005.
2. Gage B. Impact of the BBA on post-acute utilization. Health Care Fin Rev. 1999;20:103–126.
3. Ashby J, Guterman S, Greene T. An analysis of hospital productivity and product change. Health Affairs. 2000;19:197–205.
4. De Pablo P, Losina E, Phillips CB, et al. Determinants of discharge destination following elective total hip replacement
. Arthritis Care Res. 2004;51:1009–1017.
5. Salcido R, Moore RW, Schleenbaker RE, Klim G. The physiatrist and subacute rehabilitation. Phys Med Rehabil Clin N Am. 1996;7:55–81.
6. Buntin MB, Garten AD, Paddock S, Saliba D, Totten M, Escarce JJ. How much is postacute care use affected by its availability? Health Serv Res. 2005;40:413–434.
7. Kaplan SJ. Growth and payment adequacy of medicare postacute care rehabilitation. Arch Phys Med Rehabil. 2007;88:1494–1499.
8. Keith RA. Treatment strength in rehabilitation. Arch Phys Med Rehabil. 1997;78:1298–1304.
9. DeJong G, Hsieh C-, Gassaway J, et al. Characterizing rehabilitation services for patients with knee and hip replacement in skilled nursing facilities and inpatient rehabilitation
facilities. Arch Phys Med Rehabil. 2009;90:1269–1283.
10. DeJong G, Horn SD, Smout RJ, Tian W, Putman K, Gassaway J. Joint replacement rehabilitation outcomes on discharge from skilled nursing facilities and inpatient rehabilitation
facilities. Arch Phys Med Rehabil. 2009;90:1284–1296.
11. Jette DU, Warren RL, Wirtalla C. The relation between therapy intensity and outcomes of rehabilitation in skilled nursing facilities. Arch Phys Med Rehabil. 2005;86:373–379.
12. Khan F, Ng L, Gonzalez S, Hale T, Turner-Stokes L. Multidisciplinary rehabilitation programmes following joint replacement at the hip and knee in chronic arthropathy. Cochrane Database Syst Rev. 2008(2).
13. Kramer AM, Steiner JF, Schlenker RE, et al. Outcomes and costs after hip fracture
and stroke: a comparison of rehabilitation settings. J Am Med Assoc. 1997;277:396–404.
14. Kane RL, Chen Q, Finch M, Blewett L, Burns R, Moskowitz M. The optimal outcomes of post-hospital care under medicare. Health Services Res. 2000;35:615–661.
15. Munin MC, Seligman K, Dew MA, et al. Effect of rehabilitation site on functional recovery after hip fracture
. Arch Phys Med Rehabil. 2005;86:367–372.
16. Deutsch A, Granger CV, Fiedler RC, et al. Outcomes and reimbursement of inpatient rehabilitation
facilities and subacute rehabilitation programs for Medicare beneficiaries with hip fracture
. Med Care. 2005;43:892–901.
17. Walsh MB, Herbold J. Outcome after rehabilitation for total joint replacement at IRF and SNF: a case-controlled comparison. Am J Phys Med Rehabil. 2006;85:1–5.
18. Buntin MB, Partha D, Escarce JJ, Hoverman C, Paddock SM, Sood N. Comparison of Medicare Spending and Outcomes for Beneficiaries with Lower Extremity Joint Replacements. Santa Monica, CA: RAND. Working paper WR-271-MedPAC; 2005.
19. Dejong G. Bundling acute and postacute payment: from a culture of compliance to a culture of innovation and best practice. Phys Ther. 2010;90:658–662.
20. Ottenbacher KJ, Hsu Y, Granger CV, Fiedler RC. The reliability of the functional independence measure: a quantitative review. Arch Phys Med Rehabil. 1996;77:1226–1232.
21. 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.
22. Alley DE, Koster A, MacKey D, et al. Hospitalization and change in body composition and strength in a population-based cohort of older persons. J Am Geriatr Soc. 2010;58(11):2085–2091.
23. Visser M, Kritchevsky SB, Goodpaster BH, et al. Leg muscle mass and composition in relation to lower extremity performance in men and women aged 70 to 79: the Health, Aging and Body Composition Study. J Am Geriatr Soc. 2002;50:897–904.