If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
DeJong G, Hsieh C-H, Gassaway J, Horn SD, Smout RJ, Putman K, James R, Brown M, Newman EM, Foley MP. Characterizing rehabilitation services for patients with knee and hip replacement in skilled nursing facilities and inpatient rehabilitation facilities.
To characterize rehabilitation services for patients with knee and hip replacement in 3 types of postacute facilities in the U.S.
Length of stay (LOS), amount and intensity of physical therapy (PT) and occupational therapy (OT), types of therapy activities.
Average LOS was about 15 days for freestanding SNF patients, and 9 to 10 days for hospital-based SNF and IRF patients. Freestanding SNFs and IRFs provide about the same number of hours of PT and OT; the hospital-based SNF provided 27% fewer hours. Freestanding SNFs and the hospital-based SNF provided fewer hours a day than did IRFs. Joint replacement patients across all 3 types of facilities spent, on average, 70% to 75% of their PT time in just 2 activities—exercise and gait and spent 56% to 66% of their OT time in 3 activities—exercise, functional mobility, and dressing lower body.
Both freestanding SNFs and IRFs provided similar amounts of PT with a similar emphasis on exercise and gait activities. IRFs, however, provided more OT than freestanding SNFs. IRFs had shorter LOSs and more intensive therapy services than freestanding SNFs. Study freestanding SNFs exhibited greater variation in LOS and intensity of therapy than IRFs.
A KNEE OR HIP REPLACEMENT is one of the most common surgical procedures in the United States, and an increasing number of patients who acquire a joint replacement also obtain some form of postacute rehabilitation.
Much debate about the merits of postacute rehabilitation for patients with joint replacement has centered on whether one setting of care provides superior outcomes relative to another, particularly in the case of SNF and IRF care. Very little consideration, however, has been given to the content of rehabilitation care rendered to patients with joint replacement in these 2 postacute settings. This article addresses 4 questions: how SNFs and IRFs differ in (1) number of days from joint replacement to SNF or IRF admission (onset days), (2) LOS in an SNF or IRF, (3) amount and intensity of therapy (ie, hours of PT or OT a day), and (4) actual content of PT and OT rendered. The article also considers how these differences need to be taken into account when comparing SNF and IRF outcomes for patients with joint replacement. This article reports findings from the JOINTS I study conducted from 2005 to 2007. An accompanying article by DeJong et al
have attempted to examine select PT activities, mainly exercise and gait training, and their effects on functional outcomes for patients with joint replacement. These studies typically use treatment-control group designs that have small samples and allow an investigator to examine only 1 or 2 therapy activities or interventions at a time, not the full range of activities in which the patient participates. In a randomized trial, Munin et al
(N=71) examined the timing at which to begin inpatient rehabilitation after knee or hip replacement. They provided some description of both the types and amount of therapy (eg, PT, OT, recreational therapy), but not the content of therapy rendered daily in the treatment and control groups.
concluded that routinely reporting hours of service by discipline and developing a taxonomy of treatment types can help provide a better understanding of the black box of rehabilitation. In the case of joint replacement rehabilitation in SNFs and IRFs, we need to illustrate differences and/or similarities in the content, timing, intensity, and duration of care provided. Without these characterizations, both settings continue to be viewed as black boxes. In the absence of such characterizations, prudent purchasers, both government and private payers, cannot fully know what it is that they are purchasing. Purchasers and providers alike need to know which clinical activities and interventions are associated with better patient outcomes and in which setting they are most likely to find these therapeutic ingredients.
(A note to non-American readers. IRF-level care refers to care provided in hospital-based rehabilitation centers. IRFs can be freestanding rehabilitation hospitals or rehabilitation units within acute care hospitals. SNF-level care refers to care provided in SNFs, many of which also have a residential component for long-term care. SNFs, like IRFs, can be freestanding [often in association with a free-standing residential nursing home] or can be a unit within an acute care hospital absent any residential program. This latter group is known as hospital-based SNFs.)
The JOINTS study was a prospective observational cohort study involving 22 facilities from across the country. A key feature of the study was the participation of a transdisciplinary clinical practice team that consisted of each site's coinvestigator and lead clinicians from each discipline at each site, including physicians, physical therapists, occupational therapists, nurses, and case managers. The clinical practice team was involved in the entire study process. Its participation was facilitated by face-to-face planning and training meetings, and weekly conference calls supplemented with discipline-specific conference calls. The clinical practice team refined patient selection criteria for the study, developed discipline-specific POC documentation forms, reviewed severity scoring criteria, identified patient and process of care variables to be abstracted from the patient chart, and contributed front-line clinical expertise to the analysis of study data. In short, the study offered a naturalistic view of rehabilitation treatment by examining what happens during the process of care. The clinical practice team facilitated clinical buy-in and a high level of transparency across participating facilities.
Study Facility Recruitment and Participation
The primary goal of site selection was to achieve geographic diversity by recruiting at least 2 SNFs and 2 IRFs in each of 4 major census regions: northeast, south, midwest, and west.
It was not feasible to obtain a national representative sample of facilities through probability sampling because of funding limitations. Other facility recruitment and selection considerations included (1) free-standing versus hospital-based facilities, (2) profit versus nonprofit organizations, (3) high managed care versus low managed care market penetration, and (4) capacity/ability to bring at least 200 patients into the study. Initially, 11 SNFs and 11 IRFs participated in the study. Facility participation was entirely voluntary. Because we found significant data quality issues at 1 SNF and experienced a long delay in data acquisition from another SNF with very few patients, we do not include these 2 sites' data here. As we report, the study's 1 hospital-based SNF had materially different practice patterns. To characterize facility differences and practice patterns better, we present findings for 3 types of facilities: (1) freestanding SNFs, (2) hospital-based SNF, and (3) IRFs.
Patient selection criteria were broad and inclusive with the aim of capturing patient variation and enhancing generalizability of findings. Accordingly, patients who met the following criteria were included in the study: (1) 21 years of age or older, (2) underwent hip or knee replacement of any type for any reason, and (3) admitted to a participating SNF or IRF for joint replacement rehabilitation from any acute care source. For purposes of this article, we exclude patients who had a hip fracture prior to their hip replacement because they were found to be significantly different from the rest of the study population both demographically and clinically. Patient enrollment began in February 2006 and ended in February 2007.
The study used 3 key patient assessment and process of care data collection tools: (1) the CSI, (2) the FIM, and (3) the study's POC data collection forms.
The CSI is a condition-specific physiologic, functional, and psychosocial severity of illness measure that defines disease severity objectively. This instrument has been validated extensively in many inpatient, ambulatory, rehabilitation, and long-term care settings since 1982.
The CSI provides a consistent method for defining patient severity of illness based on over 2200 individual signs, symptoms, and physical findings related to a patient's diseases or conditions, not just on International Classification of Diseases–9th Revision–Clinical Modifications diagnostic coding.
CSI produces a continuous measure (0 to any positive integer) for each diagnosis. It also produces an overall continuous score that reflects interactions of severity of all the patient's health conditions. In this article, we report on the CSI score at admission (first 24 hours).
The FIM served as the study's principal measure of patient functional status on admission and discharge. One vexing issue for this study was that SNFs and IRFs use different functional assessment instruments. IRFs use the FIM embedded in the IRF–Patient Assessment Instrument, and SNFs use the MDS, a portion of which addresses domains included in the FIM. Further, SNFs and IRFs administer their respective assessment instruments at different points in time: the FIM is administered on admission and discharge, whereas the MDS is administered on the fifth and fourteenth day of a patient's stay. Also, the FIM reports the patient's most dependent behavior (burden of care), while the MDS reports the patient's usual behavior. Previous attempts to create a crosswalk between these 2 measures have not been successful.
In the absence of a reliable crosswalk between these 2 instruments, the study team elected to use the FIM to measure functional independence in both study SNFs and IRFs because it was the more parsimonious of the 2 instruments and was already being administered in 1 of the settings (IRFs) at time points needed—that is, at admission and discharge.
To ensure that the FIM was administered appropriately, IT HealthTrack, a FIM training and follow-up survey organization, was engaged to train clinical staff at each study site. SNF clinicians completed a 3-day training session in which they were taught how to use the FIM. Each clinician was required to score 100% on an examination that tested their knowledge of the FIM and its uses. This mimics FIM training procedures required of IRF clinicians. While we did not conduct formal reliability checks thereafter, we did evaluate FIM scoring to determine whether there were any anomalous FIM scores, at both the individual and facility level, relative to what else was known about individual patients—for example, age, and admission CSI—to determine whether corrective action was required.
The POC documentation system recorded activities and interventions associated with each therapy session at the time therapy was rendered or immediately thereafter. The POC documentation system was developed by discipline-specific teams of clinicians (physical therapists, occupational therapists, nurses, physicians) drawn from the study's clinical practice team and participating facilities.
Through an iterative process and many weekly teleconference calls, the PT and OT teams identified therapy activities and interventions that they believed best characterized their respective areas of practice. Activities refer to functional and medical targets associated with a therapy; interventions refer to specific methods used to address these activities. For example, physical therapists identified 16 PT activities ranging from formal assessment and pain management to exercising, transfers, and gait. Study physical therapists then identified a total of 35 distinct interventions that could be associated with an activity. Similarly, occupational therapists identified 28 therapy activities and 36 OT interventions. To assure consistency in how clinical activities were recorded, definitions for each therapeutic activity and intervention were included on the back of each POC form (see appendix 1 for the PT and appendix 2 for the OT POC forms and their respective definitions).
Therapists recorded time spent in each activity. Multiple interventions could be used to address a given activity, and the same intervention could be used to address multiple activities. For example, an occupational therapist might record that she spent 30 minutes working on lower-body dressing using multiple interventions (safety education, activity tolerance, balance training, adaptive equipment). Participating clinicians found that the 2-dimensional taxonomy of activities and interventions had heuristic value, had face validity, and captured the content of care they provided.
We did not develop a POC instrument for speech therapy because patients with joint replacement typically neither need nor use speech therapy services, but we did review the patient chart for receipt of speech therapy services, if any. Study nurses developed a POC documentation system that captured shift-specific information such as the number of times a patient was assisted with transfers and ambulation and the number of education sessions completed. Study physicians eventually abandoned the notion of a POC instrument for their discipline and chose instead simply to augment the patient chart.
After a project-wide training session, each clinical leader conducted onsite training sessions for their coworkers. The study team facilitated this training for each clinical specialty using a training manual that included instructions for completing forms, definitions for all terms used, and several case scenarios to demonstrate use of each form. A completed case scenario with correct POC entries was presented to show clinicians how the discipline-specific documentation form was to be used for a specific case. The lead clinician reviewed 2 additional case scenarios and completed the entries together with the clinicians at each site. Last, clinicians had to complete POC entries independently after reading 3 additional scenarios to evaluate their understanding of instructions. The trainer at each site then reviewed and discussed responses with their clinicians until they demonstrated full understanding of how to document the form. Supplemental 1-on-1 training and review was performed by the lead clinician whenever appropriate. Telephone conference calls throughout the data collection period provided an opportunity for clinicians to discuss implementation issues and ask questions about specific situations.
Study facility characteristics
The study team obtained data on study facility characteristics from 2 sources: (1) a publicly available provider of service file maintained by the CMS, and (2) a brief 1-page supplemental questionnaire developed by the study team to obtain data directly from facilities for the purpose of confirming provider of service data and obtaining additional facility characteristics that were not available in CMS provider of service files. The questionnaire was completed by the study site director with assistance from others at the facility who were most able to answer the question at hand.
Process of care data
Rehabilitation POC documentation forms were completed for each therapy session. Each study site identified a contact person in each discipline to facilitate timely completion of the form, monitor and track missing data entries and assure quality of form entries on a daily basis throughout the collection period. A total of 31,345 PT sessions and 28,637 OT sessions were completed.
Although the nursing POC documentation was developed by study nurses, nursing compliance in completing nursing POC documentation was limited. Only 3 SNFs and 7 IRFs contributed reliable nursing data, and thus, nursing data are not presented in this article.
The study team worked closely with the clinical practice team to identify study variables that could be obtained directly from the medical chart after patient discharge. These data included patient characteristics (severity of illness using CSI, FIM, and demographics), process characteristics in addition to POC documentation (physician care, medications), and outcome variables such as medical complications and discharge destination. Each chart abstractor completed a 3-day training session during which efficient and accurate collection of chart abstraction was described, reviewed, and practiced using a sample chart created by the study team and real charts. After the training session, each chart abstractor underwent a rigorous reliability testing process to ensure complete and accurate data collection. Chart abstraction reliability was conducted periodically throughout the data collection period. An agreement rate of 95% for all elements abstracted between each data abstractor and an experienced project team chart reviewer was required for each reliability test. For example, chart reviewers started working on their own charts after training. After an abstractor completed 10 actual patient charts, an experienced project team chart reviewer selected 2 charts (among these 10) with highest CSI scores and longer LOSs to conduct a reliability test. The experienced study team chart reviewer reviewed the full chart and compared her entries on all elements with those by the chart abstractor. If a 95% agreement was not met, the study team reviewed the concepts that the chart abstractor did not understand. The chart abstractor then continued to enter 3 more charts. The study team selected another chart for reliability. Once 95% agreement was obtained, the chart abstractor was allowed to continue with abstraction. If an abstractor entered more than 100 charts, the study team conducted an additional reliability around the one hundredth chart to assure the desired rate of agreement was still being met.
Descriptive analyses were used to examine frequencies of categorical variables as well as mean, median, and SD for continuous variables. We also conducted bivariate analyses with chi-square and analysis of variance tests to examine differences when using categorical and continuous variables. Given the large array of PT and OT activities, the various combinations in which they could be rendered, and the multiple ways in which they could be characterized, we developed graphical displays to help simplify the complex interrelationships of different therapy activities across settings and types of patients.
Table 1 presents results from the facility survey. Freestanding SNFs had almost twice the number of beds as IRFs in the JOINTS study. Most study IRFs were nonprofit, while over half of study freestanding SNFs (62.5%) were for-profit organizations. Most study SNFs and IRFs are free-standing facilities; most of the nation's IRFs, however, are hospital-based units. Study IRFs had a much higher average number of joint replacement patient admissions in 2006 than did freestanding SNFs, and the hospital-based SNF had the highest number of 2006 admissions among all participating facilities.
Mean age of the study sample was 71.2 years; the sample was mostly female (65.1%) and white (81.4%); 35.2% lived alone prior to their joint replacement (table 2). Freestanding SNF knee and hip patients were significantly older than IRF knee and hip patients. Non-Health Maintenance Organization Medicare (also known as fee-for-service Medicare) was the most common payer, especially among freestanding SNF patients; IRFs and the hospital-based SNF admitted more patients covered by private health plans. Freestanding SNF patients and IRF patients had similar rates of revisions; IRF patients were more likely to have had a bilateral replacement, especially among patients with knee replacement. IRF patients presented more medical acuity as noted by their higher CSI scores at admission and were admitted with lower functional scores as noted by their admission FIM scores. Mean BMI was 31.8 (BMI ≥30 denotes obesity). Freestanding SNFs and IRFs had similar proportions of patients with common comorbidities such as hypertension, diabetes, and morbid obesity (BMI ≥40); hospital-based SNF patients had higher proportions of these conditions. Overall, IRFs admitted a more severe case mix than other facilities as reflected in the distribution of patients in the Medicare CMGs noted in Table 2, Table 3, Table 4.
Patients who had 3 out of the following 4 possible symptoms: BMI ≥35, hypertension, diabetes, and lipid disorders.
Ischemic heart disease (%)
Functional status at admission
Motor FIM, mean ± SD
Cognitive FIM, mean ± SD
CMG groupings (%)
Acute and postacute care
Time from replacement to rehabilitation (d), mean ± SD
Postacute LOS (d), mean ± SD
NOTE. Does not include patients who had a hip replacement after a hip fracture. A–C indicate mean setting differences using the Duncan multiple comparisons test. F statistic is the value of F from analysis of variance.
The average joint replacement patient started rehabilitation stay 4 days after the actual replacement (see table 2). There was greater variation in the number of onset days within setting than between settings. Higher-volume facilities (more joint replacement patients a year) admitted patients more quickly (<4d) than lower-volume facilities (>4d).
Length of stay
Among patients with knee replacement, average LOS was about 6 days longer for freestanding SNF patients than for IRF patients (14.8 vs 8.9d) (see table 2). Among patients with hip replacement, average LOS for freestanding SNF patients was about 5 days longer than for IRF patients (15.0 vs 10.1d). Average LOS for hospital-based SNF patients approximated that of IRF patients (8.8–8.9d). Variation in LOS was much greater for freestanding SNF patients than for IRF patients or hospital-based SNF patients (see table 2 and fig 1). Adjusted for case mix, we observed much larger differences in average LOS between types of facilities. For example, average LOS among patients with knee replacement in CMG 805 and 806 was 27.0 days for freestanding SNF patients, 13.1 days for hospital-based SNF patients, and 12.5 days for IRF patients (see table 4).
Amount and intensity of therapy
Table 5 describes the kinds and amounts of therapy services rendered in each type of setting. Freestanding SNFs and IRFs provide about the same amount of total therapy over the course of a patient's stay. The hospital-based SNF provided fewer total hours of therapy because of its shorter LOS (akin to IRFs).
Table 5Rehabilitation Services Rendered in SNFs and IRFs
P (F statistic)
P (F statistic)
Minutes per session
No. PT sessions
No. PT sessions/patient/d
Total hours PT/patient
Minutes per session
No. OT sessions
No. OT sessions/patient/d
Total hours OT/patient
Total (PT and OT combined)
Minutes per session
Total hours therapy/patient
NOTE. Values are mean±SD. Excluded patients with no therapies. A–C indicate mean setting differences using the Duncan multiple comparisons test. F statistic is the value of F from analysis of variance.
In characterizing therapy intensity—for example, hours of PT or OT a day—we included only weekdays (Monday–Friday). We excluded weekend days because of different practice patterns in providing weekend therapy. IRFs provide a higher intensity of therapy—about 1 hour more of PT and OT combined a day—than did freestanding SNFs or the hospital-based SNF.
Variation in therapy intensity and length of stay
Figure 1 characterizes differences among study facilities in therapy intensity and LOS. Figure 1 has a 2-dimensional characterization that underscores practice differences between freestanding SNFs and IRFs. It also illustrates the degree of practice variation within types of facilities, especially the greater variation among participating freestanding SNFs. IRFs demonstrated less variation in intensity and LOS, perhaps because of the 3-hour rule that requires IRFs to provide 3 or more therapy hours per patient per day in order to be defined as an IRF for Medicare payment purposes. Overall, we found an inverse relationship between LOS and intensity of therapy—the longer the LOS, the less intense the services. Figure 1 also underscores the hospital-based SNF's distinct practice pattern.
Fig 2, Fig 3, Fig 4, Fig 5 provide graphical representations of the total amount of PT and OT rendered as well as the proportions of therapeutic activities used in each type of facility. Exercise and gait training were the 2 most common PT activities for patients with knee and hip replacement in both freestanding SNFs and IRFs. Approximately 70% or more of PT time was devoted to these 2 activities. Exercise activity was also the most common activity in OT and constituted about 40% of the patient's OT time in both freestanding SNFs and IRFs. Exercise constituted 15% to 16% of OT time in the hospital-based SNF, however. Apart from exercise, we found more variation in OT activity time across the 3 types of facilities.
We also identified the 4 most frequently used interventions within the 2 most common therapeutic activities in both PT and OT by type of joint replacement (knee or hip). In PT, the 4 most common interventions used in exercise activity (defined as patient's physiologic body exertion or passive exercise by therapist to improve body function) with patients with knee and hip replacement were therapeutic exercises (defined as any strength-building exercises, such as active, active assisted, resisted, and passive range of motion), manual therapy/stretching, cueing (tactile, verbal), and noninvolved lower extremity. The most common PT interventions used in gait training activity with patients with knee and hip replacement were even surface, cueing, functional endurance, and safety education.
The 4 most common OT interventions associated with exercise activity among patients with knee and hip replacement included therapeutic exercises, conditioning exercises, activity tolerance, and integration with nonsurgical limbs (including upper extremity). The most common interventions associated with functional mobility activity included safety education, DME education, activity tolerance, and balance training among patients with knee replacement. Similarly, the most common interventions among patients with hip replacement included safety education, DME education, orthopedic precautions, and activity tolerance.
Overall, we find that freestanding SNFs and IRFs deliver similar packages of PT and OT activities over the course of a patient's stay. The main differences between settings are the average LOS and the intensity of their services with less intense therapy provided in freestanding SNFs and the hospital-based SNF, and more intense therapy provided in IRFs.
The primary purpose of this article is to characterize the process of care rendered to patients receiving joint replacement rehabilitation across 3 settings of care—freestanding SNFs, hospital-based SNFs, and IRFs. On average, patients came to postacute rehabilitation 4 days after their joint replacement. Patients stayed 5 to 6 days longer in freestanding SNFs than in IRFs—not taking into account case-mix differences. Hospital-based SNF patients had about the same LOS as their IRF counterparts.
Freestanding SNF and IRF patients spend roughly the same amount of time in PT over the course of their inpatient stay; freestanding SNF patients spend about an hour or less time in OT than IRF patients. Hospital-based SNF patients spend nearly a third less time in therapy than do their freestanding SNF and IRF counterparts.
Exercise and gait accounted for more than 70% of total PT time in the 3 settings regardless of type of joint replacement. In OT treatment, however, exercise accounted for approximately 40% of treatment time in freestanding SNFs and IRFs. Use of other OT activities varied more across settings and types of patients.
Apart from differences in LOS, the distinguishing difference between SNFs and IRFs was not the total amount or content of therapy but intensity of therapy—that is, the number of therapy hours received a day—with IRF patients participating in a more intensive daily therapy regimen. This study uncovered greater variation in practice among SNFs than among IRFs when one considers both therapy intensity and LOS concurrently (see fig 1).
Implications for Policy, Practice, and Research
There is a fundamental trade-off in how the nation allocates postacute rehabilitation services—less intensive therapy over a longer LOS (eg, freestanding SNF-level care) versus more intensive therapy over a shorter LOS (eg, IRF-level care). This begs the question of which approach produces better outcomes, not taking into account other variables such as facility size, onset days, staffing, and organizational culture.
The exception to this observation is the study's hospital-based SNF, a hybrid facility that offers the intensity of a freestanding SNF and the LOS of an IRF. Hospital-based SNFs are a breed apart from other postacute facilities. Although paid like freestanding SNFs, they draw on resources of the host hospital. According to the Medicare Payment Advisory Commission, hospital-based SNFs have margins of −83.8%
To lump hospital-based SNFs with free-standing SNFs reduces the level of resolution in understanding the practice patterns, resource use, performance, and costs across the postacute spectrum.
This study was unusual in that it attempted to examine the full range of PT and OT activities and interventions used in joint replacement rehabilitation and quantified the time spent in each therapeutic activity. Other studies typically examine 1 or only a few therapy activities or interventions at a time and their association with outcomes. These studies are often small-sample studies.
The inclusive nature of the discipline-specific POC documentation system allows one to peer inside the black box of joint replacement rehabilitation in both SNFs and IRFs. More importantly, it allows one to examine the relationship of individual therapeutic activities and interventions with patient outcomes in the search for best practice in joint replacement rehabilitation. Such research can contribute greatly to the development of clinical pathways and practice guidelines that are more contextualized and nuanced to reflect real world practice than those anchored solely in randomized trials. It also can help to inform health plans, payers, and policy makers to make more informed postacute care choices.
Limitations and Generalizability of Findings
Five study limitations are especially noteworthy. First, this study does not capture the contributions of nursing services to SNF and IRF care. Nursing was well represented in the study design process. Nurses at many study facilities, however, were unable to complete nursing POC forms, although they were considerably simpler than the therapy forms.
Second, this study does not adequately capture the contributions of physician care in the postacute rehabilitation process. One of the distinguishing features of IRF-level care is presumably its 24-hour onsite access to physician care. Physicians at local sites were reluctant to develop a POC documentation system for physician care and elected instead to augment the patient chart. In the absence of such a system, we were unable to obtain standardized information across sites.
Third, while the study made every attempt to train and test therapists at the outset and regularly monitor ongoing use of study therapy POC documentation, we conducted no formal interreliability tests once the system was up and running. The consistency of reported practice patterns across sites, however, provides increased confidence in the reliability of the results. Formal interrater testing beyond the training period would have been preferred but logistically challenging.
Fourth, as noted in the Methods section, we can quantify the time spent in a given activity, but we cannot quantify fully the extent to which an intervention was used, only whether the intervention was used in association with a given activity. In short, we can identify the percentage of activities or therapy sessions in which an intervention occurred. An intervention has meaning mainly in the context of the therapeutic activity or functional goal with which it is associated.
Fifth is potential selection bias. The study was not based on a national probability sample of facilities and patients. Participation in the study was voluntary for all study facilities. Accordingly, the study was prone to attract a better breed of both SNFs and IRFs—facilities that were more motivated and equipped to participate in research and felt more secure about sharing their clinical data with outside research organizations. Rehabilitation units of acute care hospitals were not represented. Given the diverse practice patterns among SNFs, the study also could have benefited from participation of more SNFs. Nonetheless, the study did acquire a geographically diverse array of facilities. All of these considerations need to be taken into account when attempting to generalize study findings to the larger universe of all SNFs and IRFs.
At the patient level, study patients were similar to those found in other observational cohort studies of patients with joint replacement. Their age and sex distributions, BMI scores, and comorbidities were similar to those in other studies.
in a study of patients with knee replacement and revision in IRFs (n=424), reported comorbidity rates similar to those seen in this study for patients with knee replacement, including similar rates of diabetes (20.1% vs 22.6%) and ischemic heart disease (15.5% vs 15.9%).
Likewise, average LOS for study IRF patients was similar to average LOS found in previous studies.
and similar to average LOS in the 2006 Medicare data furnished by CMS to the study team (calendar year 2006 was the year most aligned with the study data collection period; Kimberly Skellan, Center for Medicare and Medicaid Services, unpublished data, 2007). Time from surgery to IRF admission is similar to national data
(computed by the Center for Post-acute Studies, National Rehabilitation Hospital, using 2005 Healthcare Cost and Utilization Project National Inpatient Sample data, Agency for Healthcare Research and Quality), but about 1 day shorter than time documented in a previous case-control study on hip replacement.
Comparisons with earlier studies offer increased confidence about the generalizability of study findings.
Study freestanding SNFs exhibited a greater diversity of practice patterns with respect to LOS and intensity of daily therapy (see fig 1) than did IRFs. IRFs provided, on average, about 1 more hour of therapy per day (PT and OT combined) than did freestanding SNFs and discharged their patients about 5 days earlier than did freestanding SNFs. Over the course of a patient's stay, freestanding SNFs and IRFs provided similar amounts of PT with a similar emphasis on exercise and gait training. IRF patients received more OT during their stay than did freestanding SNF patients. During their OT sessions, freestanding SNF and IRF patients spent about 40% of their time doing exercise activities.
This is the first known study that has attempted to identify the full range of PT and OT activities and interventions associated with joint replacement rehabilitation in SNFs and IRFs. With this characterization, we can begin to determine how duration and intensity of PT and OT overall and how their component activities and interventions are associated with patient outcomes. These associations can assist clinicians, payers, and policy makers alike to make more informed choices about the appropriate level of rehabilitation care and types of care needed for patients with joint replacements.
We thank the clinical staff at each of the 22 clinical sites represented in the JOINTS study for their role and contributions. The site study's site directors included John Bishop and Erin Propst; Charles Schauer, PhD, and Flo Singletary, MS; Hilary Siebens, MD, and Harriet Aronow, PhD; Jacalyn Lichtenstein, RN, and Lynne Wright, RN, CRRN; Julie Barth, OTR/L; Andrea Curry, RN; Jon Nordrum, Melissa P. Hake, Donna Coughlin, and Karen Kotval; Santiago Toledo, MD; Robert Krug, MD, and Steve Kunsman; Robert Bowsher, MD, and Cynthia A. Kreutz; Barbara Higgins, PT; John Jimenez; Gina Harris, PT, Natalie Russo, OT, and Marcy Howard; Steve Christensen and Lyle Black; Jean Russell, Elisa Freeman, and Andrea Maddux; Ann Cottrell, Dawn Haskell, PT, and MaryAnn Morrison, RN; Ellen Logsdon; Bernard Shore, MD, Cindy Lovetro, Karen Rowlands, and Deborah Greco; Kathy Stover; Karen L. George, OT, Suzanne Besecker, Audrey Hartz, RN, Geraldine Essick, and Pragna Doshi, PT; and Mary VandeKamp, Vienna Lafrenz, Glenda Mack, Luanne L. Cunningham, and Jane Klugman.
We thank the members of our staff who also contributed significantly to the success of this study: Cathy Ellis, PT, Naomi Greenberg, PT, Deborah Hutton, OTR/L, and Josephine Kuofie, RN (National Rehabilitation Hospital, Washington, DC).
We also acknowledge the input and contributions of all members of the study's Policy Advisory Panel and the organizations they represent: Rosaly Correa-de-Arauio, MD, PhD; Barbara Manard; Robyn Stone, PhD; Rochelle Archuleta, MSHA, MBA; Carolyn Zollar, JD; Mary Fran Delaune; Trudy Mallinson, PhD, and Anne Deutsch, PhD; Christine MacDonell; Barbara Braun, PhD; Elizabeth Sandel, MD; Susan Klanecky; Michael Weinrich, MD; Melinda Buntin, PhD; Reginald Warren, PhD; Leigh Callahan, PhD; Carl Granger, MD; and Michael Munin, MD. We also thank those nonmembers of the Policy Advisory Panel who participated as observers because of their employment with policy-making agencies and funding sources: James Bowman, MD; Ruth Brannon, MSPH, MA, and Philip Beatty, PhD; and Dexanne Clohan, MD, Justin Hunter, JD, and John Markus, JD.
Appendix 1: PT Documentation
Appendix 2: OT Documentation
Looking upstream: factors shaping the demand for postacute joint replacement rehabilitation.
Supported by the HealthSouth Corp, ARA Research Institute of the American Rehabilitation Providers Association, Brooks Health, National Rehabilitation Hospital, American Hospital Association, the Federation of American Hospitals, and others.
A commercial party having a direct financial interest in the results of the research supporting this article has conferred or will confer a financial benefit on the author or one or more of the authors.