Association between patient and facility characteristics and rehabilitation outcomes after joint replacement surgery in different rehabilitation setting for older adults: a systematic review

Abstract

Background and Purpose:

In the United States, an exponential increase in total hip arthroplasty (THA) and total knee arthroplasty (TKA) demand has occurred over the last two decades. Evidence suggesting patients receiving inpatient rehabilitation following a TKA or THA experience similar outcomes as those with rehabilitation in other settings led to dramatic shifts in post-surgical care settings owing to CMS payment reforms. A contemporary synthesis of evidence about the association between patient and facility factors and outcomes from older adults undergoing THA or TKA in the United States is needed.

Methods:

To identify eligible studies, we searched PubMed, Scopus, and CINAHL. We followed PRISMA guidelines to identify articles evaluating either patient or facility factors associated with outcomes after THA or TKA for older adults who may have been cared for in inpatient settings (i.e., inpatient rehabilitation or skilled nursing facility). Eligible articles were conducted in the United States and were published between January 1, 2000, and December 31, 2021.

Results:

We included 8 articles focused on patient factors and 9 focused on facility factors. Most included older adults and the majority were White (in those reporting race/ethnicity). Most studies evaluated outcomes at discharge and showed that patients admitted to inpatient rehabilitation facilities had either similar or better functional outcomes (mobility, self-care, Functional Independence Measure (FIM) score) and lower length of stay compared to those in skilled nursing facilities (SNFs). Few studies focused on home healthcare.

Conclusions:

The systematic review focused on older adults showed that findings in these patients are consistent with previous research. Older adults undergoing THA/TKA had acceptable outcomes regardless of post-surgical, inpatient setting of care. Research conducted after CMS payment reforms, in home healthcare settings, and in more diverse samples is needed. Given the known racial/ethnic disparities in THA/TKA and the shifts to post-surgical home healthcare with little regulatory oversite of care quality, contemporary research on outcomes of post-surgical THA/TKA outcomes is warranted.

INTRODUCTION

In the United States, an exponential increase in total hip arthroplasty (THA) and total knee arthroplasty (TKA) demand has been observed over the last two decades.^1–3 Both THA and TKA are cost-effective surgical procedures done to regain function of the hip and knee joints, reduce the degeneration of the joints in patients suffering with osteoarthritis or inflammatory arthritis, improve mobility, and improve the quality of life.⁴ Guidelines recommend rehabilitation as an integral component of faster recovery after THA and TKA,^5–9 but the frequency and intensity of exercise and mode of rehabilitation are not specified.^6,8–10 With effective rehabilitation, pain is reduced, energy levels, sleep, social, and sexual function are improved, oxygen demands during activities are reduced, and general walking ability is regained within three months of surgery,^11–14 although outcomes are dependent on patient-level factors (e.g., sociodemographic characteristics) and environmental factors (e.g., availability of specific therapeutic modalities or expert professionals).¹⁵

Post-THA and TKA surgery rehabilitation may be provided in various settings, including inpatient rehabilitation or skilled nursing facilities, outpatient rehabilitation services, home-based rehabilitation, or telerehabilitation services.^9,10 In 2015, the Centers for Medicare & Medicaid Services (CMS) implemented payment reform which dramatically altered discharge settings for patients with TKA or THA.¹⁶ Early studies on the impact of the CMS payment reform showed decreased length of hospital stays, decreased use of in-patient rehabilitation facilities, and an increased reliance on informal caregivers for home-based care.^17,18 While some studies have shown that patients receiving rehabilitation following an uncomplicated unilateral TKA^19–22 or THA^23–25 in clinics or inpatient settings experience similar outcomes as those with home-based rehabilitation, facility-level factors may influence the outcomes of THA and TKA rehabilitation.^10,15 Because of the dramatic shifts in post-surgical care setting with CMS payment reforms, a synthesis of evidence about the association between patient and facility factors and outcomes from older adults undergoing THA or TKA in the United States is needed.

The aim of this systematic review was twofold. First, we sought to provide an in-depth summary of recent literature on the relationship between patient and rehabilitation outcomes among older adults undergoing rehabilitation in an inpatient setting following a THA or TKA. Second, we aimed to synthesize the recent literature on the relationship between facility factors and rehabilitation outcomes among older adults undergoing a THA or TKA. We focused our systematic review on studies conducted in the United States because the rehabilitation settings have shifted dramatically in recent years in response to CMS payment reform. In doing so, this study addresses a gap in the literature stemming from payment policy reform affecting rehabilitation care setting after joint replacements.

METHODS

Ethics approval

The research was not considered human subjects research because all selected studies were obtained from publicly available sources.

Search Strategy

We identified articles published between January 1, 2000, and December 2021 in three research databases: PubMed, Scopus^®, and Cumulative Index to Nursing and Allied Health Literature (CINAHL). Initially, the investigative team defined three domains of interest related to patients (those who underwent hip or knee arthroplasty), the outcomes of rehabilitation, and the settings (healthcare facility such as nursing home, inpatient rehabilitation, skilled nursing facility). For each domain, the investigative team developed key words. One author (A.K.M.) worked with a research librarian at our institution to refine the search strategy. With their input, the search strategy included use of Medical Subject Headings (MeSH) terms and free text keywords for hip or knee arthroplasty (“Arthroplasty, Replacement, Knee”[Mesh] OR “knee replacement arthroplasty”) OR (“Arthroplasty, Replacement, Hip”[Mesh] OR “hip replacement arthroplasty”), treatment outcome (“Treatment Outcome”[Mesh] OR “treatment outcome” OR “rehabilitation outcome” OR “physical therapy outcome” OR “clinical outcome”), and rehabilitation setting (“Nursing Homes”[Mesh] OR “nursing home” [tiab] OR “Skilled Nursing Facilities”[Mesh] OR “skilled nursing facility” OR “long-term care” OR “extended-care facility” OR “rehabilitation setting”), and time period (2000/01/01 – 2021/12/31) (See Appendix). We checked the list of references from manuscripts identified for more literature to supplement our search strategy results.

Inclusion and Exclusion Criteria

This systematic literature review was designed and conducted using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) as a guide.^26,27 Observational and experimental studies examining one or more outcomes of rehabilitation after joint replacement for patients were eligible for inclusion. These outcomes included functional (functional independence measurement, activities of daily living, joint mobility, range of motion etc.), health status (health-related quality of life, pain etc.), and healthcare utilization (readmissions, costs, length of stay, discharge disposition etc.). Eligible studies mentioned skilled nursing facility (SNF) or inpatient rehabilitation facility (IRF) either as the setting of the study or a key variable. Studies included analyses examining either differences in outcomes based on patient level characteristics (such as pain, physical function, postoperative complications, comorbid conditions, age, sex, race, body weight/body mass index, socioeconomic status, and psychological factors) or information regarding the association between facility level factors (acute inpatient, SNF, and IRF) on outcomes. Studies conducted outside of the United States were excluded as were those that were not experimental or observational (e.g., case reports, review articles, letters, opinions, commentaries, editorials), did not examine rehabilitation outcomes, or did not include patients who underwent hip/knee arthroplasty. Because the United States lacks universal health insurance, offers a fragmented health care system, with variations in health care coverage for rehabilitation, we included only studies of patients receiving care in this context. Two of the authors (A.K.M, S.E.O) screened the titles, abstracts, and full text manuscripts following the literature search. Consensus on inclusion of the studies was reached.

Data Abstraction

Studies were sorted with respect to the information provided: 1) patient factors; or 2) facility factors. To describe the variation in setting, design, and study populations, the following information was extracted from each selected study: study location, study design, data source, setting, sample description (sample size, % men/women, mean age, % race/ethnicity). One author (A.K.M..) reviewed and abstracted information from the selected articles. An additional layer of abstraction and review was conducted by a senior author (K.L.L). Information was modified for consistency, accuracy, clarity, and additional calculations. For studies providing stratified analyses without overall summary statistics, one author (K.L.L) calculated pooled estimates (i.e., %, mean, standard deviations) which were checked and confirmed by another author (S.E.O.). Lastly, the penultimate version of these tables were reviewed by another author (S.E.O.) for accuracy. Discrepancies were resolved during a meeting between two authors (K.L.L., S.E.O.).

To summarize the information gleaned regarding patient or facility factors, the following information was extracted from each study: study objective, main factors/determinants (patient or facility) under study, main outcomes, covariates, and main findings. Two authors (A.K.M., K.L.L.) independently reviewed the studies and abstracted information from the selected articles. Additional information regarding covariates was extracted from each study by a senior author (K.L.L) to provide additional context to the (mostly) observational studies. One senior author coalesced the information from the independent reviews and organized the information consistently across the studies. The penultimate version of these tables was reviewed by another author (S.E.O.) for accuracy with discrepancies resolved during a meeting between two authors (K.L.L., S.E.O.).

Quality Assessment

Two reviewers (S.E.O., A.K.M.) independently assessed the quality of individual studies. For randomized and observational studies, we used the modified Downs and Black criteria.²⁸ The checklist was used to evaluate of the methodological quality of selected studies by examining reporting of the objectives, sample selection, external validity, and internal validity. We indicated a “0” if the article did not include any information addressing the issue, a “+” if the reported information was evaluated as study strength, or a “-” if the information was considered to be a weakness. If discrepancies in the ratings were observed, a senior researcher (K.L.L.) reviewed the study and reconciled the differences.

RESULTS

Search Results

The initial search of PubMed, SCOPUS and CINAHL databases produced 299 records after merging and deduplication. A total of 245 titles were excluded after title and abstract review, and 54 full text articles were then reviewed for eligibility. We obtained 17 full-text articles fulfilling inclusion and exclusion requirements for data extraction and analyses. The figure shows the final eligible studies which includes eight studies that compared associations between patient-level characteristics and rehabilitation outcomes,^29–36 and nine that compared associations between facility-level characteristics and rehabilitation outcomes.^37–45

Figure.

Identification of eligible studies

Characteristics of Studies

Table 1 shows the key characteristics of studies examining the effect of patient characteristics on rehabilitation outcomes. The studies are organized and displayed by study design. Most eligible studies were cohort studies. Two were retrospective cohort studies,^29,36 with one study being a chart review.²⁹ Five were prospective cohort studies.^{30,31,33–35} Three prospective cohort studies used data from the Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS) study.^33–35 One prospective cohort study was a secondary analysis of data collected from an intervention study.³⁰ Only one study was a randomized clinical trial.³²

Table 1:

Characteristics of included studies^* conducted in the United States evaluating patient factors and rehabilitation outcomes for those undergoing total knee or hip replacements.

Author (Year) Location	Study Design / Data Source (Years of data)	Setting # Patients and description # Facilities and description	Sample Description** % Female Mean Age±SD (years) % Race/Ethnicity
Vincent et al. (2010) East coast states	Retrospective cohort study (2002–2006)	5,421 patients, total knee arthroplasty 15 independent inpatient rehabilitation facilities	Female: 68.6% Age: 69.8±10.6 White: 81.3% Black: 5.4% Hispanic: 12.3%
Chu et al. (2016) United States	Retrospective cohort study Chart review (2008–2013)	94 patients, bilateral total knee arthroplasty 1 freestanding inpatient rehabilitation facility	Female: 71.3% Age: 65.6±10.2 Race/ethnicity: Not reported
Folden et al. (2007) Unspecified southeastern state	Prospective cohort study Secondary analysis of study evaluating impact of postsurgical education video (dates of study not specified)	73 patients with inpatient rehabilitation after hip repair surgery 2 hospital rehabilitation units	Female: 66.7% Age: 73.9±8.4 White: 98.4%
Tian et al. (2010) United States	Prospective cohort study Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS) I, II (2006–2007)	318 patients with hip arthroplasty (82 non-elective after fracture, 236 elective) 1 hospital-based and 5 freestanding skilled nursing facilities 6 inpatient rehabilitation facilities	Female: 72.0% Age: 72.4±11.0 White: 74.3%
Siebens et al. (2012) United States	Prospective cohort study Joint Replacement Outcome in Inpatient Rehabilitation Facilities and Nursing Treatment Sites Study (JOINTS) (2006–2007)	224 patients with hip fractures treated with hip arthroplasty 7 skilled nursing facilities 11 inpatient rehabilitation facilities	Female: 77.7% Age: 76.8±11.4 Non-white: 12.9%
Siebens et al. (2016) United States	Prospective cohort study Joint Replacement Outcome in Inpatient Rehabilitation Facilities and Nursing Treatment Sites Study (JOINTS) (2006 – 2007)	226 patients with hip fractures treated with hip arthroplasty 11 inpatient rehabilitation and 7 skilled nursing facilities	Female: Not reported Age: 76.8±11.0 Race/ethnicity: Not reported
Cogan et al. (2021) Eastern and Midwestern United States	Prospective cohort study Secondary analysis (2005–2010)	162 patients with elective hip (37.7%) or knee replacement (62.3%) 4 inpatient rehabilitation and 7 skilled nursing facilities	Female: 75.3% Age: 76.2±6.2 Non-White: 5.6% White: 94.4%
Kim et al. (2021) North Carolina	Randomized clinical trial Pilot study (IRB approval in 2015 and modification 2016)	43 patients scheduled for primary total knee arthroplasty 1 preoperative assessment clinic	Female: 44% Age: 67.2±6.2 Black: 21% White: 74%

^*Presented by design: cross-sectional, retrospective cohort, prospective cohort, and randomized trial.

^**For studies providing mean ages stratified by groups, weighted averages and pooled standard deviations were calculated.

In all but one study,³² the majority of patients were women. The average age of patients in all the studies was greater than 65 years of age. Two studies failed to report the racial/ethnic distributions of the patients.^29,33 In those that did report race/ethnicity, the majority of patients were White. Patients were seen in a variety of settings including pre-operative assessment clinics, hospitals, inpatient rehabilitation facilities, and skilled nursing facilities.

Table 2 shows the key characteristics of studies examining the effect of facility-level characteristics on rehabilitation outcomes for those undergoing hip and knee replacements. The prospective cohort design was used in all studies conducted on facility factors.^{37–39,41–45} Four studies used data from the Joints Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS Study) with the objective of comparing functional outcomes at discharge across eleven inpatient rehabilitation facilities (IRFs), 8 freestanding skilled nursing facilities (SNFs), and 1 hospital based SNF from across the United States.^38,39,43,45 Patients from two studies were discharged home for therapy after TKA.^37,44 In all but one study where the sex and racial/ethnic distribution of the study population was not reported,⁴⁴ most of the patient population were women and White. The average age in all studies was greater than 65 years of age (except one where the average age was not reported⁴⁴

Table 2:

Characteristics of included studies evaluating facility-level factors and rehabilitation outcomes for those undergoing total knee or hip replacements.

Author (Year) Location	Study Design / Data Source (Years of data)	Setting # Patients and description # Facilities and description	Sample Description % Female Mean Age±SD (years) % Race/Ethnicity
Chimenti et al. (2007) New York	Prospective study (2003–2004)	212 patients receiving home physical therapy following total knee replacement surgery	Female: 68.9% Age: 70.0±9.7 White: 92.5% Black: 7.1% Hispanic/Other: 0.4%
DeJong et al. (2009a) United States	Prospective cohort study Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS) I (2006–2007)	2,152 patients with elective hip (34.9%) or knee replacement (65.1%) 1 hospital-based and 8 freestanding skilled nursing facilities 11 inpatient rehabilitation facilities	Female: 70.6% Age: 71.2±10.2 White: 81.4%
DeJong et al. (2009b) United States	Prospective cohort study Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS) I, II (2006–2007)	856 patients with elective hip (34.5%) or knee replacement (65.5%) 1 hospital-based and 5 freestanding skilled nursing facilities 6 inpatient rehabilitation facilities	Female: 72.3% Age: 71.6±9.7 White: 77.1%
Munin et al. (2010) United States	Prospective cohort study Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (2006 to 2007)	218 patients with hip fracture treated with hemiarthroplasty or total hip arthroplasty 6 freestanding skilled nursing facilities 11 inpatient rehabilitation facilities	Female: 78.4% Age: 76.9±11.2 White: 86.7%
Herbold et al. (2011) New York	Prospective cohort study* (2006 to 2008)	102 matched pairs of patients with unilateral hip fracture, total knee replacement, or total hip replacement discharged to 5 skilled nursing facilities or an inpatient rehabilitation facility	Female: 73.0% Age: 74.1±10.5 White: 84.3% Black: 8.3% Latino: 4.9% Asian: 0.1%
Mallinson et al. (2011) Eleven unspecified states in the midwest and northeast	Prospective cohort study (2005 to 2008)	230 patients with hip (36.5%) or knee (63.5%) replacements or revisions 5 skilled nursing facilities 4 inpatient rehabilitation facilities 6 home health agencies	Female: 70.9% Age: 75.6±5.9 White: 95.2% Black: 3.9%
Tian et al. (2012) United States	Prospective cohort study Appears to be Joint Replacement Outcomes in Inpatient Rehabilitation Facilities and Nursing Treatment Sites (JOINTS) (2006–2007)	1,566 patients with elective hip (39.5%) or knee replacement (60.5%) 7 skilled nursing facilities 11 inpatient rehabilitation facilities	Female: 70.9% Age: 71.9±10.3 White: 80.7%
Padgett et al. (2018) New York	Prospective cohort study Total joint replacement registry data (2007 to 2011)	1,213 matched pairs of patients with total knee arthroplasty discharged to home matched to rehabilitation facility 492 matched pairs of patients with total knee arthroplasty discharged to rehabilitation facility or skilled nursing facilities	Female: Not reported Age: Not reported White: Not reported
Fleischman et al. (2019)	Randomized trial (2016 to 2018)	290 patients who received primary total knee arthroplasty	Female: 51.0% Age: 65±Not reported Race/ethnicity: Not reported

^*The manuscript states that this is a quasi-experimental case control study design.The matched patients with patients with joint replacements discharged to inpatient rehabilitation facilities to those discharged to skilled nursing facilities and then compared outcomes. Because they selected on the basis of exposure (care setting) and followed patients for outcomes, we classified this study as a prospective cohort study.

Summary of Findings from Studies of Patient Factors

The main patient factors on rehabilitation outcomes considered by most of the studies were demographic covariates (age, sex, and race) (Table 3).^29–36 Other key patient factors included anemia status,³⁶ patterns of rehabilitation care,³⁵ weight bearing as tolerated or restricted weight-bearing among patients,³⁴ the rate of recoveries trajectory,^30,33 and preoperative aquatic exercises.³² The main outcomes of interest by majority of the studies were the total, motor, and cognitive functional independence measure (FIM)^{29,30,33–36} Two studies were mainly interested in the functional life scale,³¹ and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score.³² Five out of the seven studies that used FIM determined functional outcome of the patients at discharge,^{29,30,33,34,36} while the other two studies determined functional outcome at 3-months post-discharge,³¹ and 8-months post-discharge.³⁵ All the patients in studies that used FIM (total, motor or cognitive) had an improvement in their FIM status at discharge and/or 3–8 months post-discharge. In the study where the patients were grouped by anemia status, results showed that there were no differences across anemia groups or by bilateral status and patients with lower hematocrit had a longer length of stay and incurred a higher total cost than their normal hematocrit counterpart.³⁶ After surgery, all the patients with faster initial recovery during therapy,³³ medium to high trajectories in therapy minutes per length of stay day,³⁰ or who were younger with lower maximum severity, who had better cognition and weight bearing as tolerated during therapy,³⁴ had an overall better prognosis, improved functional outcomes status and likely to be discharged home. Furthermore, patients who underwent preoperative water exercise after total knee arthroplasty had a better functional outcome at discharge in their WOMAC scores.³²

Table 3.

Summary of results of eligible studies of patient factors and rehabilitation outcomes

Author (Year) Study Design	Study Objective	Main Patient Factors	Main Outcome(s)	Covariates	Main findings
Vincent et al. (2010) Retrospective cohort study	Examine the relationship between anemia and rehabilitation outcomes	Anemia: very low hematocrit (<30%), low (30% to women: 0.35, men: 0.36), normal (women: >0.35, men: >0.36) Bilateral: (yes/no)	Functional outcomes at discharge: Functional independence measure (FIM), FIM sub-scores (e.g., walking, stair climbing), FIM efficiency Cost: length of hospital stay, inpatient rehabilitation, pharmacy, therapy costs	Patient factors: Age, sex, joint replacement type (primary or revision)	Functional outcomes: No differences across anemia groups overall or by bilateral status. Costs: Patients with very low hematocrit had longer length of stay and incurred higher total costs than those with normal hematocrit.
Chu et al. (2016) Retrospective cohort study	Describe functional outcomes after bilateral total knee arthroplasty	Sex, age	Functional outcomes at discharge: FIM (relative to admission) FIM efficiency Other outcomes: Readmission to hospital Length of inpatient rehabilitation facility stay	None	Relationship between patient factors (age, sex) and outcomes not shown. FIM improvement: 26.1 ±10.5 (p<0.05) FIM efficiency: 2.33 ±0.84 Re-admitted to hospital: 8.5% Length of stay: 11.7±4.2
Folden et al. (2007) Prospective cohort study	Examine the impact of patient factors on functional recovery after hip repair surgery	Age, sex, Berg Balance Scale, Fatigue Severity Scale, Fall Efficacy Scale, pain,comorbidity, 10 item CES-D, Minimal Mental Status Exam, Functional Life Scale, complications	Functional outcomes 3 months post-discharge: Independent activities of daily living (Functional Life Scale)	All patient factors were included in the model.	32% returned to presurgical functional level by 3 months. Higher balance scores were associated with higher scores on 3-month post-surgical Functional Life (β=0.46, p=<0.001) and higher MMSE scores (β=2.53, p=0.01) were associated with reduced performance.
Tian et al. (2010) Prospective cohort study (Follow-up at 6 months by design, but averaged 8 months given challenges locating and enrolling patients in the follow-up study.)	Examine association between rehabilitation patterns and outcomes for patients with hip arthroplasty	Patterns of rehabilitation care	Primary outcome at ~8 months: Motor FIM Secondary outcome: SF-12 physical component summary (PCS) Additional outcomes: Falls, medical complications, avoidable emergency room visits, rehospitalizations, and quality of life	Demographics (age, sex, race, living status) Elective vs. nonelective Surgery Health/Functional status at admission (pain, motor FIM, Cognitive FIM, days from surgery to admission) Comorbidity/Severity (Comprehensive Severity Index)	90% received rehabilitation from multiple settings, driven by initial care setting, rather than patient factors. Half had home health care after inpatient rehabilitation, 73.1% of non-elective and 83.1% of elective surgeries had home health care after skilled nursing. Patterns of care explained little of the variation in Motor FIM and SF-12 PCS at follow-up. White patients, those with higher Motor FIM, and Cognitive FIM at baseline had higher Motor FIM and SF-12 PCS at follow-up. Those with non-elective surgeries and comorbid conditions had worse Motor FIM and SF-12 PCS scores at follow-up.
Siebens et al. (2012) Prospective cohort study	Examine the association between weight-bearing as tolerated and outcomes of inpatient rehabilitation after hip arthroplasty	Weight-bearing as tolerated or restricted weight-bearing	Primary outcome at discharge: FIM: Total, Motor, Cognitive Home discharge (Yes/No)	Demographics (age, sex, race) Type of surgery Osteoarthritis, osteoporosis Highest pain (scale: 1–10) Complications Comorbidity/Severity (Comprehensive Severity Index)	79.9% were weight-bearing as tolerated No differences were observed between the WBAT and RWB groups in cognitive, motor, and total FIM scores at discharge. The patients who were younger, with lower maximum severity, had better cognition, and were weight-bearing as tolerated had a greater likelihood for home discharge.
Siebens et al. (2016) Prospective cohort study	Examine the association between initial recovery trajectory and functional outcomes at discharge after hip arthroplasty	Initial recovery trajectory (slower, moderate, faster)	Primary outcome at discharge: FIM: Total, Motor, Cognitive Home discharge (Yes/No)	Demographics (age, sex, race) Type of surgery Osteoarthritis, osteoporosis Highest pain (scale: 1–10) Complications Comorbidity/Severity (Comprehensive Severity Index)	From unadjusted models: Total FIM, Motor FIM, and Cognitive FIM at discharge increased with speed of recovery, with the lowest averages in the slower initial recovery group. While all patients in the faster initial recovery group were discharged to home, 82% of the moderate group and 66% of the slower group were discharged to home.
Cogan et al. (2021) Prospective cohort study	Examine the association between therapy minutes per length of stay day, rate of recovery, and functional outcomes after joint replacement	Therapy minutes per length of stay day (low: < 80 minutes, medium: 80 to 130 minutes, high: > 130 minutes per LOS day) Rate of recovery (low, medium, high gain trajectories)	Primary outcome at discharge: FIM Other outcomes: Self-care tasks	Demographics (sex, age, race) Social supports (marital status, living situation) Comorbidity	Therapy minutes per length of stay day were not associated with FIM at discharge. High therapy minutes per length of stay day was associated with self-care at discharge (p=0.03). Participants in medium and high trajectories improved FIM and were able to conduct self-care tasks at discharge.
Kim et al. (2021) Randomized clinical trial	Examine effectiveness of preoperative water exercise in improving outcomes after total knee arthroplasty	Preoperative aquatic exercise Sixty-minute sessions (warm-up (10 minutes), joint range of motion for flexibility and strength (20 minutes), low intensity endurance to improve cardiovascular fitness (20 minutes), and end with a cool down (10 minutes) Three times a week for 4 – 8 weeks	Primary outcome: 30-day National Surgical Quality Improvement Project defined morbidity Secondary outcomes: Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) Mobility Assessment Tool-short form (MAT-sf) Short Physical Performance Battery (SPPB) Geriatric Depression Scale-short form (GDS-sf) Montreal Cognitive Assessment (MoCA) Inflammatory markers (TNF-α, IL-6, hs-CRP)	Demographics (sex, age, race) Body mass index Charlson comorbidity score	Preoperative aquatic exercise was not associated with 30-day National Surgical Quality Improvement Project defined morbidity, and several secondary outcomes (SPPB, MAT-sf, GDS-sf, and all inflammatory markers). Preoperative aquatic exercise was associated with post-surgical improvements in total WOMAC and MoCA scores.

Summary of Findings from Studies of Facility Factors

The summary of results on facility-level factors and rehabilitation outcomes is provided on Table 4. All facility-level studies,^38–45 examined outcomes in the inpatient rehabilitation setting, but one study,³⁷ mainly focused on outcomes in the home health care setting. Four studies used the FIM score to assess functional outcomes (i.e., motor and/or cognitive FIM) at discharge or approximately 6 months post-discharge.^38,39,43,45 Two studies used passive or active knee flexion in the facilities as main outcomes for the patients,^37,40 two studies used functional statuses such as the ability to achieve functional independence such as mobility (toilet transfer, shower transfer, walking, wheelchair) and self-care (eating, grooming, bathing, dressing) as main outcomes at discharge,^41,42 and one study assessed a 2-year functional outcome using WOMAC scores (function, pain and stiffness), Lower Extremity Activity Scale, 12-item Short Form Health Survey (SF-12), as well as 6-months complications from conditions like pulmonary embolus, deep vein thrombosis, myocardial infarction etc.⁴⁴ Most studies showed that patients who were admitted to inpatient rehabilitation facilities had overall better clinical outcomes (e.g., from diseases such as pneumonia, osteoarthritis) and better functional outcomes (mobility, self-care, FIM score), as well as a lower length of stay compared to those who were in skilled nursing facilities (SNFs).^{38,39,41,43–45} However, patients in SNFs incurred a lower cost for services (i.e., had a higher payment efficiency) compared to those in inpatient rehabilitation facilities.^35,41 While comparing inpatient rehabilitation facilities and home health care in the studies, patients discharged to home health care from either acute or subacute settings generally had improvements regardless of setting.³⁷ In another study, patients discharged to inpatient rehabilitation facilities reported more fractures at 6 months compared to those who had home health care.⁴⁴ Further, self-care and mobility was higher (self-care: 0.91 points 95% CI: −0.31 to 2.14; mobility: 1.83 points; 95% CI: −0.42 to 4.08) in those who received home health care compared to those in SNFs. There was no difference observed in motor FIM between inpatient rehabilitation setting and SNFs.⁴⁵

Table 4.

Summary of results of eligible studies of facility factors and rehabilitation outcomes

Author (Year) Study Design	Study Objective	Main Factors	Main Outcome(s)	Covariates	Main findings
Chimenti et al. (2007) Prospective study	Compare episodes of care and functional outcomes of patients sent directly home to those discharged to a subacute rehabilitation setting prior to home health care	Discharged from acute setting to home health care Discharged from subacute setting to home health care	Primary: Passive and active knee flexion at discharge from home health care	Age Presence of caregiver in the home	Regardless of setting (acute or subacute) before initiating home health care, improvements were noted between initial and discharge from home health care. The number of physical therapy visits were similar across setting before initiating home health care.
DeJong et al. (2009a) Prospective cohort study	Compare functional outcomes at discharge across across inpatient rehabilitation facilities, hospital-based skilled nursing facilities, and free-standing skilled nursing facilities	Inpatient rehabilitation facility Hospital-based skilled nursing facility Free-standing skilled nursing facility	Functional status at discharge: Motor FIM	Demographics: Race Baseline status: Motor FIM at admission, cognitive FIM at admission, Comprehensive Severity Index, revision Days from surgery to onset of rehabilitation	After adjustment for covariates: Motor FIM at discharge among patients receiving care in inpatient rehabilitation facilities was higher (knee: 0.901 points p=0.014; hip: 1.639 p=0.005) than those receiving care in skilled nursing facilities. Post-acute setting explained less of the variation in outcomes than covariates.
DeJong et al. (2009b) Prospective cohort study	Compare functional and health status outcomes at 6–8 months of patients with joint replacement across inpatient rehabilitation facilities, hospital-based skilled nursing facilities, and free-standing skilled nursing facilities	Inpatient rehabilitation facility Hospital-based skilled nursing facility Free-standing skilled nursing facility	Functional status at ~6 months: Motor FIM	Demographics: Age, sex, race Clinical factors: Comprehensive Severity Index, admission motor FIM, Revision Days from surgery to onset of rehabilitation Facility factors: Volume of joint replacement patients served	Knee: Relative to patients treated in hospital-based skilled nursing facilities, patients in freestanding skilled nursing facilities had lower functional scores at follow-up (~6 month motor FIM: −2.35, p<0.001). No differences between hospital-based skilled nursing facilities and inpatient rehabilitation facilities were observed. Hip: Relative to patients treated in a hospital-based skilled nursing facility, patients in inpatient rehabilitation facilities had higher functional scores at follow-up (~6 month motor FIM: 1.95, p=0.003). No differences between hospital-based and freestanding skilled nursing facilities were observed. In both patients undergoing hip and knee replacements, post-acute setting explained less of the variation in outcomes than covariates.
Munin et al. (2010) Prospective cohort study	Compare rehabilitation services provided in inpatient rehabilitation versus freestanding skilled nursing facilities for patients with hip replacement following hip fracture	Inpatient rehabilitation facility Skilled nursing facility	Functional outcomes: FIM at discharge (motor, cognitive) Post-acute length of stay Content of physical and occupational therapy: Daily minutes spent performing the most frequent activities during physical and occupational therapy (first 9 days after admission)	Demographics: Age, sex, race, college education, lived alone, lived in nursing home, insurance Surgery: Type (Total, hemi, unknown), weight bearing status at admission Clinical factors: Body mass index, highest pain within 2 days of admission, Comprehensive Severity Index, diabetes, metabolic syndrome, hypertension, ischemic heart disease, depression, functional independence measure at admission (motor, cognitive) Days from surgery to onset of rehabilitation Facility factors: Bed size Volume of joint replacement patients served	Functional outcomes: No differences in motor or cognitive FIM measures at discharge by rehabilitation setting. Length of stay measured in days was different between settings (inpatient rehabilitation: 13.0 days versus skilled nursing facility: 24.7 days, p< 0.01). Content of physical and occupational therapy: Patients in inpatient rehabilitation facilities had fewer minutes per session, but more sessions per day than those in skilled nursing facilities (hours of physical and occupational therapy per day per patient: inpatient rehabilitation: 2.7 versus skilled nursing facility: 1.4, p<0.01).
Herbold et al. (2011) Prospective cohort study	Compare clinical outcomes by care setting (inpatient rehabilitation versus skilled nursing facility) for patients with total knee replacement and total hip replacement	Inpatient rehabilitation facility Skilled nursing facility	Functional status at discharge: Percent of patients who achieved functional independence (ambulation, stairs, lower body dressing, and toilet transfers) at discharge Transfer to acute care Discharge disposition Walker used for ambulation at discharge Home health services following discharge Length of stay Reimbursement for the rehabilitation care episode	Demographics: Age, sex, race/ethnicity Type of surgery (hip fracture, hip replacement, knee replacement) Days to post-acute care Cognitive status at admission Functional status at admission Note: Covariate balance achieved in all but two functional measures at admission (toileting transfer and stairs)	While no differences were observed between care settings for independence in lower extremity dressing and toilet transfer, a greater proportion of inpatient rehabilitation patients were able to ambulate (87.5% versus 74.0%, p=0.02) or take stairs (68.4% versus 34.7%, p<0.001) independently relative to skilled nursing facility patients. No differences were observed in proportion transferred to acute care or discharge disposition. A greater proportion of patients receiving care in skilled nursing facilities used a walker at discharge (67.7% versus 41.7%, p<0.001) and required home care (76.4% versus 33.7%, p<0.001) relative to inpatient rehab facilities. Length of stay was longer for those receiving care in skilled nursing facilities (25.5 versus 10.7 days, p<0.001) and costs lower ($10,001 versus $11,984, p=0.01) relative to inpatient rehab facilities.
Mallinson et al. (2011) Prospective cohort study	Compare clinical outcomes by care setting (inpatient rehabilitation versus skilled nursing facility versus home health agencies) for patients after lower-extremity joint replacement	Inpatient rehabilitation facility Skilled nursing facility Home health agency	Functional status at discharge: Self-care (6-items: eating, grooming, bathing, dressing upper body, dressing lower body, toileting) Mobility (7-items: tub transfer, shower transfer, bed-chair transfer, toilet transfer, walking, wheelchair, climbing stairs) Based on most dependent functional performance in the last 48 hours before discharge converted to interval-level measures.	Demographics: Age, sex Clinical factors: Type of surgery (hip/knee), revision, # of comorbid conditions, bladder incontinence, bowel incontinence Self-care / mobility at admission (based on most dependent functional performance in the first 48 hours after admission) Length of stay	After adjustment for covariates: Self-care and mobility scores at discharge among patients receiving care in inpatient rehabilitation facilities were lower (self-care: −1.90 points, 95% confidence interval: −2.84 to −0.95; mobility: −1.41 points, 95% confidence interval: −2.93 to 0.12) and those receiving home health care higher (self-care: 0.91 points 95% confidence intervals: −0.31 to 2.14; mobility: 1.83 points; 95% confidence intervals: −0.42 to 4.08) than patients receiving care in skilled nursing facilities.
Tian et al. (2012) Prospective cohort study	Compare length of stay and cost of rehabilitation care provided in inpatient rehabilitation facilities and skilled nursing facility for patients with knee or hip replacements	Inpatient rehabilitation versus skilled nursing facility	Functional status at discharge: Δ 13-item motor FIM score discharge- admission Secondary outcomes: Payment efficiency: motor FIM gain per $1000 payment. Length of stay efficiency: motor FIM gain divided by length of stay Multi-input efficiency: stochastic frontier analysis using payment and length of stay to calculation a ratio	Demographics: Age, sex, race, college education, insurance Clinical factors: Body mass index, Comprehensive Severity Index Days from surgery to onset of rehabilitation Facility factors: Bed size Volume of joint replacement patients served Note: Adequate balance achieved with propensity score matching. Analyses stratified by hip/knee replacement.	Primary outcome: No difference in Δ motor FIM between inpatient rehab and skilled nursing facility. Secondary outcomes: Payment efficiency was greater among those in skilled nursing facilities relative to inpatient rehab settings. Length of stay efficiency was greater among those in inpatient rehabilitation facilities relative to skilled nursing facilities. Multi-input efficiency was similar regardless of setting for knee replacement, but skilled nursing facilities appeared more efficient than inpatient rehabilitation for those with hip replacement. Facility factors: Low-volume facilities had the worst efficiencies and medium volume facilities had the highest efficiencies. Starting rehabilitation earlier leads to more efficient treatment.
Padgett et al. (2018) Prospective cohort study	Compare functional and patient reported outcomes after total knee arthroplasty by rehabilitation setting	Two comparisons: Discharged to inpatient rehabilitation versus home Discharged to inpatient rehabilitation versus skilled nursing facility	2-year functional outcomes: Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function, pain, and stiffness; Lower Extremity Activity Scale; 12-item Short Form Health Survey (SF-12) 6-month complications: pulmonary embolus, deep vein thrombosis, infection around joint, major bleeds, pneumonia, stroke, myocardial infarction, dislocation, fracture of or around joint, need for more surgery Early outcomes: Hospital length of stay Need for manipulation	Sociodemographics: Age, sex, living status (alone, with others, retirement or nursing home), insurance Clinical factors: Body mass index, diagnosis (osteoarthritis, post-traumatic arthritis, inflammatory disease), Elixhauser comorbidity, Deyo-Charlson comorbidity index Surgical factors: Year of surgery, type (primary, bilateral) Pre-operative measures: WOMAC function, pain, and stiffness, Lower Extremity Activity Scale, 12-item Short Form Health Survey (SF-12) Note: Adequate balance achieved with propensity score matching.	Inpatient rehab versus home: 2-year functional outcomes: No difference in 2-year outcomes after adjustment for baseline scores. 6-month complications: Patients discharged to inpatient rehab reported more fractures at 6 months (0.7% vs 0.1%; P<0 .038) than those who went home, with no difference in other complications. Early outcomes: Patients discharged to inpatient rehab had shorter length of stay than those discharged home (5.0 vs 5.4; P< 0.001). No difference in manipulation rates. Inpatient rehab versus skilled nursing facility: 2-year outcomes: No differences were observed. 6-month complications: Patients discharged to skilled nursing facilities had higher rates of pneumonia compared to those discharged to inpatient rehabilitation (2.4% vs 0.5%, P = 0.0307), with no difference in other complications. Early outcomes: No differences observed.
Fleischman et al. (2019) Randomized trial (non-inferiority)	Compare recovery among patients with primary, unilateral total knee arthroplasty across models of care	All patients received daily inpatient physical and occupational therapy until hospital discharge. Patients were randomized to: unsupervised home exercise with print materials unsupervised home exercise with web-based materials formal outpatient physical therapy	Primary: Passive knee flexion measured by surgeon unaware of random assignment with goniometer: Δ 4–6 weeks- baseline Δ 6 months- baseline Secondary outcomes: Knee Injury and Osteoarthritis Outcome Score (KOOS) Time for return to activities of daily living: Back to work Back to driving Back to walking without assistive device Back to social activities Time until discontinuation of opiate pain medications	Demographics (age, sex) Body mass index Charlson comorbidity index Length of hospital stay Note: Balance of distributions of covariates across randomized groups appears to have been achieved.	Intention to treat analyses Primary: Unsupervised home exercise program provided noninferior 6-month recovery compared with formal outpatient physical therapy. Secondary outcomes: No differences in 4–6 week or 6-month KOOS by random assignment group. Mean time back to work was 6 weeks (95% CI, 5–7), 6 weeks (95% CI, 5–7), and 7 weeks (95% CI, 6–8) for outpatient, web, and paper groups, respectively (p = 0.218). Similarly, no differences observed for driving, walking, and social activities. Mean time off opiate pain medications was 4 weeks (95% CI, 3–4), 3 weeks (95% CI, 2–4), and 4 weeks (95% CI, 3–4) for the outpatient, web, and paper groups, respectively (p = 0.020).

The review of the quality of reporting is shown in Table 5. While most studies were thorough in their reports of key elements, none discussed issues of power. Few studies reported information on adverse events with rehabilitation. Reporting on the role of confounding was variable, as well as discussions on the role of attrition in these studies.

Table 5.

Quality ratings of selected studies using modified Downs and Black criteria for randomized and cohort studies

Item	Author (Year)
	Patient factors								Facility factors
	Vincent et al. (2010)	Chu et al. (2016)	Folden & Tappen (2007)	Tian et al. (2010)	Siebens et al. (2012)	Siebens et al. (2016)	Cogan et al. (2021)	Kim et al. (2021)	Chimenti et al. (2007)	DeJong et al. (2009a)	DeJong et al. (2009b)	Munin et al. (2010)	Herbold et al. (2011)	Mallinson et al. (2011)	Tian et al. (2012)	Padgett et al. (2018)	Fleischman et al. (2019)
Quality of Reporting
Is the hypothesis/aim/objective of the study clearly described?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Are the main outcomes to be measured clearly described in the Introduction or Methods section?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Are the characteristics of the patients included in the study clearly described?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Are the exposure variables clearly described?	+	0	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Are the distributions of principal confounders in each group of subjects to be compared clearly described?	+	0	0	+	0	+	+	+	+	+	+	+	+	+	+	0	+
Are the main findings of the study clearly described?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Does the study provide estimates of the random variability in the data for the main outcomes?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	0	+
Have all important adverse events that may be a consequence of the exposure variables been reported?	+	+	+	+	0	0	0	+	0	0	0	0	0	0	0	+	+
Have the characteristics of patients lost to follow-up been described?	0	0	0	0	+	0	0	0	0	0	+	0	0	0	0	0	0
Have actual probability values been reported for the main outcomes?	0	0	+	+	+	+	+	+	+	+	+	+	+	0	+	+	+
Internal validity - bias
In trials and cohort studies, do the analyses adjust for different lengths of follow-up of patients?	+	*	*	+	+	+	+	+	0	+	+	+	0	+	+	+	-
Were the statistical tests used to assess the main outcomes appropriate?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Was compliance with the intervention/s reliable?	*	*	*	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Were the main outcome measures used accurate (valid and reliable)?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Internal validity - confounding (selection bias)
Were the patients in different exposure groups recruited from the same population?	+	+	0	0	+	0	+	+	+	+	+	+	+	+	+	+	+
Were study subjects in different exposure groups (trials and cohort studies) recruited over the same period of time?	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+	+
Were study subjects randomized to exposure groups?	*	*	*	*	*	*	*	+	*	*	*	*	*	*	*	*	+
Was there adequate adjustment for confounding in the analyses from which the main findings were drawn?	+	0	+	+	+	+	+	+	0	+	+	+	+	+	+	+	+
Power *
Did the study have sufficient power to detect a clinically important effect where the probability value for a difference being due to chance is less than 5%? Sample sizes have been calculated to detect a difference of x% and y%.	0	0	0	0	0	0	0	+	0	0	0	0	0	0	0	0	+
External validity
Were the subjects asked to participate in the study representative of the entire population from which they were recruited?	+	-	-	0	0	0	-	-	0	-	-	0	-	-	-	+	-

+/− Reported and how performed was a strength (+) or weakness (−)

0 Not reported

^*= Not applicable (N/A)

DISCUSSION

We identified 17 eligible studies (patient factors and facility-level characteristics) published between 2010 and 2021. Our systematic review revealed few studies since the CMS 2015 payment reform. Also, the patients eligible for this systematic review lacked racial/ethnic diversity. Because the measures used in predicting rehabilitation outcomes at discharge were heterogeneous (e.g., various sociodemographic factors, baseline functional status, type of physical / occupational therapy intervention, rate of recovery, and pain management), we were unable to conduct a meta-analysis. Nevertheless, several important findings and research gaps were highlighted by our systematic review.

Our systematic review synthesized evidence in older adults and showed that the receipt of rehabilitation in IRFs relative to SNFs was associated with improvements in clinical and functional outcomes, as well as a shorter length of stay.^{35,38,39,41,43,44} Some studies revealed that a higher intensity of therapy offered in IRFs and a higher physical function of the patients admitted to IRFs, contributed to this improvement in outcomes after joint replacement surgery.^11,14 However, other studies suggested that both high-intensity and low-intensity rehabilitation therapies have long-term benefits in improving strength and function after joint replacement surgery.¹³ Although patients in IRFs had better outcomes compared to SNFs, there was substantial cost savings for patients who received rehabilitation in SNFs compared to those in IRFs.^35,41 This result was similar to previous research which showed that patients discharged to IRFs had higher spending than those discharged to SNFs. This was attributed to IRFs prospectively receiving payments per discharge, with adjustments by area prices, and for patient age, functional and cognitive statuses, and diagnoses.⁴⁶ Payments for SNFs are per diem, and the cost of stay (depending on therapy intensity) is relatively small for every extra day at SNFs compared to IRFs.⁴⁷ Patients who received rehabilitation at home appeared healthier with better functional outcomes at discharge and fewer complications (e.g., fractures) compared to those discharged to IRFs and SNFs.^35,37,44 Despite the differential selection forces leading to healthier patients receiving in-home rehabilitation, post-acute care given in less intensive care settings, such as in home health care setting, may also produce acceptable outcomes. This should be interpreted carefully because most studies only examined outcomes in inpatient settings and at discharge rather than longer term outcomes. More studies evaluating the extent to which rehabilitation in home healthcare settings produces similar outcomes to SNF settings is warranted.

The studies showed more post-surgery discharges to IRFs and SNFs prior to 2015,^{31,35,36,38,39,42,43,45} compared to discharges to the home health care setting during the same period.^37,42 However, the post-surgery discharges to IRFs and SNFs after 2015,^29,30,33 were much lower than prior to 2015, compared to discharges to home health care setting after 2015.⁴⁴ The frequency of discharges to IRFs or SNFs pre-and-post-2015 for THA or TKA were very similar. Although few studies eligible for our review occurred after the major policy shift in 2015, discharges to IRFs or SNFs after 2015 decreased likely owing to changes in CMS policy. Our finding was similar to previous research based on claims data where patients with THA had decreased length of hospital stay, decreased readmissions, and were more likely to be discharged home in 2018 relative to 2011.⁴⁸ Given these shifts in care settings, research that considers factors such as race/ethnicity, advanced age, comorbidities, functional status, home environment, and availability of families/caregivers for patient care are needed to understand the extent to which patients are receiving appropriate care and satisfactory outcomes in home health post-acute care settings.

To control costs for the increasing THA/TKA demand, the CMS implemented payment reform programs which have dramatically changed the landscape of postoperative THA/TKA care by decreasing the use of in-patient rehabilitation facilities while increasing reliance on home-based care.^1–3,17,18 Despite the implementation of these CMS payment reform programs, our findings suggest that patients discharging to or receiving home health care had improvements in function regardless of settings and had higher scores in self-care and mobility, however, few studies after the main policy change were included in our review. With the increasing use of home health care,^49,50 information on THA/TKA postoperative care management among patients and how this impacts formal or informal (i.e., family members) caregivers is still lacking. Given the persistent racial disparities in THA/TKA utilization and outcomes,^51–56 the extent to whether this changing paradigm of post-acute care works equitably for all remains unclear. In addition, the lack of specific information (e.g., mode, duration, frequency) on home-based care programs were also noted. Further research to evaluate the effect of home exercise programs (i.e., functional training versus structured exercise program) and quality of healthcare professionals in home health settings among THA/TKA is warranted.

With the growing aging population and medical complexities of patients undergoing THA or TKA, it is difficult to ascertain whether the less intensive home health care setting will remain effective in managing these patients. This can be predicted if the activities conducted by physical therapists were highlighted in the studies. However, the studies lacked an elaborate description of specific types of activities or exercises by physical therapists, so we could not adequately determine if patients at home were recovering due to activities of daily living or rehab from physical therapy or occupational therapy.

Strengths and Limitations

Every phase of the conduct of this systematic review underwent a dual review process to reduce the risk of bias and ensure quality control. We worked with a research librarian to develop the search strategy. Three separate databases were searched to identify relevant articles. We included the most recent articles available through 2021. The quality ratings, risk of biases, and validity of selected studies was determined using the modified Downs and Black criteria for both randomized and cohort studies (general and significant findings).²⁸ Despite these strengths, a limitation of our systematic review is the relatively scarce number of studies in the post-2015 period and a general lack of details in the activities carried out by physical therapists or occupational therapists.

CONCLUSION

For older adults, rehabilitation at home and the SNFs may be less costly, with similar outcomes as IRFs. However, future studies to examine the trajectory of change influenced by the 2015 payment reform are necessary given the paucity of evidence for how these changes may have affected patients differently (e.g., racial/ethnic minorities, medically complicated patients). There should be more studies focused on home health care, specific activities performed at home, potential impact on the family/caregivers, and the roles played by aging and comorbidities in the home health care setting after THA or TKA