Impact of Hospital-Based Rehabilitation Services on Discharge to the Community by Value-Based Payment Programs After Joint Replacement Surgery

Abstract

Objective

The purpose of this study was to examine the impact of hospital-based rehabilitation services on community discharge rates after hip and knee replacement surgery according to hospital participation in value-based care models: bundled payments for care improvement (BPCI) and comprehensive care for joint replacement (CJR). The secondary objective was to determine whether community discharge rates after hip and knee replacement surgery differed by participation in these models.

Methods

A secondary analysis of Medicare fee-for-service claims was conducted for beneficiaries 65 years of age or older who underwent hip and knee replacement surgery from 2016 to 2017. Independent variables were hospital participation in value-based programs categorized as: (1) BPCI, (2) CJR, and (3) non-BPCI/CJR; and total minutes per day of hospital-based rehabilitation services categorized into tertiles. The primary outcome variable was discharged to the community versus discharged to institutional post-acute care settings. The association between rehabilitation amount and community discharge among BPCI, CJR, and non-BPCI/CJR hospitals was adjusted for patient-level clinical and hospital characteristics.

Results

Participation in BPCI or CJR was not associated with community discharge. This analysis found a dose–response relationship between the amount of rehabilitation services and odds of community discharge. Among those who received a hip replacement, this relationship was most pronounced in the BPCI group; compared with the low rehabilitation category, the medium category had odds ratio (OR) = 1.28 (95% CI = 1.17 to 1.41), and the high category had OR = 1.90 (95% CI = 1.71 to 2.11). For those who received a knee replacement, there was a dose–response relationship in the CJR group only; compared with the low rehabilitation category, the medium category had OR = 1.21 (95% CI = 1.15 to 1.28), and the high category had OR = 1.56 (95% CI = 1.46 to 1.66).

Conclusion

Regardless of hospital participation in BPCI or CJR models, higher amounts of rehabilitation services delivered during acute hospitalization is associated with a higher likelihood of discharge to community following hip and knee replacement surgery.

Impact

In the era of value-based care, frontloading of rehabilitation care is vital for improving patient-centered health outcomes in acute phases of lower extremity joint replacement.

Introduction

Hip and knee joint replacement accounted for over $7 billion in Medicare spending (approximately 1% of overall Medicare spending) in 2017, with approximately one-half million procedures performed.^1,2 Given the high volume and costs associated with joint replacement procedures, the Centers for Medicare and Medicaid Services (CMS) initiated value-based payment programs including the bundled payments for care improvement (BPCI) and the comprehensive care for joint replacement (CJR). In April 2013, CMS initiated the voluntary BPCI episode-based payment models. For the 4 BPCI models, Model 2 was the most common, and this included a bundle payment to the hospital for all services from acute care to post-acute care (PAC) for 90 days.³ CMS extended the voluntary BPCI into a mandatory CJR payment program starting April 1, 2016, until December 31, 2020. Both BPCI and CJR incentivize providers to coordinate care throughout the episode of care (acute care to post acute to community) to reduce health care expenditures and improve patient outcomes. The goals of these value-based programs were to improve care coordination and transitions of care and eventually reduce overall costs of care, including those associated with institutional-based PAC following joint replacement surgery. The primary driver for achieving such efficiency of care was based on provision of (prospective or retrospective) lump-sum payment for an entire episode of care, from surgery to post acute to outpatient and home-based care. One of the ways that costs could be contained was by reducing utilization of PAC in institutional settings (inpatient rehabilitation facilities [IRF] and skilled nursing facilities [SNF]). Thus, it is likely that individuals treated in hospitals participating in BPCI or CJR have a greater likelihood of discharging to the community. However, little research has investigated this pathway. Understanding this from a rehabilitation care perspective could better inform such payment reforms in the future.

Studying outcomes for individuals receiving care from the providers in these bundled payment models is challenging because provider participation is not random and is prone to self-selection bias. Thus, the evidence on hospital characteristics by BPCI and CJR participation and its impact on patient outcome is mixed.^4–8 Hospitals participating in BPCI reported to have higher Medicare patient volume, higher number of beds, and were more likely to be a teaching hospital compared with CJR hospitals in 1 study.⁶ However, another study reported no significant differences in hospital joint replacement volume between BPCI participating and non-BPCI hospitals.⁹ A recent study also demonstrated that BPCI hospitals lowered total Medicare spending per joint replacement episode, primarily by reducing PAC without significant changes in 30-day and 90-day unplanned readmissions, emergency department visits, and mortality compared with non-BPCI hospitals.¹⁰ Also, CJR payment models resulted in 3% reduction in (statistically significant) discharge to SNF and IRF compared with non-CJR hospitals.¹¹

In the era of value-based programs, hospitals may discharge patients (directly) to the community with home health care versus SNF or IRF. One modifiable factor potentially associated with community discharge rates in value-based care models is increasing the provision of rehabilitation services during acute hospitalization. Evidence suggests that intensive hospital-based rehabilitation services are associated with functional independence for individuals after joint replacement^12–15 and may allow safe discharge directly to the community. Both BPCI and CJR payment programs have been shown to affect the utilization of PAC services.^16,17 However, their impact on the utilization of hospital-based rehabilitation services and how this affects the likelihood of discharge to the community versus institutional-based PAC settings remains largely unknown.

The purpose of this study was to analyze the impact of hospital-based rehabilitation services on community discharge after acute hospitalization by hospital participation status in BPCI and CJR following hip and knee joint replacement. We hypothesized that hospitals participating in the BPCI or CJR value-based care models would have higher rates of community discharge after joint replacement, and the greater use of rehabilitation services would explain the magnitude of these differences.

Methods

Data

The study was a retrospective cohort study and secondary analysis of the 100% Medicare inpatient claims data in standard analytical file format for the years 2016 to 2017. Inpatient claims were linked to PAC assessment files (which included the Inpatient Rehabilitation Facility–Patient Assessment Instrument for IRF, Minimum Data Set 3.0 for SNF, and Outcome and Assessment Information Set for home health) to create the community discharge variable. The Provider of Services file was used to retrieve hospital characteristics, and publicly available files from the CMS website were used to identify hospitals participating in BPCI Model-2 and CJR.^1,3,18 A data use agreement was established with the CMS, and institutional review board approval was obtained prior to conducting this study.

Study Population

The study cohort included Medicare fee-for-service claims for individuals over 65 years of age who underwent total knee replacement (TKR), total hip replacement (THR), and partial hip replacement from January 1, 2016, to November 30, 2017. Medicare severity diagnosis-related groups 469 and 470 were used to identify patients with knee and hip replacement. Further, international classification of diseases (ICD)-10 procedure codes were used to identify TKR and THR (total and partial hip replacements; presented in (Suppl. Tab. 4). Individuals were excluded if they (1) died during the index hospitalization, (2) left against medical advice, (3) had a missing length of stay during acute hospitalization, (4) were transferred to another acute care hospital, or (5) were enrolled in Medicare Advantage plans during the study time frame.

Exposure Measures

Hospital Participation in Value-Based Programs

Hospitals were categorized into 3 groups: (1) hospitals participating in the voluntary BPCI Model 2 program (BPCI); (2) hospitals participating in the mandatory CJR program (CJR); and (3) hospitals participating in neither the BPCI model 2 nor CJR programs (non-BPCI/CJR). Model 2 of BPCI includes a retrospective payment for an episode of care, which comprises acute care hospital plus the PAC and all related services up to 90 days post-hospital discharge.³

Hospital-Based Rehabilitation Services

The amount of hospital-based rehabilitation services, including physical therapist and occupational therapist (OT) services, was the primary independent variable of interest. Using methodology developed from the previous research,^19,20 revenue center codes were used to identify evaluation and treatment for physical therapist and OT services during acute hospital stays. We assigned 15 minutes for each treatment unit and 30 minutes for each evaluation unit. Physical therapist and OT evaluation and treatment minutes were summed to calculate the amount of hospital-based rehabilitation services. To adjust for the confounding effect of a patient receiving more therapy with a longer length of hospital stay, total therapy was divided by hospital length of stay to compute rehabilitation services per day. Hospital-based rehabilitation services were then categorized into tertiles of low (<40 min/d), medium (40–65 min/d), and high (>65 min/d) for patients with hip replacement, and low (<45 min/d), medium (45–68 min/d), and high (>68 min/d) for patients with knee replacement.

Outcome Measures

The primary outcome of interest was discharge to community after acute hospitalization. We defined community discharge as a discharge directly to home with or without home health care after their acute hospital stay. Inpatient hospital claims were linked with post-acute assessment files (Inpatient Rehabilitation Facility–Patient Assessment Instrument, Minimum Data Set 3.0, and Outcome and Assessment Information Set) and home health care claims to determine community discharge. The community discharge category included individuals discharged to private home/apartment, board/care, assisted living, group homes, and transitional living settings. Discharge to an institutional post-acute setting such as IRF, SNF, intermediate care, or long-term care hospitals was not considered as a community discharge.

Covariates

To control for case-mix differences and hospital-level confounders, we included patient-level sociodemographic information (ie, age [continuous], sex, race/ethnicity [non-Hispanic White, Black, Hispanic, Others], hospital length of stay [continuous], and dual eligibility into Medicare and Medicaid [yes/no]) and co-morbidity (ie, Elixhauser comorbidity score as continuous score) as well as the cause of the joint replacement (ie, osteoarthritis or fracture/other reasons) and type of replacement (total hip and partial hip replacement).²¹ We included hospital characteristics such as profit status (yes/no), teaching status vs non-teaching, safety-net status (yes/no), number of hospital beds (categorical), and volume of joint replacement surgeries (low, medium, high categories).

Statistical Analysis

Data and analyses were stratified for THR and TKR (Tab. 1). Chi-square test for categorical variables and t test for continuous variables was used to test the association of covariates by hospital participation status (BPCI, CJR, non-BPCI/CJR). We reported continuous variables with mean (SD) and categorical variables with frequencies. In our analysis, acknowledging the nested structure of the data, where patients (level 1) were nested in hospitals (level 2), we constructed generalized linear mixed models. These mixed models determine differences in amount of rehabilitation services per day across hospitals participating in BPCI, CJR, or non-BPCI/CJR groups and adjusted for within-cluster correlation to examine the association between the amount of hospital-based rehabilitation services with odds of community discharge. Models were controlled for age, sex, race, dual eligibility, Elixhauser comorbidity index, hospital beds, profit status of the hospital, teaching status of the hospital, safety-net status of the hospital, joint replacement volume treated at the hospital, and hospital participation in bundle payment. In addition, we adjusted for the interaction effect between hospital participation (BPCI, CJR, nonparticipant) and amount of rehabilitation (low, medium, high). Adjusted odds ratios with a 95% CI were reported for the patients with THR and TKR separately. All statistical analyses were performed using SAS 9.4 (SAS Inc, Cary, NC, USA) and STATA 15 (StataCorp LLC, College Station, TX, USA).

Table 1.

Patient and Hospital Characteristics Following Hip and Knee Replacement by Hospital Participation in Different Value-Based Programs^a

	Hip Replacement N = 264,755				Knee Replacement N = 376,146
Characteristics	BPCI	CJR	Non-BPCI/CJR	P	BPCI	CJR	Non-BPCI/CJR	P
No. (%)	21,995(8.3)	55,135(20.8)	187,625(70.9)	—	31,084(8.3)	74,200(19.7)	270,862(72.0)	—
Patient-level characteristics, %
Age,mean (SD), y	76.6(7.9)	76.7(7.9)	76.7(7.9)	.0696	73.8(5.8)	73.9(5.9)	73.8(5.9)	.0008
Sex assigned at birth, %
Male	35.6	36.0	35.5	.2873	36.5	36.0	37.3	<.0001
Female	64.4	64.0	64.5		63.5	64.0	62.7
Race, %
Non-HispanicWhite	89.9	90.2	91.8	<.0001	86.2	86.5	88.9	<.0001
Black	4.8	3.3	3.5		6.0	4.0	4.3
Hispanic	2.3	2.8	1.7		3.8	4.5	3.0
Others	3.0	3.7	3.0		4.0	5.0	3.8
Dual coverage, %
Yes	6.7	7.0	7.2	.0080	4.6	5.6	5.0	<.0001
No	93.3	93.0	92.8		95.4	94.4	95.0
Length of stay, mean (SD)	4.5(3.3)	4.4(2.8)	4.5(2.8)	.0004	4.0(3.2)	4.1(2.9)	4.1(3.0)	.0004
Elixhauser comorbidity index,Mean (SD)	2.4(1.7)	2.3(1.7)	2.4(1.7)	<.0001	2.3(1.5)	2.2(1.5)	2.2(1.5)	<.0001
Cause of joint replacement
Osteoarthritis	71.4	72.1	69.6	<.0001	98.6	98.3	98.4	.0001
Fracture/other	28.6	27.9	30.4		1.4	1.7	1.6
Surgical procedure for hip replacement only
Partial hipreplacement	21.9	21.6	23.5	<.0001	—	—	—
Total hipreplacement	67.1	67.5	65.6		—	—	—
Other surgicalprocedures	11.0	10.9	10.9		—	—	—
Discharge destinations
IRF	6.5	6.4	7.3	<.0001	2.1	2.5	2.7	<.0001
SNF	32.9	31.5	31.1	<.0001	21.9	22.1	19.4	<.0001
Home health	39.0	38.9	30.5	<.0001	51.3	47.8	39.8	<.0001
Home	21.6	23.3	31.1	<.0001	24.8	27.6	38.1	<.0001
Community discharge
Yes	60.6	62.2	61.6	.0004	76.1	75.5	77.8	<.0001
No	39.4	37.8	38.4		23.9	24.5	22.2
Hospital-based rehabilitation, mean (SD)
PT/d	39.0(22.3)	40.3(21.6)	38.1 (19.7)	<.0001	47.1(22.8)	48.0(22.3)	45.0(20.0)	<.0001
OT/d	16.9(14.9)	16.4 (15.5)	16.3 (14.2)	<.0001	13.7(14.9)	14.2(15.7)	13.2(14.6)	<.0001
Total therapy/d	55.9(32.6)	56.7 (32.0)	54.4 (28.1)	<.0001	60.9(31.3)	62.2(31.0)	58.2(27.4)	<.0001
Hospital characteristics
Profit status
Yes	16.9	18.4	14.5	<.0001	17.3	22.3	17.1	<.0001
No	67.5	71.7	74.9		65.7	67.7	72.8
Other/ unknown	15.6	9.9	10.6		17.0	10.0	10.1
Teaching status
Nonteaching	44.3	47.1	55.0	<.0001	45.2	48.0	59.1	<.0001
Teaching	56.0	52.9	45.0		54.8	52.0	40.9
Safety net, %
Yes	21.0	16.1	13.4	<.0001	19.6	16.7	12.4	<.0001
No	79.0	83.9	86.6		80.4	83.3	87.6
Hospital beds
1 (≤123)	8.2	13.8	22.9	<.0001	7.9	19.4	28.5	<.0001
2 (124–228)	21.4	19.5	24.4		21.3	18.6	23.7
3 (229–370)	27.6	28.3	23.4		28.5	26.3	21.8
4 (≥371)	42.9	38.3	29.3		42.4	35.7	26.0
LEJR volume
Low	9.7	13.0	10.1	<.0001	7.9	9.6	6.7	<.0001
Medium	38.1	33.2	36.6		33.4	31.0	34.6
High	52.2	53.8	53.3		58.8	59.4	58.7

^a BPCI = bundled payments for care improvement; CJR = comprehensive care for joint replacement; IRF = inpatient rehabilitation facility; LEJR = lower extremity joint replacement; OT = occupational therapy; PT = physical therapy; SNF = skilled nursing facility.

Sensitivity Analyses

Additional analyses were conducted to present patient and hospital characteristics by rehabilitation dosing, that is, low, medium, and high groups. In addition, multilevel multivariate logistic regression models were computed separately for hip replacement and knee replacement to examine the association between hospital-based physical therapist and OT services per day with the likelihood of community discharge. All regression models were controlled for age, sex, race, dual eligibility, Elixhauser comorbidity index, hospital beds, profit status of the hospital, teaching status of the hospital, safety-net status of the hospital, joint replacement volume treated at the hospital, and hospital participation in bundle payment. We examined physical therapist and OT minutes in separate models for each stratified group of THR and TKR. Models for physical therapist minutes included main effect of physical therapist services received; main effect of hospital participation in BPCI, CJR, or nonparticipant; and interaction effect between hospital participation (BPCI, CJR, nonparticipant) and amount of physical therapy (low, medium, high). Models for OT minutes included main effect of OT services received; main effect of hospital participation in BPCI, CJR, or nonparticipant; and interaction effect between hospital participation (BPCI, CJR, nonparticipant) and amount of occupational therapy (low, medium, high).

Role of the Funding Source

The funder played no role in the design, conduct, or reporting of this study.

Results

The final sample included 640,901 patients who received either a THR or TKR. Table 1 compared the demographic and clinical characteristics of patients with THR and TKR stratified by hospital participation in BPCI, CJR, and non-BPCI/CJR. Only 8.3% and 20.8% of the patients with THR received care in BPCI and CJR hospitals, respectively. A higher proportion of patients in CJR hospitals with THR were discharged to community, followed by non-BPCI and BPCI hospitals. Approximately 8.3% and 19.7% of the patient with TKR received care in BPCI and CJR hospitals, respectively. Overall, patients after TKR received more rehabilitation services per day compared with patients after THR. A higher proportion of patients with TKR were discharged to the community in non-BPCI/CJR hospitals, followed by BPCI and CJR hospitals.

Higher amounts of daily rehabilitation services were associated with an increased likelihood of community discharge, with higher odds in those with high amounts of rehabilitation services (>65 min/d for individuals with hip replacement and >68 min/d for individuals with knee replacement) compared with medium (40–65 min/d and 45–68 min/d for hip and knee replacement, respectively) and low (<40 min/d and 45 min/d for hip and knee replacement, respectively). There was no significant association between patients receiving rehabilitation services in BPCI and CJR hospitals and community discharge after hip and knee replacement compared with non-BPCI/CJR hospitals (Tab. 2). Patients from all hospitals participating in BPCI/CJR models had similar likelihood of discharge to the community compared with non-BPCI/CJR.

Table 2.

OR and 95% CIs for Community Discharge and Amount of Rehabilitation Services and Hospital Participation in Value-Based Programs^a,b

Variables	Hip Replacement N = 259,595 OR for Community Discharge (95% CI)	Knee Replacement N = 367,856 OR for Community Discharge (95% CI)
Hospital participation in value-based programs
Non-BPCI/CJR	Reference	Reference
BPCI	0.93 (0.83 to 1.04)	0.95 (0.82 to 1.10)
CJR	0.98 (0.90 to 1.06)	0.91 (0.82 to 1.00)
Amount of rehabilitation services per day
Low	Reference	Reference
Medium	1.24 (1.20 to 1.29)	1.06 (1.03 to 1.10)
High	1.76 (1.68 to 1.84)	1.35 (1.29 to 1.41)

^a BPCI = bundled payments for care improvement; CJR = comprehensive care for joint replacement; OR = odds ratio.

^b Results are adjusted for patient sociodemographic characteristics, comorbidity, cause of joint replacement, and hospital characteristics. Bolded values represent statistically significant association.

Irrespective of hospital participation in BPCI/CJR models or non-BPCI/CJR, we observed a monotonic increase in likelihood of community discharge with amounts of rehabilitation services per day for patients with hip replacement (Tab. 3). Compared with those with a low amount of rehabilitation services per day, those with a medium amount of rehabilitation services per day had a 20% to 28% increased odds of community discharge, and those with a high amount of rehabilitation services per day had a 69% to 90% increased odds of discharge to the community. The pattern was not consistent for patients with TKR (Tab. 3). Compared with a low amount of rehabilitation services per day, individuals receiving a medium amount of rehabilitation services per day was significantly associated with increased odds of community discharge in CJR model and non-BPCI/CJR model. Compared with those with a low amount of rehabilitation services per day, those receiving a high amount of rehabilitation services per day had a 22% to 56% increased odds of community discharge.

Table 3.

OR and 95% CIs for Community Discharge With Amount of Rehabilitation Services by Hospital Participation in Value-Based Programs^a,b

	Hip Replacement N = 259,595 OR (95% CI)			Knee Replacement N = 367,856 OR (95% CI)
	BPCI	CJR	Non-BPCI/CJR	BPCI	CJR	Non-BPCI/CJR
Amount of rehabilitation services per day
Low	Reference	Reference	Reference	Reference	Reference	Reference
Medium	1.28 (1.17 to 1.41)	1.25(1.18 to 1.33)	1.20(1.16 to 1.24)	0.97(0.89 to 1.06)	1.21(1.15 to 1.28)	1.03(1.00 to 1.06)
High	1.90(1.71 to 2.11)	1.69(1.57 to 1.81)	1.69(1.63 to 1.76)	1.29(1.16 to 1.43)	1.56(1.46 to 1.66)	1.22(1.18 to 1.27)

^a BPCI = bundled payments for care improvement; CJR = comprehensive care for joint replacement; OR = odds ratio.

^b Results are adjusted for patient sociodemographic characteristics, comorbidity, cause of joint replacement, and hospital characteristics. Bolded values represent statistically significant association.

Sensitivity Analyses

The results of the additional analysis are reported in the supplementary materials, including patient and hospital characteristics by rehabilitation dosing in Supplementary Table 1. In addition, adjusted odds ratio with a 95% CI for patients with THR and TKR is presented in Supplementary Table 3, demonstrating an association between hospital-based physical therapist and OT services and that of community discharge. Overall, increased amounts of hospital-based physical therapist and OT services were significantly associated with increased odds of discharge to the community in patients with hip and knee replacement, irrespective of an individual hospital’s participation in BPCI/CJR model.

Discussion

The findings suggest that a hospital’s participation in value-based payment programs was not significantly associated with community discharge for patients after THR and TKR. However, we found a significant positive association between a higher amount of rehabilitation services during acute hospitalization with a greater likelihood of community discharge in patients after hip and knee joint replacement. To our knowledge, this is the first study that compared hospital-based rehabilitation services and discharges to the community across hospitals participating in the BPCI and CJR programs. Our findings provide further evidence to support clinical guidelines that have asked for early initiation of rehabilitation services to maximize patient outcomes and have policy implications on promoting rehabilitation services for patients in the hospital after joint replacement to maximize the likelihood of community discharge immediately after an acute hospital stay.

Community discharge is an important performance indicator for value-based payment programs and is consistent with the national quality strategy priorities for care coordination and cost effectiveness.²² Both the BPCI and CJR programs changed PAC delivery after joint replacement in the United States, primarily by shifting discharge disposition settings after acute hospitalization.^16,17 To reduce the cost of institutional-based PAC, both the BPCI and CJR programs encourage discharge to the community with the home health or outpatient care, which are often less costly than institution-based PAC.^23,24 The current study showed no association with participation in value-based payment programs and odds of community discharge. A similar proportion of individuals after THR were discharged to the community for each value-based payment program (60.6%, 62.2%, and 61.6% for BPCI, CJR, and non-BPCI/CJR, respectively), but with TKR, a slightly higher proportion of individuals in the non-BPCI/CJR were discharged to community (77.8%) compared with the value-based payment programs (76.1% for BPCI and 75.5% for CJR). Similarly, the likelihood of community discharge by value-based payment program shows that the amount of rehabilitation services significantly impacts discharge destinations after adjusting for hospitals’ participation in BPCI/CJR.

Rehabilitation professionals have needed expertise in maximizing functional recovery and planning for a successful transition from the hospital to the community. This is critical because discharge to the community directly from the hospital may reduce the risk of complications, emergency visits, and readmissions.^17,25,26 Prior studies conducted in single hospital settings have shown a dose–response relationship between post-operative rehabilitation and improvement in function after joint replacement.^12,27 Hospital rehabilitation professionals often help coordinate acquisition of necessary durable medical equipment, provide caregiver education, and recommend safe discharge destination.²⁸ These roles are likely to become more important with shortening lengths of hospital stay after joint replacement.^29,30 Our findings support models of care that include a higher amount of rehabilitation services to help facilitate community discharge in individuals after THR and TKR.

Both BPCI and CJR payment programs incentivize hospitals to reduce the hospital length of stay.³¹ A shorter hospital length of stay limits the time that a physical therapist and OT can work with an individual to address goals and optimally prepare for discharge. Thus, timely and appropriate rehabilitation services and including in-depth assessment by rehabilitation clinicians in the discharge planning process is crucial to have better care coordination with post-acute settings and improve the quality of continuum care.³²

With the shorter hospital lengths of stay and shifts to reduce PAC use, findings from this study also provided insights to improve early initiation of rehabilitation services and the discharge planning process from the time of surgery and the immediate post-operative period. Thus, our study adds to prior work showing the benefit of post-operative hospital-based rehabilitation services in improving patient outcomes following THR and TKR during an acute phase.

Limitations

Our results should be interpreted considering several limitations. First, we used Medicare claims data, and the limitations of administrative data are well reported.³³ Specific to this study, claims data lack information on the clinical severity (eg, pain measures), functional status during acute hospitalization, and timing for rehabilitation services initiation. Additionally, these results are limited in the generalizability to those who are not Medicare beneficiaries on Fee-for-Service (FFS). Secondly, information was not available for those individuals who were unable to participate in rehabilitation due to complications or those who declined rehabilitation services. Finally, the entirety of clinical decision-making for discharge planning (eg, rehabilitation potential, availability of PAC) is not included in administrative data but is certainly a ripe area for future investigation. These limitations in clinical care and processes may have impacted discharge location rather than having fewer rehabilitation minutes.

With the introduction of value-based payment programs, there is a critical need to evaluate the impact of these new payment programs on service provision and outcomes. Although the focus of most of the research has been on cost savings,^34–37 examination of the impact on process of care and patient-centered outcomes is equally important. Our study addresses the impact of hospital participation (or not) in one of the value-based payment programs and how that affected discharge to the community for individuals after hip and knee joint replacement. We also explored the provision of rehabilitation services during acute hospitalization for these individuals. Future work should examine other patient-centered outcomes such as functional recovery and 30-day hospital readmission as well as cost-effectiveness of rehabilitation care and long-term outcomes. The rehabilitation research community must take advantage of the substitution of care occurring due to these value-based payment models to examine the “opportunity cost” of switching to community settings over institutional settings, taking into consideration some of the long-term outcomes.

Author Contributions

Concept/idea/research design: A. Kumar, I. Roy, M. Warren, S.D. Shaibi, A.M. Karmarkar

Writing: A. Kumar, I. Roy, M. Warren, S.D. Shaibi, M. Fabricant, J.R. Falvey, A. Vashist, A.M. Karmarkar

Data collection: A. Kumar, I. Roy, M. Fabricant

Data analysis: A. Kumar, I. Roy, M. Warren, M. Fabricant, A. Vashist, A.M. Karmarkar

Project management: A. Kumar, I. Roy, A.M. Karmarkar

Fund procurement: A. Kumar, A.M. Karmarkar

Providing facilities/equipment: A. Kumar

Consultation (including review of manuscript before submitting): A. Kumar, M. Warren, S.D. Shaibi, J.R. Falvey, A. Vashist

Funding

This study was supported by National Institute of Health grants K01-HD086290, R03AG060345, U54MD012388, K76AG074926, and P30AG028747.

Ethics Approval

A data use agreement was established with CMS, and institutional review board approval was obtained prior to conducting this study.

Disclosures and Presentations

The authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest and reported no conflicts of interest.

A portion of this study was presented as an abstract at the American Physical Therapy Association’s Combined Sections Meeting; February 14, 2020; Denver, CO, USA.