Sage Journals: Discover world-class research

Abstract

Introduction:

To examine racial variations in access to postacute care (PAC) and rehabilitation (Rehab) services following elective total knee arthroplasty and whether where patients go after surgery for PAC/Rehab is associated with 30-day readmission to acute care facility.

Materials and Methods:

Sample consisted of 129 522 patients discharged from 169 hospitals in the State of Pennsylvania between fiscal years 2008 and 2012. We used multinomial regression models to assess the relationship between patient race and discharge destination after surgery, for patients aged 18 to 64 years and for those aged 65 and older. We used multivariable (MV) regression and propensity score (PS) approaches to examine the relationship between patient discharge destination after surgery for PAC/Rehab and 30-day readmission, controlling for key individual- and facility-level factors.

Results:

Lower proportions of younger patients compared to those older than 65 were discharged to inpatient rehabilitation facilities (IRFs; 5.8% vs 12.6%, respectively) and skilled nursing facilities (SNFs; 15.2% vs 32.7%, respectively) compared to home-based Rehab (self-care; 23.3% vs 14.2%, respectively). Compared to whites, African American patients had significantly higher odds of discharge to IRF (age < 65, odds ratio = 2.04; age ≥ 65, odds ratio = 1.64) and to SNF (age < 65, odds ratio = 2.86; age ≥ 65, odds ratio = 2.19) and discharge to home care in patients younger than 65 years (odds ratio = 1.31). The odds of 30-day readmission among patients discharged to an IRF (MV odds ratio = 7.76; PS odds ratio = 8.34) and SNF (MV odds ratio = 2.01; PS odds ratio = 1.83) were significantly higher in comparison to patients discharged home with self-care.

Conclusion:

African American patients with knee replacement are more likely to be discharged to inpatient Rehab settings following surgery. Inpatient Rehab is significantly associated with 30-day readmission to acute care facility.

Keywords

racial disparity access hospital readmissions post-acute care total knee arthroplasty patient-centered outcomes

Introduction

Total knee arthroplasty (TKA) is widely regarded as a safe and highly successful treatment option for end-stage osteoarthritis (OA) of the knee. As the US population ages, use of TKA is expected to accelerate.¹ In 2015, the number of annual TKAs performed in the United States is projected to exceed 1 million procedures.² This figure marks a 2-fold increase in TKA utilization over the past decade and an increase in annual hospital charges from US$7.42 billion in 2005 to US$40.8 billion in 2015.^2,3 As Medicare is the single largest payer for these procedures, the Centers for Medicare and Medicaid Services (CMS) has implemented a series of cost-containment changes to its reimbursement policies. These include the Inpatient Prospective Payment System (IPPS) and the Hospital Readmissions Reductions Program (HRRP). Beginning in fiscal year (FY) 2015, CMS will impose reimbursement penalties through the HRPP—up to 3% of total Medicare billings—on IPPS hospitals found to have “worse than expected” TKA readmissions.

According to the latest data from the Healthcare Cost and Utilization Project, approximately 120 000 (4.8%) TKA procedures resulted in early readmission in 2009 to 2012.⁴ Prior research suggests 30-day readmission rates for TKA have steadily increased since the early 1990s,⁵ with at least 1 study reporting the most dramatic rise among minority, and in particular African American (AA) patients: from 6% higher than white patients in 1991 to 24% in 2008.⁶ Often, these readmissions signal inadequate discharge planning or poor care continuity between hospital clinicians and postacute care (PAC) or rehabilitation (Rehab) service providers,⁷ yet little is known about racial variations in the patterning of PAC/Rehab care after surgery. Furthermore, PAC/Rehab encompasses a wide range of processes, locations, and intensities of care,⁸ and despite growing demand for these services, little research on hospital readmission has been reported for patients according to rehabilitation setting.

Thus, our study sought to answer 2 questions with implications for efforts to reduce avoidable TKA readmissions: First, are there racial differences in where elective primary TKA recipients receive PAC/Rehab services? Second, is the location of PAC/Rehab services (discharge destination) associated with the likelihood of all-cause 30-day readmission to an acute care hospital?

Methods

Study Sample and Data Set

We used the Pennsylvania Health Care Cost Containment Council (PHC4) database to identify a cohort of patients who underwent elective primary TKA performed in the State of Pennsylvania in the 12 fiscal years (Fys) 2001 to 2012. Cases with TKA were identified using the International Classification of Diseases, Ninth Revision (ICD-9) code 81.54 for primary TKA.

Study Measures

The primary study outcome was the type of PAC to which the patient was discharged. This was defined using the patient’s discharge destination according to the setting in which services were delivered. The categories were home with routine self-care (home-self), home under care of organized home service (home care), skilled nursing facility (SNF), and inpatient rehabilitation facility (IRF). The primary exposure of interest was patient race. Because of the clustered nature of the data, and in order to account for distinct associations between patient-level (within-cluster) race and facility-level (between-cluster) race, we specified both patient- and facility-level measures and included both in our primary analyses.⁹ Facility-level race was defined as the proportion of AA patients at each facility. Patient race was specified as the difference between the patient race indicator and the facility mean proportion of AA patients. Other patient-level variables examined as potential confounders were gender, age (categorized as 18-44, 45-64, 65-84, and 85 and older), insurance type (categorized as Medicaid, Medicare/government, or private), and the occurrence of any of the following major complications during the index stay: venous thromboembolism, myocardial infarction, surgical wound infection, and prosthetic device complication. These individual complications were identified using ICD-9 codes (see Appendix A).

Facility-level rural status was assigned based on rural–urban commuting area codes.¹⁰ Facility procedure volume was determined for each study year based on the number of TKA procedures performed that year. We categorized facility procedure volume into 3 levels: low (less than 200), medium (200-400), and high (more than 400 TKAs performed). Patient comorbidity was summarized using the Elixhauser comorbidity index, a modification of the Elixhauser comorbidity index that combines the 30 Elixhauser indicators into a single numeric score.¹¹ We also examined readmission within 30 days, defined as an inpatient admission within 30 days of discharge from the index surgical hospitalization. The study was approved by the Institutional Review Board at the Philadelphia VA Medical Center.

Data Analysis

Descriptive statistics for patient demographic and clinical characteristics were calculated for the study sample overall, by race, and by Rehab destination. Comparisons of these characteristics in the study cohort were tested using chi-square tests after stratifying the sample by age (<65 years old and >65 years old) to account for age-based differences in Medicare eligibility. Unadjusted associations between patient race and Rehab destination were estimated with multinomial mixed effects regression models, with the 4-level Rehab type as the dependent variable, patient- and facility-level race as independent variables, and facility-level random effects. Multivariable models for the association of race and Rehab destination also included patient gender, age, type of insurance, presence of in-hospital complications, Elixhauser score, hospital rurality, and facility procedure volume. These variables were chosen a priori based on previous research and availability of data in the PHC4 database.

We examined the association between Rehab destination and 30-day readmission. Unadjusted associations between 30-day readmission and Rehab type were estimated with mixed effects logistic regression models, with random facility intercepts. Multivariable mixed effects models also included patient gender, age, race, type of insurance, presence of in-hospital complications, Elixhauser score, facility-level race, facility rurality, and facility procedure volume.

In order to further reduce bias due to nonrandomized procedure for patient referral to type of Rehab service, we conducted a propensity score (PS) analysis.¹² We first developed a logistic regression model for discharge to each PAC or Rehab type (IRF, SNF, or home care) relative to home self. We included patient-level race, gender, age, type of insurance, an interaction of age with insurance type, Elixhauser indicators, in-hospital complications, and facility procedure volume and rural status and estimated a propensity score for each Rehab type for each patient. After evaluating and confirming that observed characteristics were adequately balanced within PS quintiles,¹³ associations between type of Rehab service and 30-day readmission were estimated within PS strata. We report the pooled within-strata comparisons,¹² expressed as odds ratios (OR) for readmission, for each of the Rehab types (IRF, SNF, and home care) relative to home-self. Standard errors and 95% confidence intervals (CIs) were determined using the bootstrap method with 1000 bootstrap samples. Associations between where patient went for PAC or Rehab and 1-year mortality were not determined due to the very low mortality rate. All analyses were performed using SAS 9.3.

Results

Baseline Characteristics of the Study Sample

We identified 322 912 primary TKA procedures performed at 192 hospitals in the state of Pennsylvania between FYs 2001 and 2012. All TKA records indicating a discharge to a nonstudy Rehab destination (n = 1388), patient race not white or black (n = 15 086), or where the index surgery was primary bilateral knee replacement (n = 27 410) were excluded from the analysis. Additional exclusions are detailed in Figure 1.

Figure 1.

Study sample flow and construction of the analytic sample.

In 2004, eligibility criteria for inpatient rehabilitation, commonly referred to as the 75% rule, were revised by CMS.¹⁴ This policy modification resulted in significant shifts of Rehab utilization—reducing use of IRF and increasing use of SNF between 2004 and 2007. Trends in use of Rehab services types stabilized and remained relatively constant between 2008 and 2012, suggesting that implementation of the policy revision was complete. Therefore, our analysis focuses on the cohort of patients who underwent surgery between 2008 and 2012. We identified 129 522 primary TKAs performed at 169 nongovernmental (ie, non-Veterans Affairs) acute care hospitals in the state of Pennsylvania between FYs 2008 and 2012. The final analytic sample consisted of 121 449 (94%) TKA procedures performed on white patients and 8073 (6%) performed on AA patients. Procedures were stratified by patient age (n = 56,575 [43.7%] < 65 and n = 72,947 [56.3%] > 65). The TKAs for AAs involved patients who were younger, more likely to be female, and have Medicaid or Medicare/Government insurance status when younger than 65 years (P < .001 for all comparisons). Nearly half of all TKA procedures resulted in discharge to home with health care services (n = 61 102 [47.17%]). An additional 18% of TKA recipients were discharged to home with self-care (n = 23 507). Remaining procedures were discharged to institutional PAC settings (n = 32 450 [25.05%] to SNF; n = 12 463 [9.62%] to IRF). Selected demographic and clinical characteristics of the study cohort are summarized by race in Table 1, and they are summarized by type of Rehab services in Table 2.

Table 1.

Selected Baseline Characteristics by Race.^a

Race	White				Black				Overall
	<65		≥65		<65		≥65		<65	≥65
	N	%	N	%	N	%	N	%	N	N
Age category	51 693	100	69 756	100	4882	100	3191	100	56 575	72 947
Item
Gender
Female	31 958	61.82	45 236	64.85	3486	71.41	2476	77.59	35 444	47 712
Male	19 735	38.18	24 520	35.15	1396	28.59	715	22.41	21 131	25 235
Age
<45	2086	4.04	–	–	356	7.29	–	–	2442	–
45-64	49 607	95.96	–	–	4526	92.71	–	–	54 133	–
65-84	–	–	65 940	94.53	–	–	3075	96.36	–	69 015
>85	–	–	3816	5.47	–	–	116	3.64	–	3932
Insurance type
Medicaid	2852	5.52	94	0.13	1207	24.72	48	1.5	4059	142
Medicare/Govt	6668	12.9	62 094	89.02	1035	21.2	2718	85.18	7703	64 812
Private	42 173	81.58	7568	10.85	2640	54.08	425	13.32	44 813	7993
Rehab type
IRF^b	2759	5.34	8817	12.64	504	10.32	383	12	3263	9200
SNF^b	6833	13.22	22 048	31.61	1753	35.91	1816	56.91	8586	23 864
Home health care	2394	56.86	28 696	41.14	2176	44.57	836	26.2	31 570	29 532
Home self-care	12 707	24.58	10 195	14.62	449	9.2	156	4.89	13 156	10 351

Abbreviations: IRF, inpatient rehabilitation facility; SNF, skilled nursing facility.

^aIn all categories the p-value is P < 0.001.

Table 2.

Selected Baseline Characteristics by Discharge Destination.

	IRF		SNF		Home health care		Home self-care		Overall
	N	%	N	%	N	%	N	%	N
Discharge destination	12 463	100	32 450	100	61 102	100	23 507	100	129 522
Item
Patient race
White	11 576	92.88	28 881	89	58 090	95.07	22 902	97.43	121 449
Black	887	7.12	3569	11	3012	4.93	605	2.57	8073
Elixhauser score^b
(Mean, SE)	0.66	0.06	0.41	0.02	−0.21	0.01	−0.26	0.02	0.02	0.01
(Median, IQR)	0	5	0	5	0	2	0	2	0	2
Facility race^c
(Mean, SE)	0.07	0	0.09	0	0.06	0	0.03	0	0.06		0.01
(Median, IQR)	0.02	0.06	0.03	0.07	0.03	0.06	0.02	0.03	0.03		2
30-day readmission
No	9111	73.1	29 609	91.24	58 844	96.3	22 675	96.46	120 239
Yes	3352	26.9	2841	8.76	2258	3.7	832	3.54	9283

Abbreviations: IQR, interquartile range; IRF, inpatient rehabilitation facility; SNF, skilled nursing facility; SE, standard error.

^aIn all categories the P value is P < .001.

^bCalculated using weights from van Walraven et al.¹¹

^cReported as facility-level % black.

Race and PAC/Rehab Discharge Destination

Unadjusted odds of IRF referral versus home-self were significantly higher for AA patients compared to whites (<65, OR = 2.48 [95% CI, 2.10-2.93]; >65, OR = 1.71 [95% CI, 1.37-2.13]). The AA race also strongly predicted discharge to SNFs (<65, OR = 3.53 [95% CI, 3.08-4.04]; >65, OR = 2.40 [95% CI, 1.97-2.92]) and discharge to home care in patients younger than 65 years (OR = 1.31 [95% CI, 1.15-1.49]). After adjusting for patient gender, age, type of insurance, presence of in-hospital complications, Elixhauser score, hospital rurality, and hospital procedure volume, the odds of discharge to IRF versus home-self in AA patients remained significantly higher than the odds in whites (<65, OR = 2.06 [95% CI, 1.74-2.45; P < .0001]; > 65, OR = 1.76 [95% CI, 1.40-2.21; P < .001]) as did their odds of discharge to SNF versus home-self (<65, OR = 3.04 [95% CI, 2.64-3.50; P < .001]; >65, OR = 2.31 [95% CI 1.89-2.82; P < .001]) and discharge to home care versus home-self in patients younger than 65 years (OR = 1.32 [95% CI, 1.16-1.50; P < .0001]). Reported ORs according to age are summarized in Figure 2.

Figure 2.

Adjusted odds of IRF, SNF, and home health care referral in African American patients by Medicare eligibility. IRF indicates inpatient rehabilitation facility; SNF, skilled nursing facility.

Postacute Care/Rehab Discharge Destination and 30-Day Readmissions

Figure 3 summarizes the reported unadjusted and adjusted odds ratios for 30-day readmission by IRF, SNF, and home care discharge destinations (referent is home-self for all models). Unadjusted analysis found the highest odds of 30-day readmission among patients discharged to IRFs (OR = 10.06 [95% CI, 9.21-10.99; P < .0001]) followed by patients discharged to SNFs (OR = 2.53 [95% CI, 2.32-2.77; P < .0001]). The strength of the observed association was marginally diminished after adjusting for patient gender, age, race, type of insurance, presence of in-hospital complications, Elixhauser comorbidities, hospital rurality, and hospital procedure volume: discharge to IRF (OR = 7.76 [95% CI, 7.08-8.51; P < .0001]) and SNF (OR = 2.01 [95% CI, 1.83-2.21; P < .0001) significantly predicted 30-day readmission. After stratifying by propensity score quintile for each PAC type, the odds of 30-day readmission among patients discharged to IRF (OR = 8.34 [95% CI, 7.5-9.3; P < .0001]) and SNF (OR = 1.83 [95% CI, 1.65-2.03; P < .0001]) were found to be significant. In both multivariable and propensity-adjusted analyses, discharge to home care following TKA did not increase the likelihood of 30-day readmission (P = .62 and P = .87, respectively).

Figure 3.

Unadjusted, adjusted, and propensity-adjusted odds ratios of 30-day readmission by rehabilation type.

Discussion

In this sample of nearly 130 000 patients who underwent elective primary TKA in the Commonwealth of Pennsylvania, we identified several distinct trends with implications for efforts to reduce TKA readmissions. First, we found significant racial variation in discharge destination after surgery. Compared to white patients and irrespective of Medicare eligibility age, AA patients had significantly higher odds of being discharged to institutional rehabilitation (IRFs and SNFs) following TKA. Second, we found patients discharged to IRFs and SNFs were readmitted to an acute care hospital 8.34 and 1.38 times more often, respectively, than patients discharged home.

Attempts to decrease future readmission rates will require an understanding of the complex interaction of hospital processes that influence readmission and its predisposing factors within patient populations. Numerous prior studies have examined these aspects independently¹⁵; however, few have ventured to examine provider practices and patient-level factors in tandem or in depth. Thus, little is known about underlying mechanisms that account for how nonclinical factors—such as racial minority status or AA patient race—serve to increase the risk of readmission following TKA.

Intuitively and from other types of health disparities research, minority status is a proxy for many underlying “root cause” aspects that lead to poorer patient outcomes^16,17,18,19—aspects that include limited health care coverage and, in turn, limited access to specialists and Rehab services.^20,21 Instructively, most joint replacement studies that have documented racial disparities in readmission rates have used Medicare data in which such coverage-based access limitations are mitigated.^22,23 Thus, for the bulk of TKA recipients, racial differences in access to PAC/Rehab services are unlikely to sufficiently explain observed differences in readmission. The present findings offer an alternative plausible factor: racial variation in the utilization of higher level institutional PAC/Rehab services.

In a health care climate that increasingly emphasizes the importance of resource management and high-value care,¹⁹ there is a striking absence of standardized, evidence-based criteria to help providers determine the most cost-effective and clinically beneficial modality of PAC/Rehab services.^24,25,26,27 As pressure mounts to reduce readmissions, many well-meaning providers may err on the side of higher than necessary PAC/Rehab care as permitted by adequate health care (most often, Medicare) coverage.²⁵ Our findings suggest that such decision making may particularly influence the trajectory of AA patients with TKA and, ultimately, may serve cross purposes, specifically to increase the risk of rehospitalization for these patients.

Our study is one of several that has examined PAC/Rehab utilization patterns after joint replacement surgery and found significant variation in the use of institutional services by race. Freburger et al,²⁸ in a multistate analysis of hospital discharge data, noted AA patients covered by Medicare of private insurance were more likely to use institutional PAC relative to white patients in the same coverage group (OR = 1.69). Further, among all patients receiving institutional PAC, AAs from a majority of states were more likely than whites to use SNF care as opposed to IRF care. Earlier research supports these differences: In a descriptive analysis of Medicare data, Buntin⁸ reported a larger percentage of nonwhite TKA recipients received institutional PAC relative to whites. While comparisons of our findings to the previous research are complicated by the recent phase-in of the newly revised 75% rule—a time of major shifts in rehab utilization patterns—our study confirms that the increased likelihood of AA patients receiving institutional PAC/Rehab services persist in the present policy environment.

To our knowledge, our study is the first to examine both racial variation in PAC/Rehab destination and subsequent risk of readmission by PAC/Rehab setting using a single sample of TKA recipients. However, the specific contribution PAC/Rehab setting to the risk of readmission has been documented in prior joint replacement studies. Similar in part to the present analysis, articles by Bini et al²⁹ and Ramos et al.³⁰ examined readmission rates for patients who underwent primary joint replacement surgery and were discharged to either institutional or home-based rehabilitation settings. Consistent with our results, both analyses found higher risk for readmission in patients who received SNF or IRF care compared to those discharged home with healthcare services.

As noted in both Bini et al²⁹ and Ramos et al,³⁰ there is potential for PAC/Rehab readmission findings to be confounded by indication: As institutional rehabilitation facilities typically serve patients with the greatest needs for supervised care,³¹ it is a reasonable expectation that IRF and SNF patients would be readmitted to acute care hospitals at higher rates than patients discharged home. A major contribution of our study is the use of propensity-adjusted analysis to balance key factors—such as patient age and comorbidity score—that were different between PAC/Rehab patient groups (in effect, mimicking the setting of a randomized, clinical trial).¹² Although administrative data do not provide adequate detailed information to examine all of what shapes clinical decisions on where patients are discharged after surgery, the marked difference in readmission rates between those who received IRF or SNF care and those who went home in our analysis cannot be entirely explained by differences in patient care needs.

There are important limitations to consider when interpreting the results of our study. First, the present analysis used a large administrative database that was designed for hospital performance assessment but contains inadequate information on potential confounding variables including body mass index, preoperative pain, function, and radiographic severity. Second, our sample consisted of only patients who underwent surgery and postacute rehabilitation in the Commonwealth of Pennsylvania. Our conclusions may not be readily extrapolated to other regions or states of the country, as there is significant regional variation regarding the rates of TKAs performed.³² Fourth, it is conceivable that coincident geographic variation in available PAC/Rehab facilities and clustering of patient subgroups by race may play a role in the chosen level of care postdischarge. As many AA patients dwell and receive care in Pennsylvania’s large, urban metropolitan areas, it is possible that these patients consider or are offered different rehabilitation scenarios than their rural counterparts, which could affect our results. Finally, there is recent evidence which suggests that some patients (approximately 4%)³³ receive multiple levels of PAC/Rehab services during an episode of care.³⁴ The degree to which this potential crossover effect may have impacted our finding is unknowable yet must be considered in any interpretation of our findings.

Conclusion

The present study found that AA patients who underwent primary elective TKA across 169 Pennsylvania hospitals were more likely to utilize SNF and IRF compared to white patients and that discharge to institutional scenarios for rehabilitation entails greater odds of 30-day readmission compared to patients discharged home. Future prospective studies should evaluate the criteria currently used by providers and patients to form decisions about postacute setting after TKA and how patient race impacts this process. More work needs to be done to investigate why patients discharged home with home health or self-care are less likely than those admitted to IRF or SNF to be readmitted to an acute care hospital within 30-days after elective TKA.

Footnotes

Appendix A

Authors’ Note

No author has affiliations with or involvement in any organization or entity with a direct financial interest in the subject matter or materials discussed in this article. This article is not under consideration elsewhere. There is no conflict of interest in connection with this report. All the authors listed contributed significantly to this work and have read and approved the article. They take full responsibility for the content of this manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Ibrahim is supported by Grant Number K24AR055259 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases.

References

Kane

Saleh

Wilt

. Total knee replacement. Evidence Report/Technology Assessment (Summary). 2003;(86):1–8. AHRQ Publication No. 04-E006-2.

Kurtz

Ong

Lau

Mowat

Halpern

. Projections of primary and revision hip and knee arthroplasty in the united states from 2005 to 2030. J Bone J Surg Am. 2007;89(4):780–785. doi:89/4/780.

Kurtz

Ong

Schmier

. Future clinical and economic impact of revision total hip and knee arthroplasty. J Bone Joint Surg Am. 2007;89(suppl 3):144–151. doi:89/suppl_3/144.

HCUP National Inpatient Sample (NIS). Healthcare Cost and Utilization Project (HCUP) 2009-2012. Agency for Healthcare Research and Quality, Rockville, MD. Web site. http://hcupnet.ahrq.gov/. Accessed January 1, 2015.

Cram

Kates

Singh

Wolf

. Total knee arthroplasty volume, utilization, and outcomes among Medicare beneficiaries, 1991-2010. JAMA. 2012;308(12):1227–1236. doi:1362022.

Singh

Rosenthal

Ibrahim

Cram

. Racial disparities in knee and hip total joint arthroplasty: An 18-year analysis of national medicare data. Ann Rheumat Dis. 2014;73(12):2107–2115. doi:10.1136/annrheumdis-2013-203494.

Ottenbacher

Karmarkar

Graham

. Thirty-day hospital readmission following discharge from postacute rehabilitation in fee-for-service medicare patients. JAMA 2014;311(6):604–614. doi:10.1001/jama.2014.8.

Buntin

. Access to postacute rehabilitation. Arch Phys Med Rehabil. 2007;88(11):1488–1493. doi:S0003-9993(07)01450-5.

Weingarten

Riedinger

Sandhu

. Can practice guidelines safely reduce hospital length of stay? Results from a multicenter interventional study. Am J Med. 1998;105(1):33–40.

10.

University of Washington. Rural Health Research Center. Rural Urban Commuting Area Codes. Web site. http://depts.washington.edu/uwruca/. Accessed January 1, 2015.

11.

Van Walraven

Austin

Jennings

Quan

Forester

. A modification of the Elixhauser comorbidity measures into a point system for hospital death using administrative data. Med Care. 2009;47(6):626–633.

12.

Rosenbaum

. Observational studies (2nd ed.). New York, NY: Springer Science; 2002.

13.

Austin

. The relative ability of different propensity score methods to balance measured covariates between treated and untreated subjects in observational studies. Med Decis Mak. 2009;29(6):661–677.

14.

Centers for Medicare and Medicaid Services: Final rule Medicare program; changes to the criteria for being classified as an inpatient rehabilitation facility. Fed Regist. 2004;69(89):25752–25776.

15.

Hanchate

Zhang

Felson

Ash

. Exploring the determinants of racial and ethnic disparities in total knee arthroplasty: Health insurance, income, and assets. Medical Care. 2008;46(5):481–488. doi:10.1097/MLR.0b013e3181621e9c.

16.

O’Neil

Lake

Merrill

Wilson

Mann

Bartnyska

. Racial disparities in hospitalizations for ambulatory care-sensitive conditions. Am J Prev Med. 2010;38(4):381–388. doi:10.1016/j.amepre.2009.12.026.

17.

Biello

Rawlings

Carroll-Scott

Browne

Ickovics

. Racial disparities in age at preventable hospitalization among U.S. adults. Am J Prev Med. 2010;38(1);54–60. doi:10.1016/j.amepre.2009.08.027.

18.

Joynt

Orav

Jha

. Patient race, site of care, and 30-day readmission rates among elderly Americans. JAMA. 2011;305(7):675–681. doi:10.1001/jama.2011.123.

19.

McHugh

Carthon

JMB

Kang

. Medicare readmissions policies and racial and ethnic health disparities: a cautionary tale. Policy Polit Nurs Pract. 2010;11(4):309–316. doi:10.1177/1527154411398490.

20.

Metzl

Roberts

. Structural competency meets structural racism: Race, politics, and the structure of medical knowledge. Virtual Mentor. 2014;16(9):674–690. doi:10.1001/virtualmentor.2014.16.09.spec1-1409.

21.

Babitsch

Gohl

von Lengerke

. Re-revisiting Andersen’s Behavioral Model of Health Services Use: a systematic review of studies from 1998–2011. GMS Psycho Soc Med. 2012;9:Doc11. doi:10.3205/psm000089.

22.

Hanchate

Zhang

Felson

Ash

. Exploring the determinants of racial and ethnic disparities in total knee arthroplasty: Health insurance, income, and assets. Med Care. 2008;46(5):481–488. doi:10.1097/MLR.0b013e3181621e9c

23.

Ibrahim

. Racial variations in the utilization of knee and hip joint replacement: An introduction and review of the most recent literature. Curr Orthop Pract. 2010;21(2):126–131. doi:10.1097/BCO.0b013e3181d08223.

24.

Sacks

Lawson

Dawes

Gibbons

Zingmond

. Which patients require more care after hospital discharge? An analysis of post-acute care use among elderly patients undergoing elective surgery. J Am Coll Surg, 2015;220(6):1113–1121. doi:http://dx.doi.org/10.1016/j.jamcollsurg.2015.02.029.

25.

Ackerly

Chen

Ferris

. “How to Rehabilitate Medicare’s ‘Post Acute’ Services.” Harvard Business Review; 2013 [updated October 25th 2013;]. Web site. http://hbr.org/2013/10/how-to-rehabilitate-medicares-post-acute-services/. Accessed June 25, 2015.

26.

Gabrowski

Huckfeldt

Sood

Escarce

Newhouse

. Medicare post acute care payment reforms have potential to improve effeciencey of care, but many need changes to cut costs. Health Aff. 2012;31(9):1941–1950.

27.

Kramer

Holthaus

Gage

Green

Saliba

Coleman

. (2006). Uniform Patient Assessment for Post-Acute Care: Final Report. Aurora, CO: Centers for Medicare & Medicaid Services and the Iowa Foundation for Medical Care; 2006 [updated January 25th 2006;]. Web site. https://www.cms.gov/medicare/quality-initiatives-patient-assessment-instruments/qualityinitiativesgeninfo/downloads/qualitypacfullreport.pdf. Acccessed June 25, 2015.

28.

Freburger

Holmes

. Disparities in post-acute rehabilitation care for joint replacement. Arthritis Care Res. 2011;63(7):1020–1030. doi:10.1002/acr.20477.

29.

Bini

Fithian

Paxton

Khatod

Inacio

Namba

. Does discharge disposition after primary total joint arthroplasty affect readmission rates? J Arthrop. 2010;25(1):114–117. doi:10.1016/j.arth.2008.11.007.

30.

Ramos

Karia

Hutzler

Brandt

Slover

Bosco

. The effect of discharge disposition on 30-day readmission rates after total joint arthroplasty. J Arthrop. 2014;29(4):674–677. doi:10.1016/j.arth.2013.09.010.

31.

Braddom

. Medicare funding for inpatient rehabilitation: How did we get to this point and what do we do now? Arch Phys Med Rehabil. 2005;86(7):1287–1292.

32.

Skinner

Weinstein

Sporer

Wennberg

. Racial, ethnic, and geographic disparities in rates of knee arthroplasty among Medicare patients. N Engl J Med. 2003;349(14):1350–1359, doi: 10.1056/NEJMsa021569.

33.

Hogan

. Medicare beneficiaries’ use of post-acute care trends, 1996 to 2002. Draft working paper developed for MedPAC contract. LLC, Vienna, VA: Direct Research; 2004.

34.

DeJong

. Are we asking the right question about postacute settings of care? Arch Phys Med Rehabil. 2014;95(2), 218–221.