Regardless of the definition, most of the spending variation between high-cost and low-cost hospitals is due to postacute care. CJR refers to the Comprehensive Care for Joint Replacement model, and Hospital Compare refers to the 90-day total hip arthroplasty and/or total knee arthroplasty payment measure. Hospital quartiles were created by categorizing hospitals into 4 groups based on mean 90-day payments for joint replacement.
CJR refers to the Comprehensive Care for Joint Replacement model, and Hospital Compare refers to the 90-day total hip arthroplasty and/or total knee arthroplasty payment measure.
eTable 1. Attribution algorithm for the CJR and Hospital Compare 90-day episode definitions for joint replacement compared
eTable 2. Subgroup analysis of alternative episode definitions
eTable 3. Analysis of readmission rates
eTable 4. Structural characteristics of hospitals in Michigan hospitals vs all acute care hospitals in the United States
eFigure. Correlation in average 90-day episode payments between the broad CJR and the clinically-narrow Hospital Compare episode definitions
Customize your JAMA Network experience by selecting one or more topics from the list below.
Ellimoottil C, Ryan AM, Hou H, Dupree JM, Hallstrom B, Miller DC. Implications of the Definition of an Episode of Care Used in the Comprehensive Care for Joint Replacement Model. JAMA Surg. 2017;152(1):49–54. doi:10.1001/jamasurg.2016.3098
Will unrelated expenditures resulting from the Comprehensive Care for Joint Replacement (CJR) model’s broad definition of an episode of care affect hospital performance?
In this Medicare claims–based analysis that included 23 251 Michigan patients who underwent joint replacement during the period from 2011 through 2013, there was a strong correlation between the CJR model’s broad definition and a clinically narrow definition of an episode of care. Moreover, the average 90-day payment difference between these 2 types of episodes of care was small.
In the context of joint replacement bundled payments, these data suggest that hospital performance will be consistent whether a broad or clinically narrow definition is used.
Under the Comprehensive Care for Joint Replacement (CJR) model, hospitals are held accountable for nearly all Medicare payments that occur during the initial hospitalization until 90 days after hospital discharge (ie, the episode of care). It is not known whether unrelated expenditures resulting from this “broad” definition of an episode of care will affect participating hospitals’ average episode-of-care payments.
To compare the CJR program’s broad definition of an episode of care with a clinically narrow definition of an episode of care.
Design, Setting, and Participants
We identified Medicare claims for 23 251 patients in Michigan who were Medicare beneficiaries and who underwent joint replacement during the period from 2011 through 2013 at hospitals located in metropolitan statistical areas. Using specifications from the CJR model and the clinically narrow Hospital Compare payment measure, we constructed episodes of care and calculated 90-day episode payments. We then compared hospitals’ average 90-day episode payments using the 2 definitions of an episode of care and fit linear regression models to understand whether payment differences were associated with specific hospital characteristics (average Centers for Medicare & Medicaid Services–hierarchical condition categories risk score, rural hospital status, joint replacement volume, percentage of Medicaid discharges, teaching hospital status, number of beds, percentage of joint replacements performed on African American patients, and median income of the hospital’s county). We performed analyses from July 1 through October 1, 2015.
Main Outcomes and Measures
The correlation and difference between average 90-day episode payments using the broad definition of an episode of care in the CJR model and the clinically narrow Hospital Compare definition of an episode of care.
We identified 23 251 joint replacements (ie, episodes of care). The 90-day episode payments using the broad definition of the CJR model ranged from $17 349 to $29 465 (mean [SD] payment, $22 122 [$2600]). Episode payments were slightly lower (mean payment, $21 670) when the Hospital Compare definition was used. Both methods were strongly correlated (r = 0.99, P < .001). The average payment difference between these 2 types of episodes of care was small (mean [SD], $452 [$177]; range, $73-$1006). In our multivariable analysis, we found that the hospital characteristics examined had a minimal impact or no impact on the payment differential.
Conclusions and Relevance
The average 90-day episode payments determined by both definitions of an episode of care were strongly correlated, and there was a small payment differential for most hospitals. In the context of joint replacement bundled payments, these data suggest that hospital performance will be consistent whether a broad or clinically narrow definition of an episode of care is used.
The Centers for Medicare & Medicaid Services (CMS) recently announced a bundled payment program based on the episode of care for patients undergoing lower-extremity joint replacement.1,2 This initiative—the Comprehensive Care for Joint Replacement (CJR) program—was designed to improve transitions of care and coordination across multiple health care professions and, therefore, reduce the number of episode payments associated with this commonly performed procedure. Accordingly, the model will hold hospitals financially accountable not only for joint replacement–related payments but also for nearly all medical and surgical expenditures that occur during the initial hospitalization until 90 days after hospital discharge (ie, an episode of care).
Supporters of this broad definition of an episode of care argue that hospitals should be held responsible for expenses that derive from both surgical complications and suboptimal management of medical comorbidities during and after the index admission. For example, in the CJR program, payments associated with a readmission for acute exacerbation of congestive heart failure that occurs 83 days after hospital discharge will be attributed to the episode of care during the joint replacement. Policymakers at CMS and elsewhere are optimistic that, by defining episodes to include a very broad range of postdischarge services, this payment reform will accelerate coordination and optimization of medical and surgical care for Medicare beneficiaries undergoing joint replacement surgery.
Although many stakeholders contend that this broad definition of an episode of care is necessary to drive the higher level of care, others have expressed concern that this approach will unfairly penalize hospitals for postdischarge expenditures that are more directly related to the patient’s underlying health condition than to the joint replacement procedure.1 Notably, during the same time period when the CJR model was announced, CMS also publically released the total hip arthroplasty and/or total knee arthroplasty payment measure.3 Similar payment measures for acute myocardial infarction, pneumonia, and congestive heart failure have been publically reported on the Hospital Compare website4 (CMS will likely begin reporting on the total hip arthroplasty and/or total knee arthroplasty payment measure in 2017). In contrast to the CJR model’s broad definition of an episode of care, the Hospital Compare payment measure attributes a much more narrow set of postdischarge claims to the joint replacement–related episode of care. Although nearly all claims are attributed to the joint replacement–related episode of care within 30 days of the procedure, only claims related to specific postoperative complications are attributed for 31 to 90 days (eg, mechanical complications and periprosthetic joint infections). Accordingly, for this measure, if a patient is readmitted with an acute exacerbation of congestive heart failure more than 30 days after hospital discharge, the claim is not included in the Hospital Compare payment measure. Proponents of the clinically narrow definition of an episode of care used by Hospital Compare argue that this attribution method is better suited for joint replacement because it holds health care professionals accountable for the complications and care delivery most directly relevant to the surgery, without unfairly penalizing hospitals that take care of patients with more frequent or more severe medical comorbidities.
In reality, however, little is known about the degree to which these different definitions of an episode of care actually affect hospital payments for joint replacements. A better understanding of this issue would provide clinicians, hospital administrators, and policymakers with insight on the direct financial implications of CMS’s selection of a broad definition of an episode of care for the CJR program. Moreover, information on the implications of claim attribution can be used to guide the development of future bundled payment programs. In this context, we used Medicare claims data from 2011 through 2013 to calculate and compare average 90-day episode payments using the broad definition of an episode of care in the CJR model and the clinically narrow definition of an episode of care from Hospital Compare. We also examined whether the observed payment differences between these 2 definitions were associated with specific hospital characteristics.
We analyzed Medicare claims for beneficiaries in the state of Michigan who underwent lower-extremity joint replacement (Medicare severity diagnosis related group codes 469 and 470) during the period from 2011 through 2013. We included in our analysis only those patients who had surgery in a hospital meeting the following criteria: (1) the hospital was geographically located in a metropolitan statistical area (an inclusion criteria for the CJR program), and (2) the hospital performed more than 20 cases during the study period.
To ensure the availability of complete claims from the index hospitalization until 90 days after hospital discharge, we excluded patients who received a joint replacement after October 1, 2013. We also excluded beneficiaries who were not continuously enrolled in Medicare Parts A and B, those who died during the episode of care, and those who had health maintenance organization coverage or were eligible for Medicare because of end-stage renal disease or disability. In addition, we excluded patients who received a primary diagnosis of hip fracture (12.3% of episodes).
We used data from the 2012 American Hospital Association annual survey and the Census Bureau to assess hospital characteristics for facilities included in this analysis. This study was deemed exempt from review by the institutional research board at our health system because the data were deidentified.
We defined 90-day episodes of care according to specifications from both the CJR program and the Hospital Compare total hip arthroplasty and/or total knee arthroplasty payment measure.1,5 Fundamentally, these definitions of an episode of care differ in the scope of postdischarge services that are considered related to the joint replacement surgery. The definition of an episode of care in the CJR program includes a very wide range of medical and surgical postdischarge expenditures. Throughout this article, we also refer to the definition in the CJR programs as a “broad” definition. The Hospital Compare definition of an episode of care, in contrast, includes a more narrow range of postdischarge claims that are more directly related to complications resulting from joint replacement surgery. Herein, we also refer to the definition used by Hospital Compare as a “clinically narrow” definition. Details for both definitions are summarized in eTable 1 in the Supplement.
For the broad definition of an episode of care, we aggregated all hospitalization, professional, and post–acute care claims from the index admission through 90 days after hospital discharge and then removed claims that matched the exclusion list published by CMS in the CJR Final Rule.1 For the clinically narrow definition of an episode of care, all index hospitalization and post–acute care claims within 30 days were attributed to the bundle; however, this definition only includes claims “related” to the joint replacement surgery after 30 days.5
Next, we calculated the 90-day payments for each surgical episode according to both the broad and clinically narrow definitions. To do this, we first aggregated the payments received for each claim included in the respective episode definitions. We then removed intentional payments by CMS for disproportionate share hospitals, indirect medical education, and new technologies. Next, following the methods used by CMS in the CJR Final Rule, we excluded episodes with 90-day payments lower than $4000 and capped payments that were higher than 2 SDs above the mean at the 95th percentile. Finally, we accounted for national price trends by adjusting payments by the consumer price index.
We assessed the effect of the 2 definitions of an episode of care by calculating the correlation between each hospital’s average 90-day episode payments using the CJR and Hospital Compare algorithms. In addition to the statewide analysis, we performed several subgroup analyses to confirm the stability of our findings. To perform these analyses, we stratified hospitals by average CMS–hierarchical condition categories (HCC) risk score, rural hospital status, joint replacement volume, percentage of Medicaid discharges, teaching hospital status, number of beds, percentage of joint replacements performed on African American patients, and median income of the hospital’s county. The CMS-HCC risk score includes data on age, sex, 70 potential comorbidities, and dual eligibility status and the original reason for Medicare entitlement derived by reviewing 12 months of inpatient, outpatient, and select professional claims. We selected subgroups based on hospital characteristics that may influence determinants of total episode cost (eg, readmissions).6-8 For each subgroup, we calculated average 90-day episode payments using both the CJR definition and the Hospital Compare definition and then calculated the correlation between the 2 definitions for all hospitals in these subgroups.
For each hospital, we first calculated the magnitude of the difference between average 90-day payments determined by the CJR and Hospital Compare definitions. Next, we fit a linear regression model with heteroscedastic robust standard errors to examine associations between hospital characteristics and the payment differential. For these models, the payment differential was used as our dependent variable. Similar to the subgroup analyses, the average CMS-HCC risk score, rural hospital status, joint replacement volume, percentage of Medicaid discharges, teaching hospital status, number of beds, percentage of joint replacements performed on African American patients, and median income of the hospital’s county were used as independent variables.
We performed our primary analysis without including fractures because patients undergoing hip fracture surgery are typically different from patients undergoing elective joint replacement. Accordingly, they have been excluded from CMS’s Hospital Compare measure. However, because hip fractures are included in the CJR program, we performed a sensitivity analysis with hip fractures included to confirm the stability of our primary findings. All analyses were performed using statistical software (Stata 13/SE; StataCorp) and at the 5% significance level. Our sensitivity analysis (with hip fracture included) yielded no substantive changes to our primary findings (eTable 2 in the Supplement).
We identified 23 251 patients who underwent a lower-extremity joint replacement (740 cases for Medicare severity diagnosis related group code 469 and 22 511 cases for Medicare severity diagnosis related group code 470) during the period from January 1, 2011, through October 1, 2013, at 64 hospitals in Michigan. Using the broad definition of an episode of care, we found that 90-day payments varied widely across hospitals, ranging from $17 349 to $29 465 (mean [SD] payment, $22 122 [$2600]). Although the magnitude of variation was similar, the mean 90-day episode payment was lower ($21 670) when calculated using the Hospital Compare definition (Table). The most expensive quartile of hospitals and the least expensive quartile of hospitals differed in 90-day episode payments by $6383 and $6209 using the CJR and Hospital Compare definitions, respectively. As illustrated in Figure 1, this difference was largely due to variation in post–acute care expenditures (eg, skilled nursing facility payments, inpatient rehabilitation, and home health care).
We identified a very strong correlation between average 90-day episode payments using the broad and clinically narrow definitions (r = 0.99, P < .001), indicating extreme consistency in a hospital’s costs with regard to episodes of care using both definitions (eFigure in the Supplement). Moreover, the mean (SD) difference in 90-day episode payment between the broad and clinically narrow definitions was only $452 ($177 [range, $73-$1006]), representing only 2% of the average total episode payment based on the broad definition (Figure 2).
In univariate regression analyses, hospital CMS-HCC risk score (P = .03) and proportion of African American patients treated (P < .001) were associated with a greater difference in payments between the 2 definitions of an episode of care. In our multivariable analysis, only proportion of African American patients remained significantly associated with the payment differential (P = .04). While this variable was statistically significant, the effect size was small. For each percentage point increase in patients who were African American, there was a $4.00 (95% CI, $0.60-$7.75) increase in the payment difference between the CJR and Hospital Compare definitions of an episode of care. Our sensitivity analysis (with hip fractures included) yielded no substantive changes to our primary findings.
In this analysis, we directly compared 2 definitions of an episode of care used by CMS to measure 90-day episode payments for patients undergoing lower-extremity joint replacement surgery. Specifically, the CJR model’s broad definition was compared with the clinically narrow Hospital Compare definition. Using both definitions, we found wide variation in 90-day episode payments (largely driven by differences in post–acute care utilization). We identified a strong correlation between 90-day episode payments calculated using both definitions; moreover, the magnitude of the payment difference was small relative to the overall episode costs. Collectively, these findings suggest that because average 90-day episode payments were similar between the broad and clinically narrow episode definitions, it is likely that hospital performance will be consistent whether a broad or clinically narrow definition is used.
Explanations for the tight correlation, and small absolute difference, between payments based on the broad definition and payments based on the clinically narrow definition stem mainly from the epidemiology of postdischarge care after joint replacement. Total 90-day episode payments for joint replacement are composed mostly of expenditures for the index hospitalization and from postacute care. These payments are also included in more clinically narrow definitions.9,10 The main difference with the CJR model’s broad definition is that almost all medical readmissions are attributed to the episodes of care related to the joint replacement. However, these events are rare after a joint replacement; in our study and in prior work, at 90 days, the overall unplanned readmission rate after joint replacement was approximately 6% to 8% (eTable 3 in the Supplement), and only 18% to 25% of these readmissions result from medical causes.11-13 Accordingly, regardless of the definition used, hospitals should focus their resources on reducing post–acute care expenditures to perform well on the CJR model and the Hospital Compare payment measure.
Our study has several limitations. First, it is possible that our analysis of Michigan hospitals may not be generalizable to hospitals nationally. However, we believe that the results would be no different in a national sample for 2 primary reasons. First, all hospitals in metropolitan statistical areas selected for the CJR model are required to participate in the program. Accordingly, the CJR-participating hospitals vary greatly in regard to teaching status, number of beds, and other structural characteristics. Similarly, Michigan hospitals are also diverse according to these same characteristics (eTable 4 in the Supplement). Second, we performed subgroup analyses around many of the structural characteristics that could potentially affect the generalizability of our results and found no difference in our primary findings. Our second limitation is that while we considered 2 definitions of an episode of care for joint replacement, many other definitions are considered by commercial payers. However, the 2 definitions evaluated in this study are most relevant to policymakers because they are currently being used in CMS performance programs. Finally, our finding of the deep congruence between the broad and clinically narrow definitions may not hold true for conditions in which patients have more comorbidities and in which medical readmissions may constitute a greater proportion of the variation in 90-day episode payments (eg, vascular surgery procedures and coronary artery bypass graft).
These limitations notwithstanding, our results have several important implications. Our finding that average 90-day episode payments based on the broad definition used in CJR are similar to those calculated using the more clinically narrow Hospital Compare definition should reassure both policymakers and hospital administrators that the definition alone will not unfairly penalize facilities that treat sicker patients with multiple medical conditions. Furthermore, because it does not rely on nuanced, condition-specific clinical details, the broad definition of an episode of care for CJR represents a much more rapidly scalable model for CMS because it inevitably moves toward bundled payment programs and episode payment measures for other conditions.14 However, our findings in no way imply that risk standardization is unnecessary when measuring 90-day episode payments for joint replacement. While readmissions for medical conditions after joint replacement are infrequent, age and comorbidities will still affect 90-day episode payments through several mechanisms, including longer length of stay, greater use of postacute care, and greater frequency of inpatient complications.
For hospital leaders and orthopedic surgeons practicing at one of the approximately 800 hospitals involved in the CJR program, our findings should reduce the understandable concerns about being unfairly penalized for readmissions and other expenditures that are not directly related to joint replacement procedures. Instead, our findings clarify the key point that a hospital’s success with episode payments for joint replacement will depend to a large extent on its ability to reduce post–acute care spending without compromising patient outcomes.
Our collective findings suggest that the broad definition of an episode of care used by CMS for the CJR model will not unintentionally penalize hospitals that treat patients with multiple medical conditions because 90-day readmissions resulting from conditions unrelated to the joint replacement are infrequent. Moving forward, however, future research in this area should focus on whether a broad vs clinically narrow definition would affect similar bundled payment programs for other clinical conditions associated with a higher prevalence of significant comorbidities. Patients undergoing these procedures may have a higher rate of unrelated readmissions and other expenditures. Such analyses are particularly important if CMS applies the CJR model’s broad definition ubiquitously to future bundled payment programs. Moreover, it will be important for policymakers and researchers to monitor the unintended consequences of bundling, including care stinting and patient selection. These concerns would exist regardless of the definition used but may be more apparent if hospitals believe that medically complex patients will affect their bottom line under the broad definition of an episode of care. In the end, the goal of any bundled payment program should be to motivate hospitals and clinicians to reduce episode payments and improve the delivery of care. To achieve this laudable goal in an equitable manner, policymakers should aim to define episodes of care in a way that does not lead to unintended consequences such as reducing access for beneficiaries or financially penalizing hospitals for treating patients with inherently expensive therapies.
Accepted for Publication: June 9, 2016.
Corresponding Author: Chad Ellimoottil, MD, MS, Institute for Healthcare Policy and Innovation, Department of Urology, University of Michigan, 2800 Plymouth Rd, Bldg 16, Ann Arbor, MI 48109 (email@example.com).
Published Online: September 28, 2016. doi:10.1001/jamasurg.2016.3098
Author Contributions: Dr Ellimoottil had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Ellimoottil, Ryan, Dupree, Hallstrom, Miller.
Acquisition, analysis, or interpretation of data: Ellimoottil, Hou, Dupree, Hallstrom, Miller.
Drafting of the manuscript: Ellimoottil, Dupree, Miller.
Critical revision of the manuscript for important intellectual content: Ellimoottil, Ryan, Hou, Dupree, Hallstrom.
Statistical analysis: Ellimoottil, Ryan, Hou.
Administrative, technical, or material support: Ellimoottil, Ryan, Hou, Dupree.
Study supervision: Miller.
No additional contributions: Hallstrom.
Conflict of Interest Disclosures: Dr Miller receives salary support from Blue Cross Blue Shield of Michigan for his role as the director of the Michigan Urological Surgery Improvement Collaborative and the Michigan Value Collaborative. Dr Hallstrom receives salary support from Blue Cross Blue Shield of Michigan as the codirector of the Michigan Arthroplasty Registry Collaborative Quality Initiative. Dr Dupree receives salary support from Blue Cross Blue Shield of Michigan for his role as the codirector of the Michigan Value Collaborative and as a resource physician for the Michigan Urological Surgery Improvement Collaborative. No other disclosures are reported.
Funding/Support: This research was supported by the Agency for Healthcare Research and Quality (grant 1F32HS024193-01 to Dr Ellimoottil and grant K01HS018546 to Dr Ryan) and by funding from the National Cancer Institute (grant 1-R01-CA-174768-01-A1 to Dr Miller).
Role of the Funder/Sponsor: The funding organizations had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.