Simulating the Impact of the Transforming Episode Accountability Model

ABSTRACT

Objectives: To simulate the financial impact of Medicare’s Transforming Episode Accountability Model (TEAM) on CMS and assess variation in incentive payments across hospital types and geographic regions from 2026 to 2030.

Study design: We developed a simulation model to project hospital-level incentive payments and CMS savings under alternative TEAM design scenarios. The model incorporated hospital and patient characteristics, clinical episode categories, and target prices to estimate reconciliation payments, accounting for TEAM’s mandatory participation and episode structure. Analyses included projections at the national level and by hospital type and state.

Methods: We applied random forest models using data from Model Year 4 (2021) of the Bundled Payments for Care Improvement Advanced (BPCI-A) program obtained through a Freedom of Information Act request. We identified 710 acute care hospitals required to participate in TEAM based on the CMS participant list after linking the data to BPCI-A target price and Inpatient Prospective Payment System impact files (2018-2021).

Results: Assuming a $500 per-episode reduction in spending, TEAM is projected to yield $2.6 million in savings in 2026 and $46.6 million in 2030. In 2030, safety-net hospitals are projected to receive the highest average reconciliation payments ($962 per episode), followed by rural ($620) and urban non–safety-net hospitals ($77). Nebraska is projected to receive the highest average payments, whereas Oregon may face the largest penalties.

Conclusions: TEAM is projected to produce modest Medicare savings with substantial variation in hospital payments, highlighting the importance of balancing incentives across hospital types and regions.

Am J Manag Care. 2026;32(10):557-562.

doi:10.37765/ajmc.2026.90027

Takeaway Points

Using national data and machine learning–based methods, we projected that TEAM would save $2.6 million in 2026 and $46.6 million by 2030—still a modest amount relative to total Medicare spending—with substantial variation in hospital payments.

• Safety-net hospitals are projected to receive the highest reconciliation payments compared with rural and urban non–safety-net hospitals.
• Reconciliation payments exhibit substantial variation across hospital types and geographic regions.
• Policy makers should carefully balance financial risk and reward to promote equitable and effective cost control.

The Transforming Episode Accountability Model (TEAM), which was launched in January 2026 by CMS, is a mandatory episode-based payment model for hospitals. The model builds upon previous initiatives, such as the Bundled Payments for Care Improvement Advanced (BPCI-A) and the Comprehensive Care for Joint Replacement (CJR) models. TEAM aims to improve the quality of care and reduce Medicare spending, with a focus on 5 major episodes of care: lower-extremity joint replacement, hip or femur fracture surgery, spinal fusion, coronary artery bypass grafting, and major bowel procedures.¹

As with previous bundled payment models, participating hospitals are assigned an episode-specific spending target for selected clinical conditions. Participants continue to receive Medicare fee-for-service payments, which are reconciled with the predetermined target price after each performance period. Participants receive bonuses if spending is below the target and incur penalties if spending exceeds the target. As with the BPCI-A model, TEAM incorporates quality performance measures by adjusting reconciliation payments based on the Composite Quality Score (CQS).

Nevertheless, there are differences between the 2 models. First, BPCI-A was a voluntary program, whereas TEAM is mandatory. All acute care hospitals in selected Core-Based Statistical Areas (CBSAs) are required to participate in TEAM (certain hospitals that participated in BPCI-A or CJR through the end of those models were given a 1-time opportunity to opt in to TEAM). Second, although TEAM evaluates spending in the 30-day postdischarge period, BPCI-A focused on the 90-day postdischarge period. Third, TEAM focuses only on the 5 surgical procedures mentioned above, whereas BPCI-A included a wider range of both medical and surgical episodes.

The mandatory nature of TEAM may address important limitations of voluntary participation models such as BPCI-A. In BPCI-A, voluntary participation limited its impact in 2 important ways. First, a relatively small share of eligible patient episodes were subject to the program. In Model Year 4, approximately 3.3 million hospital discharges and outpatient procedures were eligible for BPCI-A, but only 18.9% were under the model.² The limited inclusion of episodes in BPCI-A had limited gross savings. Second, hospitals selectively participated in BPCI-A and chose episodes more likely to yield bonus payments.³ Although evidence suggests that BPCI-A reduced medical spending, bonus payments to hospitals exceeded these reductions, resulting in net losses to CMS.^2,4-9 With mandatory participation in TEAM, hospitals cannot selectively choose episodes that maximize financial advantage. This design feature should limit excess bonus payments to participants.

Nonetheless, the financial impact of TEAM is unknown. The experience of hospitals under BPCI-A is a good starting point to understand the impact of TEAM for 2 reasons: (1) TEAM shares many design features with BPCI-A, and (2) BPCI-A included episodes and participants from regions that substantially overlap with TEAM regions.

Incentive Structure of TEAM

Under TEAM, hospitals have the option to participate in 3 tracks, which vary in terms of risk and reward. Track 1, which is available to all hospitals in 2026 (performance year 1) and to safety-net hospitals through 2028, offers hospitals a low level of rewards with no downside risk. Specifically, it includes a 10% stop-gain and allows up to a 10% CQS adjustment on positive reconciliation amounts. Track 2, available in 2027 (performance year 2) through 2030 (performance year 5), offers a low level of reward and risk, defined by a 5% stop-gain and stop-loss, with up to a 10% CQS adjustment for positive reconciliation amounts and a 15% CQS adjustment for negative reconciliation amounts. In contrast, Track 3, available from 2026 through 2030, presents a high level of reward and risk, featuring a 20% stop-gain and stop-loss, with up to a 10% CQS adjustment applied to both positive and negative reconciliation amounts.¹ Under TEAM, all participants can choose between Track 1 and Track 3 for 2026. In 2027 and 2028, safety-net hospitals can choose among Tracks 1, 2, and 3, while rural, Medicare-dependent, sole community, and essential access community hospitals can choose between Tracks 2 and 3. From 2029 through 2030, safety-net hospitals and these other eligible hospitals can choose between Tracks 2 and 3.¹ All other hospitals are limited to Track 3.¹

The Current Study

In this context, we used a variety of data sources to simulate the financial impact of TEAM on both CMS and participating hospitals, based on the experience with BPCI-A. Rather than providing definitive predictions of realized TEAM results, our goal was to generate policy-relevant projections that could inform how TEAM’s design features may affect reconciliation payments, net CMS spending, and the distribution of financial incentives across hospital types and geographic regions during the early implementation period.

METHODS

Data Sources, Study Population, and Variables

We used information on target prices and reconciliation payments for hospital participants from Model Year 4 (2021) of BPCI-A to assess the likely impact of TEAM. The analytic data set was structured at the hospital-episode level, where each observation represents a hospital’s participation in a specific episode under BPCI-A. We assessed the experiences of hospital participants in Model Year 4 alone because we anticipated that the design of BPCI-A Model Year 4 would be more similar to TEAM than that of other model years. Specifically, like TEAM, Model Year 4 limited hospitals’ ability to selectively choose advantageous episodes by requiring participation across specific service lines. In addition, Model Year 4 applied a retrospective adjustment to target prices based on national spending trends that we expect to be included in TEAM.

Information on hospitals’ participation, target prices, and reconciliation payments for the BPCI-A Model Year 4 was obtained through a Freedom of Information Act request.These data include episode-level information on baseline spending, performance period spending, reconciliation payments, and episode counts for participating hospitals. We then identified acute care hospitals required to participate in TEAM using the CMS TEAM participant list. Hospitals were included if they were located in one of the CBSAs designated for mandatory participation under TEAM. This process identified 741 hospitals that would be required to participate in the model.

Next, we linked these hospitals to the Inpatient Prospective Payment System impact files for 2018-2021 and the BPCI-A target price files to obtain hospital and patient characteristics that were used as predictors in the simulation model. These characteristics included bed size, teaching status, safety-net status, urban/rural location, geographic region, percentage of Medicare patients, ratio of Medicare operating costs to Medicare-covered charges, and case mix index. Bed size was defined as small (0-250 beds), medium (251-500 beds), large (501-850 beds), and extra large (851 or more beds). A safety-net hospital was defined as a hospital in which more than 60% of Medicare fee-for-service inpatient admissions during the fiscal year involved beneficiaries with either full or partial Medicaid enrollment.¹⁰

Hospitals were excluded if key variables required for the simulation model were missing, including hospital characteristics or target price data in the linked data sets. After applying these criteria, 710 of the 741 hospitals required to participate in TEAM were included in the analytic sample used for simulation.

Simulating the Impact of TEAM on CMS Reconciliation Payment and Net Financial Changes Under Different Design Scenarios

Our goal was to simulate the impact of TEAM on reconciliation payments across different scenarios based on the model’s proposed approaches. To estimate these payments under TEAM, we used reconciliation payment data from Model Year 4 of BPCI-A. Because BPCI-A was a voluntary program, hospitals could choose not to participate at all or only join specific clinical episode categories. Therefore, simulating TEAM requires predicting reconciliation payments for hospital episode categories that were not observed in BPCI-A, making a flexible prediction approach appropriate for this analysis.

To achieve this, we used the random forest method—a robust machine learning approach that constructs multiple decision trees during training and aggregates their predictions.^11,12 We used random forests as a predictive tool for simulation rather than to estimate causal effects of hospital characteristics on reconciliation payments. We selected this approach because reconciliation payments under bundled payment models may reflect nonlinear relationships and interactions that are difficult to specify a priori. Random forests are well suited to this setting because they can flexibly model these relationships without requiring prespecified functional forms. Prior studies in health services research have used random forest models to predict health care outcomes.^13-15 By incorporating hospital and patient characteristics, the category and type of clinical episodes, and the target price, we estimated the average reconciliation payment across all hospitals. We partitioned our data set into 80% for training and 20% for testing, applied 5-fold cross-validation for model tuning, then evaluated prediction accuracy in the held-out testing sample using mean absolute error (eMethods [eAppendix available at ajmc.com]). Hospitals participating in the BPCI-A Model Year 4 were used to train the prediction model. Model performance was evaluated on the test sample, and the trained model was then applied to hospitals required to participate in TEAM to generate predicted reconciliation payments for the simulation.

To simulate the impact of TEAM, we limited the category of clinical episodes to the 5 conditions proposed in the model. Because TEAM evaluates spending in the 30-day postdischarge period, we converted the 90-day postdischarge spending into a 30-day equivalent by applying a scaling multiplier. The average 30-day costs were $21,000 for lower-extremity joint replacement, $40,500 for hip or femur fracture surgery, $42,000 for spinal fusion, $50,000 for coronary artery bypass grafting, and $29,000 for major bowel procedures.¹⁶ 

We then constructed 3 distinct participation scenarios based on the program’s 3 tracks (eAppendix Table 1). In all 3 scenarios, we assumed that all hospitals would choose to participate in Track 1 in 2026, since Track 1 carries no downside risk yet provides rewards when hospitals perform well. This is consistent with provider choices in other alternative payment models.^17,18

In Scenario 1, we assumed that all rural, Medicare-dependent, sole community, and essential access community hospitals would choose Track 2 (with less risk and less reward) over Track 3, whereas safety-net hospitals would remain in Track 1 through 2028. Beginning in 2029, safety-net hospitals were included in the Track 2 and Track 3 assignments. In Scenario 2, half of these eligible hospitals chose Track 2, and the other half chose Track 3. In Scenario 3, three-fourths of these eligible hospitals chose Track 3. Individual hospitals were assigned to tracks based on their propensity to participate in BPCI-A (eMethods). For each year, we estimated the average reconciliation payments per episode overall, by hospital type, and by state. Based on various estimates of the impact of bundled payments on medical spending, we assumed that reductions in episode spending would be either $100 per episode, $500 per episode, or $1000 per episode.^{2,4,6-9,19-24} Combining these alternative estimates of gross spending reductions with estimates of reconciliation payments, we calculated the net change in CMS spending.

RESULTS

Hospital Characteristics

We identified 710 acute care hospitals required to participate in TEAM starting January 2026 (Table 1) and compared their characteristics with those of 2530 acute care hospitals not required to join TEAM. Among the hospitals required to participate in TEAM, 67.9% were small, 17.5% were teaching hospitals, 83.8% were located in urban areas, and 6.2% were safety-net hospitals. Among hospitals not required to participate, 76.7% were small, 9.1% were teaching hospitals, 72.7% were located in urban areas, and 5.7% were safety-net hospitals. Compared with the nonrequired hospitals, the required hospitals had a lower Medicare share (0.35 vs 0.37), a higher case mix index (1.71 vs 1.65), and a similar operating cost to charge ratio (0.27 vs 0.28).

Estimating the Average Reconciliation Payment per Episode and Net Change in CMS Spending Under the TEAM Design

The estimated reconciliation payments per episode (payment amounts rounded for readability throughout) are $481, $152, $160, $160, and $169 from 2026 to 2030, respectively (Table 2). Lower reconciliation payments after 2026 resulted from a required shift toward downside risk. In all years, CMS is projected to pay financial bonuses to participating hospitals, with safety-net and rural hospitals consistently receiving higher reconciliation payments than urban non–safety-net hospitals. For instance, in 2030, safety-net hospitals are projected to receive $962 per episode, rural hospitals $620, and urban non–safety-net hospitals $77.

Across all 5 years, hospitals in Nebraska ($3495 per episode), Nevada ($1888), Kansas ($1433), South Dakota ($1028), and Texas ($947) were projected to receive the highest reconciliation rewards per episode from CMS (Figure; eAppendix Table 2). In contrast, hospitals in Oregon (–$644) and Idaho (–$577) were projected to incur the highest reconciliation penalties per episode. In comparison, California ($8.6 million), New Jersey ($8.0 million), and New York ($7.9 million) were projected to receive the highest aggregate average yearly reconciliation payments, whereas Massachusetts (–$2.5 million) and Oregon (–$1.8 million) were projected to receive the lowest. These variations reflect regional differences in hospital characteristics and expected performance under TEAM.

Based on assumed gross savings in medical spending of $100, $500, and $1000 per episode, we estimated the net change in CMS spending. In 2026, CMS would experience net spending changes of $53.7 million (losses), –$2.6 million (savings), and –$73.1 million (savings), respectively. In 2030, CMS is projected to experience changes of $9.7 million (losses), –$46.6 million (savings), and –$117.1 million (savings) under these spending-reduction assumptions.

The estimated reconciliation payments per episode are $164, $172, $174, and $183 for Scenario 2 and $164, $173, $174, and $184 for Scenario 3 from 2027 to 2030, respectively (eAppendix Table 3). These estimates are similar to those in Scenario 1.

DISCUSSION

Results from our simulation suggest that TEAM could lead to modest changes in CMS spending, ranging from annual net savings of $117 million to annual net losses of $54 million between 2026 and 2030. The potential $117 million in savings in 2030 corresponds to approximately 0.02% of projected Medicare hospital care spending.^17,25,26 Bonus payments are projected to vary substantially across hospitals, with urban non–safety-net hospitals projected to receive the lowest incentive payments and safety-net hospitals projected to receive the highest. Driven by differences in bonus payments across hospital characteristics, reconciliation amounts are also projected to vary substantially across states and regions.

To our knowledge, this is the first analysis to simulate and project the financial payments that CMS may incur under TEAM and how bonus payments may vary across hospitals. The modest financial impact of TEAM projected in our simulation is consistent with previous studies of bundled payment programs like BPCI, BPCI-A, and CJR. This study is also consistent with prior research showing variation in hospitals’ receipt of bonuses across characteristics.^6,27 This may be due to CMS’s target pricing methodology, which groups hospitals into peer cohorts. Hospitals serving more complex, low-income populations tend to have higher baseline spending, resulting in higher target prices and greater opportunities to generate savings.⁶ Differences in bonus payments across hospital characteristics are relatively consistent across all years, suggesting structural advantages for certain hospital types.

Our results could provide policy makers with valuable insights for improving the design of TEAM. For instance, in Track 1 of TEAM, all hospitals face no downside risk and low levels of reward for 1 year.¹ With smaller incentives under this track, the potential for savings on medical spending may be more limited. This highlights the need for careful balancing of risk and reward in program design. Additionally, more meaningful differentiation between Track 2 and Track 3 may be warranted, given that only a small share of hospitals are projected in our simulation to reach the stop-loss or stop-gain threshold that distinguishes Tracks 2 and 3. Moreover, the projected variation in reconciliation payments across hospital types and geographic regions suggests that hospitals may face different financial incentives and constraints under the model depending on their characteristics and local market conditions. Policy makers may therefore consider whether adjustments to incentive levels or risk thresholds across tracks are needed to ensure that financial incentives are appropriately aligned across different hospital types and geographic regions while maintaining financial stability for participating hospitals.

More broadly, the realization of savings under TEAM will depend on how hospitals respond to the financial incentives embedded in the model. BPCI-A was voluntary, and hospitals that chose to participate may have been better prepared to implement approaches that reduce episode spending. As a result, savings observed in BPCI-A may not fully translate to hospitals required to participate in TEAM. However, if hospitals participating in TEAM adopt comparable approaches to managing episode spending, they may achieve similar savings over time. Conversely, potential savings may be smaller if hospitals face limited opportunities to reduce episode spending within a shorter 30-day episode window. Differences in hospital resources, care coordination capabilities, and strategic responses to financial incentives may also influence the magnitude of savings. As TEAM implementation progresses, observed spending, reconciliation payments, quality performance, and hospital responses will provide direct evidence on the model’s financial effects. Early results that are directionally consistent with our projections would lend support to the simulation framework and reinforce the relevance of the policy considerations identified in this study.

Our study has several limitations. First, we used data on bonus payments from Model Year 4 of BPCI-A to forecast the TEAM results. This approach is limited because the data set is from 2021 and pertains to a predecessor program. However, given the many similarities in the design of the 2 programs, data from BPCI-A were the best source for understanding the impact of TEAM. Second, we used the preliminary target price from BPCI-A instead of the final target price.

This choice may limit the accuracy of our simulation, as the preliminary target price is based on historical data and projections and lacks adjustments for actual performance-period data and factors such as the Peer Group Trend Factor Adjustment. Nevertheless, the target price was used solely to estimate the stop-loss and stop-gain thresholds. Few hospitals are affected by these stop-loss and stop-gain limits. Third, the study primarily focuses on hospital and episode-level characteristics, and we did not incorporate detailed patient-level data due to data limitations. This could limit the ability to capture variations in patient populations across hospitals, potentially influencing reconciliation payments under TEAM. Nevertheless, we included a case mix index to account for differences in patient complexity across hospitals.

Fourth, our model may be subject to selection bias because it relies on reconciliation payment data from hospitals that participated in the voluntary BPCI-A program, which may differ systematically from nonparticipating hospitals. In addition, our projections rely on national Medicare spending trends rather than hospital- or market-specific episode spending trends, which may not fully capture local variation in spending patterns across hospitals or regions. Finally, our analysis focuses on cost avoidance and does not assess potential impacts on care quality, which is an important aspect of value-based payment models. Nonetheless, evaluation evidence to date suggests that bundled payment has made little to no appreciable impact on measured quality.

CONCLUSIONS

Our analysis suggests that TEAM may have a relatively minor impact on CMS spending over its first 5 years. Bonus payments to hospitals are projected to be associated with hospital characteristics, potentially leading to systematic payment differences across states and regions. These findings highlight the need for policy makers to carefully balance risk and reward structures to manage costs effectively while ensuring equitable financial outcomes for hospitals.

Author Affiliations: Department of Health Services, Policy, and Practice, Brown University School of Public Health (ZZ, JDB, AMR), Providence, RI.

Source of Funding: This study was funded by the National Institutes of Health (NIH) under award numbers R01AG047932 and R01HL160588.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (ZZ); acquisition of data (ZZ, JDB, AMR); analysis and interpretation of data (ZZ, JDB, AMR); drafting of the manuscript (ZZ, JDB, AMR); critical revision of the manuscript for important intellectual content (ZZ, JDB, AMR); statistical analysis (ZZ, JDB, AMR); provision of patients or study materials (ZZ); obtaining funding (JDB, AMR); and supervision (JDB, AMR).

Address Correspondence to: Zehui Zhou, MPH, Doctoral Candidate, Brown University School of Public Health, 121 South Main St, Providence, RI 02903. Email: zehui_zhou@brown.edu.

REFERENCES

1. Transforming Episode Accountability Model (TEAM). CMS. 2026. Accessed September 9, 2026. https://www.cms.gov/priorities/innovation/innovation-models/team-model
2. Henke R, Bradley A, Bacon K, et al. CMS Bundled Payments for Care Improvement Advanced Model: Fifth Annual Evaluation Report. CMS. 2024. Accessed April 15, 2025. https://www.cms.gov/priorities/innovation/data-and-reports/2024/bpci-adv-ar5
3. Berlin NL, Gulseren B, Nuliyalu U, Ryan AM. Target prices influence hospital participation and shared savings in Medicare bundled payment program. Health Aff (Millwood). 2020;39(9):1479-1485. doi:10.1377/hlthaff.2020.00104
4. Joynt Maddox KE, Orav EJ, Zheng J, Epstein AM. Year 1 of the Bundled Payments for Care Improvement–Advanced model. N Engl J Med. 2021;385(7):618-627. doi:10.1056/nejmsa2033678
5. Joynt Maddox KE, Orav EJ, Zheng J, Epstein AM. Medicare’s Bundled Payments for Care Improvement Advanced model: impact on high-risk beneficiaries. Health Aff (Millwood). 2022;41(11):1661-1669. doi:10.1377/hlthaff.2022.00138
6. Shashikumar SA, Gulseren B, Berlin NL, Hollingsworth JM, Joynt Maddox KE, Ryan AM. Association of hospital participation in Bundled Payments for Care Improvement Advanced with Medicare spending and hospital incentive payments. JAMA. 2022;328(16):1616-1623. doi:10.1001/jama.2022.18529
7. Dummit L, Tripp A, Bergman S, et al. CMS Bundled Payments for Care Improvement Advanced Model: Year 2 Evaluation Report. CMS; 2021. Accessed April 15, 2025. https://www.cms.gov/priorities/innovation/data-and-reports/2021/bpci-yr2-annual-report
8. Somers J, Bradley A, Bacon K, et al. CMS Bundled Payments for Care Improvement Advanced Model: Third Evaluation Report. CMS; 2022. Accessed April 15, 2025. https://www.cms.gov/priorities/innovation/data-and-reports/2022/bpci-adv-ar3
9. ‌Somers J, Bradley A, Bacon K, et al. CMS Bundled Payments for Care Improvement Advanced Model: Fourth Evaluation Report. CMS; 2023. Accessed April 15, 2025. https://www.cms.gov/priorities/innovation/data-and-reports/2023/bpci-adv-ar4
10. BPCI Advanced Target Price Specifications: Model Year 4. CMS; 2022. Accessed April 15, 2025.
    https://www.cms.gov/priorities/innovation/media/document/bpciadvanced-targetprice-specs-my4-v3
11. Breiman L. Random forests. Mach Learn. 2001;45(1):5-32. doi:10.1023/a:1010933404324
12. Athey S, Imbens GW. Machine learning methods that economists should know about. Annu Rev Econ. 2019;11(1):685-725. doi:10.1146/annurev-economics-080217-053433
13. Wang J, Xu S, Wang Y, El Bouhissi H. An analytics framework for healthcare expenditure forecasting with machine learning. Healthc Anal. 2025;8:100428. doi:10.1016/j.health.2025.100428
14. Rose S. A machine learning framework for plan payment risk adjustment. Health Serv Res. 2016;51(6):2358-2374. doi:10.1111/1475-6773.12464
15. Kumagai N, Jakovljević M. Random forest model used to predict the medical out-of-pocket costs of hypertensive patients. Front Public Health. 2024;12:1382354. doi:10.3389/fpubh.2024.1382354
16. Mechanic RE, Perloff J, Koppel D. Mandatory Medicare bundled payment and the future of hospital reimbursement. Am J Manag Care. 2024;30(suppl 13):SP1050-SP1058. doi:10.37765/ajmc.2024.89653
17. Report to the Congress: Medicare and the Health Care Delivery System. MedPac; 2024. Accessed April 16, 2025. https://www.medpac.gov/document/june-2024-report-to-the-congress-
    medicare-and-the-health-care-delivery-system
18. Joynt Maddox KE, Orav EJ, Zheng J, Epstein AM. Participation and dropout in the Bundled Payments for Care Improvement initiative. JAMA. 2018;319(2):191-193. doi:10.1001/jama.2017.14771
19. Shashikumar SA, Zheng J, Orav EJ, Epstein AM, Joynt Maddox KE. Changes in cardiovascular spending, care utilization, and clinical outcomes associated with participation in Bundled Payments for Care Improvement–Advanced. Circulation. 2023;148(14):1074-1083. doi:10.1161/circulationaha.123.065109
20. Ryan AM, Shashikumar SA, Chopra Z, Joynt Maddox KE, Buxbaum JD. Reconciliation payments in the Bundled Payments for Care Improvement Advanced program and reductions in clinical spending needed for CMS to avoid financial losses. JAMA. 2023;329(14):1221-1223. doi:10.1001/jama.2023.1435
21. Dummit LA, Kahvecioglu D, Marrufo G, et al. Association between hospital participation in a Medicare bundled payment initiative and payments and quality outcomes for lower extremity joint replacement episodes. JAMA. 2016;316(12):1267-1278. doi:10.1001/jama.2016.12717
22. Barnett ML, Wilcock A, McWilliams JM, et al. Two-year evaluation of mandatory bundled payments for joint replacement. N Engl J Med. 2019;380(3):252-262. doi:10.1056/nejmsa1809010
23. CMS Comprehensive Care for Joint Replacement Model: Performance Year 2 Evaluation Report. CMS; 2019. Accessed April 4, 2025. https://www.cms.gov/files/document/cjr-secondannrptpdf.pdf
24. Navathe AS, Troxel AB, Liao JM, et al. Cost of joint replacement using bundled payment models. JAMA Intern Med. 2017;177(2):214-222. doi:10.1001/jamainternmed.2016.8263
25. National health expenditure data: projected. CMS. June 2026. Accessed September 9, 2026. https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure-data/projected
26. American Hospital Association. Total number of hospital admissions in the U.S. from 1946 to 2022 (in 1,000). Statista. January 12, 2024. Accessed March 19, 2025. https://www.statista.com/statistics/459718/total-hospital-admission-number-in-the-us/
27. Muldoon LD, Cohen D, Pelizzari P, Lampert M, Williams E. BPCI Advanced: Key Considerations for Prospective Model Participants. Milliman; 2019. Accessed April 11, 2025. https://www.milliman.com/en/insight/bpci-advanced-key-considerations-for-prospective-model-participants