Analyzing 30-Day Episode Costs for Total Joint Arthroplasty: Identifying Targets for Cost Reduction Under Bundled Payment Models.

Abstract

The Bundled Payments for Care Improvement Initiative works to more effectively coordinate care and reduce costs across healthcare services. A better understanding of the primary drivers of healthcare spending that fall within the bundled payment window will be beneficial to prioritizing quality care and minimizing unexpected costs, especially with the implementation of the Transforming Episode Accountability Model. This retrospective cohort study used data from January 1, 2022, to December 31, 2022. We identified commercially managed Medicare patients who underwent lower extremity surgery using Medicare Severity Diagnosis Related Groups (MS-DRG) codes of 469 and 470. A random sample of 100 patients was selected from each MS-DRG cohort. Postoperative services occurring 30 days from a patient's discharge date were identified and described. The total average 30-day episode of care costs for patients who had an MS-DRG code of 470 was $25,245 per surgical procedure. Of this cost, 84.0% was due to the index hospitalization, averaging $21,207 per case. Home health visits were the costliest postoperative service. The 30-day episode of care costs were much higher among patients who had an MS-DRG code of 469, totaling $34,462 per surgical procedure. The index hospitalization averaged $25,802 per case, compromising 74.9% of the total. Outside of the index hospitalization, readmissions and services skilled nursing facilities services were the two most costly postoperative contributors. Patients classified as MS-DRG 469 incurred higher costs than patients who had the MS-DRG 470 code. This is mostly due to increased postoperative costs such as readmissions and skilled nursing facility stays. This data highlights the importance of interventions aimed at more effectively preventing readmission and managing post-discharge care as hospitals transition to bundled payment for Medicare reimbursement.