Impact of an Episode-Based Payment Initiative by Commercial Payers in Arkansas on Procedure Volume: an Observational Study.

BACKGROUND: Episode-based payment (EBP) is gaining traction among payers as an alternative to fee-for-service reimbursement. However, there is concern that EBP could influence the number of episodes. OBJECTIVE: To examine how procedure volume changed after the introduction of EBP in 2013 and 2014 under the Arkansas Health Care Payment Improvement Initiative. DESIGN: Using 2011-2016 commercial claims data, we estimate a difference-in-differences model to assess the impact of EBP on the probability of a beneficiary having an episode for four procedures that were reimbursed under EBP in Arkansas: total joint replacement, cholecystectomy, colonoscopy, and tonsillectomy. PARTICIPANTS: Commercially insured beneficiaries in Arkansas serve as our treatment group, while commercially insured beneficiaries in neighboring states serve as our comparison group. INTERVENTIONS: Statewide implementation of EBP for various clinical conditions by two of Arkansas' largest commercial insurers. MAIN MEASURES: For a given procedure type, the primary outcomes are the annual rate of procedures (number of procedures per 1000 beneficiaries) and the probability of a beneficiary undergoing that procedure in a given quarter. KEY RESULTS: The relationship between EBP and procedure volume varies across procedures. After EBP was implemented, the probability of undergoing colonoscopy increased by 17.2% (point estimate, 2.63; 95% CI, 1.18 to 4.08; p < 0.001; Arkansas pre-period mean, 15.29). The probability of undergoing total joint replacement increased by 9.9% (point estimate, 0.091; 95% CI, - 0.011 to 0.19; p = 0.08; Arkansas pre-period mean, 0.91), though this effect is not significant. There is no discernable impact on cholecystectomy or tonsillectomy volume. CONCLUSIONS: We do not find clear evidence of deleterious volume expansion. However, because the impact of EBP on procedure volume may vary by procedure, payers planning to implement EBP models should be aware of this possibility.