That is the question that Carroll and co-authors try to answer in their latest NBER working paper (WP #23926). They examine the Arkansas Health Care Payment Improvement Initiative (APII), which is a state-wide, multi-payer episode-based program. Unlike most episode-based payment (EBP) models, provider participation in the program was mandatory (as of 2013). NBER summarizes the program as follows:
The APII initially covered five types of health care episodes, including perinatal care. Like many modern EBP programs, the APII employs a retrospective payment model, where providers are paid FFS while they oversee episodes, but face reconciliation payments at the end of the year. The provider’s annual average spending per episode (adjusted for patient risk factors) is calculated based on episodes for which they served as Principal Accountable Provider (PAP). Each PAP’s average episode spending is then deemed to be either commendable, acceptable, or unacceptable based on pre-determined thresholds. PAPs with unacceptable ratings are responsible for half of the spending beyond the acceptable level, while those with commendable ratings can share in half of the savings.
The authors use a difference-in-difference approach comparing cost in Arkansas before and after the APII implementation to changes in spending for neighboring states for whom no EBP was implemented. Costs were measured for episodes constructed from claims data for perinatal episodes. The authors found that:
…in the first full year of EBP implementation, spending per episode declined by 3.8 percent, or $403, in Arkansas relative to the control states. The savings were driven by slower spending growth in Arkansas after EBP implementation, while spending growth continued on a similar trajectory in the control states.
Over 80 percent of these savings stem from a large (6.6 percent) reduction in spending on inpatient facility care. The researchers find that this decline was largely driven by changes in the price of inpatient care rather than in the quantity of care. While unable to test directly for a mechanism underlying this effect, they suggest that a change in referral patterns is a likely cause. Outside of inpatient facility care, the implementation of EBP led to few changes in perinatal care. Declines in physician spending and outpatient spending were small and statistically insignificant, as were changes in utilization, including caesarean section rates and the length of inpatient stays. In terms of quality measures, EBP implementation was associated with improvements in chlamydia screening rates but no other changes.
I would guess that most policy wonks and academics (yourself included) would have assumed that EBP would have little effect on price, but would incentivize providers to reduce health care utilization. In the case of perinatal care, one may hypothesize that the number of cesareans would go down. Perhaps it is easier for provider’s managers to dictate to physicians which providers they should use (i.e., lower cost ones) rather than to try to mandate to physicians that they need to change their practice patterns. This is just a hypothesis, but it will be interesting to see if the results from Arkansas are also found in other EBP initiatives.
- Carroll, Caitlin, Michael Chernew, A. Mark Fendrick, Joe Thompson, and Sherri Rose. Effects of Episode-Based Payment on Health Care Spending and Utilization: Evidence from Perinatal Care in Arkansas. No. w23926. National Bureau of Economic Research, 2017.