Prior Authorization as a Strategy to Reduce Unnecessary Utilization of Repetitive Scheduled Non-emergency Ambulance Transportation

Background: Amid concerns that repetitive scheduled non-emergency ambulance transportation (RSNAT) services are overused in the Medicare program, in 2014 the Centers for Medicare and Medicaid Services launched a three-year prior authorization model for these services in three states. The model states were chosen because utilization was high relative to other states, and because higher levels of improper utilization were suspected. In 2015, the model was expanded to six neighboring states. We evaluate the impact of prior authorization as a strategy to reduce unnecessary RSNAT utilization and expenditures, and assess the model’s impact on total Medicare expenditures, quality of care, and adverse outcomes.

Study Design: We used Medicare claims data and a multivariate difference-in-differences design to assess the impact of the prior authorization requirement on key outcomes among Medicare beneficiaries with end-stage renal disease (ESRD) and/or pressure ulcers of the skin. We selected comparison states based on their similarity to model states on state-level demographic and health characteristics, and on RSNAT capacity and utilization measures. We also weighted individual Medicare beneficiaries to ensure balance between the treatment and comparison groups on observable characteristics.

Population Studied: We studied utilization and expenditures for Medicare beneficiaries with ESRD and/or pressure ulcers in nine treatment states and 13 comparison states from January 2012 through June 2017.

Principal Findings: We find that the prior authorization model reduced RSNAT service utilization and expenditures by over 65 percent among beneficiaries with ESRD and/or pressure ulcers, with larger impacts observed among beneficiaries with ESRD. Total Medicare fee-for-service expenditures also declined by almost two percent for the study population as a whole, although expenditures increased for beneficiaries with pressure ulcers. This increase was primarily due to increased long term care hospital expenditures, possibly a result of community-residing beneficiaries moving into other settings to access care upon the introduction of prior authorization. Utilization of emergency department services and emergency ambulance transport did not change overall; we found small increases in utilization among beneficiaries in Year 2 states but no change among beneficiaries in Year 1 states. Among beneficiaries with ESRD, we found a small decrease in scheduled dialysis and a small increase in emergency dialysis. These findings were also concentrated in states which were not selected for high baseline RSNAT utilization rates.

Conclusions: The prior authorization model had a dramatic impact on RSNAT service use and expenditures for beneficiaries with ESRD and/or pressure ulcers. However, the impacts were larger in states with initially high RSNAT use, suggesting that savings from nation-wide implementation may be lower than those estimated in the model. Estimated impacts on access to treatment were small, and concentrated in states without initially high RSNAT use.