Evaluation of Prior Authorization in Medicare Nonemergent Ambulance Transport
Some Medicare-reimbursed services are overused or improperly used, resulting in payments for unnecessary services.
To determine if prior authorization of services vulnerable to improper use is associated with reduced use and costs without changing patient access or health outcomes.
Design, Setting, and Participants
This study involved repeated cross-sectional evaluation with a state-level matched control group construction and inverse propensity score weighting at the Medicare beneficiary level. Eight states plus the District of Columbia requiring prior authorization were compared with 13 matched comparison group states not subject to prior authorization. Observations on approximately 1.7 million Medicare beneficiaries spanned January 2012 through December 2019. Depending on their state of residence, this included 3 or 4 preintervention years and 4 or 5 postintervention years.
Ambulance suppliers were directed to request prior authorization for Repetitive, Scheduled, Non-Emergent Ambulance Transport (RSNAT) services; failure to do so resulted in prepayment claim review. The goal of prior authorization is to reduce use of nonemergency ambulance transports that do not meet Medicare coverage criteria.
Main Outcomes and Measures
Primary outcomes included total cost of care, RSNAT use rates and expenditures, unplanned hospital admission, emergency department admission, and emergency ambulance use per beneficiary-year. All measures were constructed from Medicare claims.
Approximately 1.7 million Medicare beneficiaries were observed in the study. The results showed that prior authorization was associated with a 2.4% reduction in total annual expenditures for a total of $1530 per beneficiary-year (95% CI, −$1775 to −$1285; P < .001); a 61% reduction in the probability of RSNAT use for a total of 4.1 percentage points per beneficiary-year (95% CI, −4.26 to −3.94; P < .001); a 77% reduction in RSNAT expenditures for a total of $1136 per beneficiary-year (95% CI, −$1179 to −$1093; P < .001); a 1.4% reduction in the probability of emergency department use by 0.99 percentage points per beneficiary-year (95% CI, −1.17 to −0.81; P < .001); no change in the probability of emergency ambulance use (0.07 percentage points, 95% CI, −0.15 to 0.29; P = .50); a 2.6% reduction in the probability of unplanned hospital admission for a total of 1.53 percentage points per beneficiary-year (95% CI, −1.71 to −1.35; P < .001); and a 19% annual increase in the probability of emergency dialysis use for a total of 1.4 percentage points per beneficiary with end-stage renal disease (95% CI, 1.28 to 1.60; P < .001).
Conclusions and Relevance
In this difference-in-differences analysis of Medicare beneficiaries, the results suggest that the RSNAT Prior Authorization Model was associated with reduced costs with little or no change in the quality or access indicators examined. Targeted approaches to prior authorization may be an appropriate control measure for Medicare services vulnerable to improper use.