Causal Effects of Utilization Management on Health Care Utilization

Utilization management is becoming increasingly prevalent in the Medicare Part D program, but existing studies of the effects of utilization management are confined to specific subpopulations including low-income individuals and nursing home residents. In this paper, I analyze the effects of utilization management on health care utilization among enrollees in Medicare Part D plans, excluding dual eligible and low-income individuals. I measure utilization management in a plan as the share of a person’s predicted drug utilization that is subject to three forms of utilization management—prior authorization, step therapy, and quantity limits. I identify the causal effect of utilization management by instrumenting for utilization management with the weighted average of the utilization management shares across all plans offered in an individual’s market, where I weight each formulary by the share of drugs covered (a salient characteristic in plan choice).

I find that utilization management policies have heterogeneous effects on Medicare spending, with prior authorization increasing spending in Parts A, B, and D, while step therapy decreases spending in Parts A and B and increases spending in Part D, and quantity limits decrease spending in Parts A, B, and D. Effects on health care utilization are also varied, with prior authorization increasing emergency room visits and hospitalization, while step therapy has the opposite effect, and quantity limits decreasing emergency room utilization and hospitalizations on the extensive, but not the intensive, margin.

I next turn to identifying the degree to which these effects of utilization management are mediated by prescription drug use using a causal mediation model based on work by Dippel, Gold, Heblich, and Pinto (2017). Across almost all outcomes, I find that prescription drug utilization, measured by the number of 30-day equivalent prescriptions, decreases health care spending and utilization. I also find that majority of the effect of step therapy and quantity limits is mediated by prescription drug spending, but this is not true for prior authorization requirements, which continue to have a measurable and clinically significant effect on spending and health care utilization. The remaining direct effect of prior authorization is particularly strong on Medicare Part B spending, indicating that prior authorization may be affecting drug utilization that substitutes for ambulatory care.