Although little is known about its effectiveness as a cost containment policy, most state Medicaid programs have instituted prior authorization (PA) before dispensing specific drugs to discourage the use of relatively more expensive medications. This project will examine the impact of the use of prior authorization in Michigan in March 2002 on the use and costs of antidepressants (AD) and other health services by disabled Medicaid enrollees who are also enrolled in Medicare. Careful selection and dosing of antidepressant agents, sometimes after initial treatment failures, can improve response rates, increase compliance, decrease possible adverse drug events, and improve long-term health outcomes. This is particularly true for individuals with comorbid mental and physical illness. No previous study has evaluated the effects of prior authorization on use of antidepressant medications or non-drug health services. Using multiple years (1999-2004) of Medicaid and Medicare claims data, we will evaluate whether the administrative barriers imposed by the PA policy in Michigan (Ml) resulted in unintended outcomes such as reductions in the use of antidepressant agents, reduced duration of treatment, increased risk of AD-drug interactions, or an increase in the use of other, more expensive, health care services. The study population will consist of community-dwelling permanently disabled fee-for-service Ml Medicaid enrollees aged 18-64. We will use an interrupted time series design to control for pre-policy trends in use of medications and health services in Michigan (Ml) and in an identically identified control cohort in Indiana (IN), which excluded AD medications from its prior authorization program. Approximately 32,625 and 39,375 enrollees will meet the study criteria in Ml and IN, respectively, for inclusion in the Phase I analyses of trends in and predictors of antidepressant use at baseline. The analyses for Phases II and III will examine the impact of the policy on antidepressant drug use, incidence of drug-drug interactions, use of other health care services, and total health care costs. We will restrict the analyses in Phases II and III to patients who initiated antidepressant therapy in the 12 months before and after the implementation of the PA policy in Michigan (N equals approximately MI:5,758; IN:6,950). All analyses will be stratified by type of disability (mental, physical, developmental).
