The Effect of the Affordable Care Act’s Medicaid Expansion on Access to Care from Licensed Substance Abuse Clinics

This study estimates the extent to which access to substance abuse treatment as well as the types of treatment are impacted by the Affordable Care Act (ACA). To causally identify the effect on access to substance abuse treatment facilities, we exploit both geographic and temporal variation created by the state expansions to Medicaid after the ACA. Medicaid expansion increased access to substance abuse treatment among those who became newly eligible for Medicaid, which in turn influences if an individual treatment facility decides to accept Medicaid as a form of payment. Furthermore, individual facilities may decide to change the forms of treatment used either shifting towards or away from medically assisted therapy.

Recent studies have shown that the expansion of Medicaid increased the likelihood of Medicaid being used as a form of payment for treatment substantially and an increase in Medicaid-reimbursed prescriptions for medications (Maclean and Saloner 2017). This also leads to an increase in Medicaid spending on medications to treat opioid use disorder (Clemans-Cope, Epstein and Kenney 2017). Yet many of the administrative databases used for these studies are constrained to using state and year estimates. We improve on the existing literature by using a unique panel dataset we created containing geocoded information for the near census of licensed substance abuse facilities in the country between the years 2005 and 2017. The rich administrative facility panel data contain information on the types of insurance accepted as well as the types of treatment by the individual facility. For identifying the effects on acceptance of Medicaid and the use of pharmacological therapies we leverage both geographic and temporal variation in the presence of a facility within a county and the number of facilities across counties.

We find that the Medicaid Expansion impacted the likelihood of an individual facility accepting Medicaid as well as the likelihood of the facility using buprenorphine. There was no effect on methadone. Our model is robust to the inclusion of state fixed effects, zip code fixed effects and individual substance abuse facility fixed effects. These results are important from a policy standpoint because there is an active and vigorous debate over whether the recent Medicaid Expansion was a contributor to the current opioid epidemic.