Association for Public Policy Analysis & Management

Mental illnesses and substance use disorders (SUDs), collectively referred to as ‘behavioral health’, are prevalent and lead to morbidity/mortality, healthcare costs, employment problems, and relationship difficulties for affected individuals. In addition to internal costs, behavioral health conditions impose negative externalities on broader society in terms of healthcare use, crime, disability payments, and so forth.

While mental illness and SUDs imposes large costs on both affected individuals and society, these illnesses can be effectively treated with appropriate use of healthcare services. However, most individuals who would benefit from behavioral health treatment do not receive it; with lack of insurance coverage and inability to pay for treatment being critical barriers.

In this study, we provide the first evidence on the effect of losing public insurance on behavioral healthcare utilization. We exploit plausibly exogenous variation in insurance coverage generated by one of the largest disenrollments in the history of Medicaid: a 2005 reform in Tennessee that lead to 190,000 enrollees losing insurance (‘TennCare’). The individuals who lost TennCare are similar to those who became eligible through ACA Medicaid expansion. To study reform effects, we employ rich survey data from the National Survey of Drug Use and Health and administrative data on hospitalizations. We estimate differences-in-differences models in which we compare geographically similar U.S. states to Tennessee before and after the reform.

A contribution of our study is that we are able to study the effect of losing, as opposed to gaining, insurance. While the literature has focused primarily on estimating the effect of gains due to available sources of exogenous variation in coverage, the effects of insurance gains and losses are likely not symmetric. For instance, when an insured individual is diagnosed with a particular behavioral health condition, insurance likely provides access to a provider. Interactions with the provider provides formal treatment, which improves health, and information on how to manage the condition. While access to formal treatment likely declines after an insurance loss, health information on condition management is likely retained. Thus, there is a paucity of research on the effects of losses; in particular, there is no evidence within the context of behavioral health.

The second contribution of this paper focuses on data quality. Data from the nationwide inpatient sample (NIS) is used in several papers that deal with mental health and other health outcomes. This database is region and nationally represented but sometimes used to perform a state-level analysis. We offer a path forward to use this data for state-level analysis and compare our results without this adjustment.

Our preliminary findings suggest that post-disenrollment, mental health care use declined and unmet need increased in Tennessee relative to comparison states. We do not find strong evidence of changes for SUD-related hospitalizations, however. We further show that Medicaid coverage and use of Medicaid to pay for treatment declined in Tennessee relative to comparison states post-disenrollment. These findings are immediately important given current policy debates within the U.S. on whether or not to maintain ACA Medicaid expansions to non-disabled low-income childless adults.