Association for Public Policy Analysis & Management

Background: The process of deinstitutionalization has radically transformed care for people with serious mental illness (SMI) in the United States. Indeed, between 1955 and 2016 the number of beds in state psychiatric hospitals has decreased more than 96 percent on a population-adjusted basis. One factor contributing to this is Medicaid’s “Institutions for Mental Diseases (IMD) Exclusion,” which prohibits reimbursement for most acute psychiatric and substance abuse care delivered to beneficiaries aged 21 through 64 in specialty psychiatric hospitals and residential facilities. By restricting payment for care in particular settings, the IMD exclusion creates a financial incentive for states to limit care in IMD settings and instead rely on non-IMD alternatives, including community mental health services but also hospital emergency departments (EDs) and inpatient beds. Since general hospital beds and emergency rooms are more expensive and less clinically appropriate settings than dedicated psychiatric facilities for the treatment of acute SMI, policymakers interested in the budgetary and public health effects of the IMD exclusion have focused on care in the hospital setting. Previous studies have found little evidence that psychiatric bed availability or the IMD exclusion affect the number of patients with SMI using emergency rooms or general hospital beds. However, these studies generally do not include a control group and they focus on short-run outcomes that may not capture supply responses by providers.

Method: The study population consists of individuals needing acute care for SMI or a substance use disorder. There are five study outcomes for which I estimate the effect of the IMD exclusion: the probability that an individual uses a hospital ED; disposition from the ED; time spent in the ED; length of stay for those admitted as inpatients; and total hospital charges. I measure these outcomes in seven states using the all-payer State Inpatient Database (SID) and State Emergency Department Database (SEDD), which combined include all ED visits and inpatient admissions to community hospitals in those states. I estimate the effects of the IMD exclusion using two complementary methods. The first approach is a triple difference model relying on quasi-experimental variation in the applicability of the IMD exclusion resulting from the cutoff at age 21, while controlling for unrelated changes in outcomes that occur at age 21. This approach estimates long-run effects of the exclusion, since the age cutoff has existed unaltered for decades. The second approach uses a differences-in-differences model relying on multiple changes over the sample period in state Medicaid waivers that effectively nullify the IMD exclusion for Medicaid managed care enrollees when in effect. The regression models also include controls for hospital characteristics from the American Hospital Association annual survey, including the number of dedicated psychiatric and other beds.

Results and Discussion: Results are not available as of the abstract submission deadline. To find that the IMD exclusion affects hospital care outcomes would corroborate the assertions of states, mental health advocates, and emergency physicians. Alternatively, to find no effect would reinforce the arguably counterintuitive results of the existing empirical literature using a markedly different approach.