Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

A substantial number of private inpatient psychiatric hospital closures over the past decade: nearly 40% of private specialty psychiatric hospitals closed between 1995 and 2004 and these closures were only slightly offset by new entry. Similar trends are observed in public psychiatric hospitals though much of the closure occurred earlier: by 1995 15.5% of the state psychiatric hospitals that existed in 1995 closed by 2004. These changes have the potential to create a variety of negative effects related to reduced access to care. If closure is undertaken with a focus solely on costs and without the understanding that alternative care in the community must exist, society may be made worse in the long run. However, it is also possible that, depending on the pattern of closure and the characteristics of the institutions that close, patient care could be improved. Documenting the circumstances under which closure of state psychiatric hospitals can lead to improved societal outcomes and increased societal benefits is an important goal of this application. The purpose of our research is to estimate the determinants of public and private inpatient psychiatric facility closure. A key area of our research will focus on determine the extent closure occurs due to low demand versus political/budget (for public facilities) or reimbursement that is too low despite strong demand (for private facilities).

We analyze a 17-year panel of hospitals from the 1988-2005 American Hospital Association (AHA) Annual Survey data for the national study of the determinants of psychiatric hospital closure. The AHA data represents a national survey of hospitals that includes specialty psychiatric hospitals and institutions, regardless of ownership. The AHA data will be merged with Area Resource File (ARF) data from 1988-2005, containing information on county-level demographics and the health-care infrastructure of the county; HMO market-level data (1988-2005); and data on state funding of inpatient and community mental health care available from the National Association of State Mental Health Program Directors Research Institute, Inc. (NRI) State Mental Health Agency Revenues and Expenditures Study. In addition, we have collected information on whether or not mental health is carved-out from the Medicaid Program, and whether there were changes in the way mental health care was reimbursed during the period.

We will estimate discrete time hazard models to model the determinants of public and private psychiatric hospital closure. The primary determinants are political, the competitive landscape, availability of treatment substitutes, and reimbursement factors. Thus the probability of closing a psychiatric hospital can then be modeled as:

Pr (Close t+1) = f(Politicalt, Competitiont, Treatment Substitutest, Reimbursementt, Demographics t, Industry t),

where closure in the following year (time period t +1) is modeled as a function of the previous year's characteristics (time period t). This conforms to the plausible assumption that the decision to close occurs before the actual closure, thus we use the last full year of data (time=t) to explain closure in the subsequent year (t+1).