Who is treated in psychiatric scatter beds in general hospitals? An imputational algorithm.

We imputed where mental, alcohol, and drug disorders are treated in general hospitals (GHs). The 1980 Hospital Discharge Survey, enhanced with other information, provided data. An algorithm based on multiple discriminant analysis and clinical assumptions imputed which GH patients were treated in a psychiatric unit (PU), a chemical dependency unit (CDU), or a general medical/surgical scatter bed (SB). SBs were used much less than expected. Over 60% of GH episodes occurred in specialized sites: PUs (63%), CDUs (15%), or specialty hospitals (22%). Most SB episodes occurred in GHs with no specialized unit (75%). Patients treated in scatter beds differed systematically from those in units of the same hospital and those in hospitals with no specialized units (e.g., less psychosis and psychiatric complexity, more physical comorbidity). The de facto system of assignment of disorders to hospital sites appears rational. Implications for public policy regarding inpatient treatment are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)