The impact of seclusion on the treatment outcomes of psychotic in-patients.

The controversial practice of seclusion for the treatment of the acutely emotionally ill continues despite a lack of proven effectiveness. This retrospective cohort study examined the impact of seclusion alone and in combination with other important patient-centred treatment variables (medications), disease severity variables, psycho-social and demographic variables that best predicted outcome (as defined by length of hospital stay and reduced pathology at discharge) for a group of consecutively admitted psychotic in-patients. Comparisons of baseline and outcome measures for secluded and non-secluded cohorts were made via t-test analysis, and patient-related variables were entered as groups for stepwise regression analysis to determine those factors that could best predict outcome. Data were collected via a comprehensive chart review of 85 patients with psychosis, who were consecutively admitted over an 8-month period to a psychiatric in-patient unit of a general community hospital in southern Ontario, Canada. Bigelow's Psychiatric Symptom Assessment Scale (PSAS) was applied to the chart admission and discharge data to determine both initial pathology scores and pathology change scores. Investigator-developed global rating scales measured the early pharmacological and psycho-social variables of interest. Although seclusion itself did not appear to be associated with any more or less improvement in mental status at discharge, seclusion was associated with a 12-day increase in length of hospital stay, raising additional questions about its therapeutic effectiveness. Early medication response together with seclusion predicted 16% of the total variance of length of hospital stay (P = 0.0011) while baseline pathology, being married and being employed together predicted 66% of reduced pathology at discharge (P = 0. 00001). Secluded subjects experienced an earlier age of illness onset, were employed or full-time students, exhibited poorer attitude towards hospital, were more likely to be substance or drug abusers, received greater doses of chlorpromazine-equivalent medications, and had longer lengths of hospital stay than their non-secluded cohorts. Two predictive regression equations were formulated on the basis of these results. This study has raised several questions for further prospective investigation, has validated previous findings in the literature, and has produced new findings that may be of interest to administrators and clinicians providing quality care to this high-risk population.