Les variables associées à la décision d’hospitalisation

Abstract

The psychiatric assessment conducted in a psychiatric emergency service and the subsequent admission decisions have major psychological, physical and fiscal effects on patients, families and the community. Despite the importance of the task, there is no agreement on the areas of clinical assessment the psychiatric emergency service should focus on. The purpose of this study was to identify important variables that influence admission decisions in psychiatric emergency services in order to provide information facilitating the development of assessment guidelines for use in the psychiatric emergency services. Residents at the psychiatric emergency service rated prospectively 326 patients (sex ratio: 1.91, mean age: 34.55 years (S.D. = 11.68), unemployment: 74.5%, low educational level: 60.7%; low socioeconomic level: 77.9%, married: 29.4%, rural residence: 51.8%, of the 326 patients, 218 (66.9 %) were hospitalized) on 12 clinical and demographic variables, including gender, marital status, socioeconomic level, unemployment, danger to self, danger to others, referred to by psychiatric physicians, self referred, psychotic signs, current diagnostic of major mental illness, involuntary consultation, and previous hospitalisation. First, a chi square analyse was conducted on the 12 variables to determine whether they significantly differentiated patients admitted from those released. A p value below 0.05 was used. Next, a binary logistic regression procedure was employed to develop a model for an admission depending on five variables. The SPSS software program was used for these analyses. The model's performance was evaluated in two ways: the proportion of cases correctly classified and the level of signification of the Hosmer–Lemeshow test. A logistic regression produced a model with five variables that significantly predicted admission, in order of importance they were danger to self, psychotic signs, current diagnostic of major mental illness, involuntary consultation, and danger to others. The model correctly classified 93.9 percent of the cases. The level of signification of the Hosmer–Lemeshow test was 0,724 (chi-square = 5.307, ddl = 8). A review of results of similar studies in other populations shows that these findings are supported in the literature. As a result, these five variables should be included in admission Guidelines for emergency psychiatric services staff. In this study, the admission decisions were usually justified.