Résumé
Short periods of hospitalization for first episode psychosis are generally considered preferable, although research evidence is inconsistent. Clinical and social determinants of length of hospital stay (LOS) were examined in 121 admissions for first episode psychosis using standardized assessment measures, in five clinical units of a French psychiatric hospital. LOS varied from 4 to 371 days, and considerably between units. LOS was examined both as a continuous variable and dichotomized comparing short (<31 days) to long (>31 days) stays. In the multivariate analyses, change in antipsychotic medication and the unit head psychiatrist's preference for longer stays were significantly associated with both measures of LOS, indicating effects on LOS per SE and not only with respect to a threshold duration of stay. Of the clinical factors at admission, the only borderline significant association found was between the severity of negative symptoms and LOS on a continuum. Despite some justification for longer stays with respect to discharge conditions, the persisting association with the head psychiatrist's preference for long or short stays strongly suggests a need for greater evidence-based rationalization of practice.