Predicting psychiatric admission from an emergency room: Psychiatric, psychosocial, and methodological factors

The determinants of psychiatric hospitalization, especially nonspecific issues, have been studied extensively. The methodological problem of correlational non-cross-validated find-ings are seen as contributing to our uncertainties about the critical issues involved in the decision to admit. A cross-validated multiple regression analysis of determinants of hospi-talization in an inner-city municipal hospital revealed two determinants of hospitalization: Severity of schizophrenic symptoms and active suicidal and/or homicidal ideation. They accounted for one third of the variance without shrinkage. Quasi-experimental designs may be the most efficient means of studying the remaining unexplained variance.