We found a substantially higher 43-fold risk of suicide in women and 21-fold in men withan alcohol-related admission and a pre-existing psychiatric diagnosis (Table 3). Patients admitted with no known psychiatric morbidity were still found to have a substantial risk compared to subjects not admitted, at 8-fold and 24-fold in men and women, respectively.Model checking of the Schoenfeld residuals found that none of the models violated the proportional hazards assumption.