Predicting suicide in patients who have received acute psychiatric inpatient treatment - the longitudinal and...

Suicidality (suicidal ideation and -plans (SI) and deliberate self-harm (DSH)), has been extensively studied in different patient populations. Fewer studies have examined predictors of suicide within an extensive longitudinal frame, which is problematic since findings from studies including participants with non-lethal suicidal behavior are not easily generalizable to patients that complete suicide. We aimed to address this gap by examining demographic and clinical factors as predictors of suicide in a longitudinal prospective cohort study of 7,000 consecutively admitted patients. Mean observation time was 4.1 years. Male gender, more frequent admissions during the follow-up period, and having SI, DSH and substance-induced psychosis at the final admission before suicide, directly predicted shorter time to suicide. Shorter time to suicide was indirectly predicted by borderline personality disorder (BPD) via DSH and by psychotic depression and neurotic disorders via SI. In contrast, longer time to suicide was indirectly predicted by substance-induced psychosis and psychotic disorders via reduced probability of SI and DSH at the final admission. Patients diagnosed with BPD had more frequent admissions and higher levels of SI and DSH across admissions, which predicted shorter time to suicide via increased risk of DSH at the final admission. Similarly, patients diagnosed with neurotic disorders had a higher level of SI and DSH across admissions, with shorter time to suicide via SI and DSH at the final admission. Findings suggest different pathways to suicide for different clinical subgroups, and this is discussed in relation to current theoretical models that emphasize suicide as a transdiagnostic phenomenon.