S1. SUICIDE RELATED BEHAVIORS IN A SCHIZOPHRENIA SAMPLE

Abstract

Background: Lifetime risk of completed suicide and suicide attempt are high in people with schizophrenia. We assessed the prevalence and predictors of suicidal behaviors in a large sample of schizophrenia cases and tested the predictive power for suicide attempts of non-suicidal self-injury (NSSI), mood disorder, suicidal ideation, and certain clinical characteristics of schizophrenia. Methods: Data on suicidal behaviors were abstracted from medical records in 626 participants diagnosed with schizophrenia or schizoaffective disorder. This abstraction involved binary coding for facets of suicidal ideation, single suicide attempt (“patient took all medication including blood pressure medication”), and multiple suicide attempt (“patient jumped out of a moving car, jumped off a bridge, overdosed, and attempted to hang himself”). We coded for mild, moderate (“patient cut himself and placed hands in hot water to burn himself”), and severe (“patient cut tendon in arm and required 100 stitches”) non-suicidal self-injury. Risk factors such as history of drug abuse, history of mood disorder, age at prodromal onset, and specific psychotic symptoms were also coded. We used logistic regression analyses to test the independent and overlapping predictive associations of NSSI, presence of mood disorder symptoms, suicidal ideation, with suicide attempts. All analyses controlled for age, sex, and race. Results: We found a history of suicidal ideation in medical records for 75% of participants and of suicide attempt in 44%. Multiple attempts were found in 21%. Additionally, 30% of our sample had a history of NSSI, and 59% of those with a history of NSSI attempted suicide at least once. Prodromal age was a significant predictor of suicide attempt in our sample (p = .0006, OR = .92). In a model simultaneously considering suicidal ideation, NSSI, and mood disorder history, suicidal ideation was an independent predictor of attempt (p = 1.5 E-11, OR = .20), as was NSSI (p = 8 E-6, OR = .42). Although significant when modeled alone, history of mood disorder was not an independent predictor of suicide attempt when modeled with NSSI and suicidal ideation (p = .34, OR = .84). When looking at symptom types, command hallucinations significantly predicted suicide attempt in our sample (p = .01, OR = 1.65). Other symptom variables, including history of disparaging auditory hallucinations and persecutory delusions were not associated with suicide attempts in these data. A two-group comparison of suicide attempt in schizophrenia and schizoaffective disorder was significant (p = .01, OR = .56), with higher rates of suicide in the schizoaffective group. Discussion: Suicidal ideation and attempt are very common in our sample. A documented history of suicidal ideation and non-suicidal self-harm were strongly associated with suicide attempt. Additionally, earlier prodromal age of onset and command hallucinations were found to be a significant risk factors. While mood disorder alone predicted attempt, it was not independently significant in a model that also included ideation and NSSI.