Awareness of Disorder and Suicide Risk in the Treatment of Schizophrenia: Results of the International Suicide Prevention Trial

The methods of this study and the main findings concerning predictors of suicide events have been previously described in detail (11, 12). In summary, InterSePT is a 2-year randomized, open-label, blinded-rater trial that included 980 patients with schizophrenia (N=609) or schizoaffective disorder (N=371) who were recruited from 67 centers in 11 countries. Of the total group, 477 subjects received olanzapine, 479 received clozapine, 21 received no treatment, and three dropped out of the study before medication compliance could be verified. All patients were evaluated as being at high risk for suicide because of a suicide attempt or hospitalization to prevent suicide in the 3 years before inclusion or because of severe suicidal ideation with depressive symptoms or command hallucinations for self-harm in the week before inclusion. After providing informed written consent, the patients received weekly clinical monitoring and assessments for suicide-related events over 6 months and then at biweekly intervals for 1.5 years. Suicide events in both patient groups were defined as a suicide attempt or a hospitalization to prevent suicide but not ideation alone, and these judgments were made by blinded, independent evaluators of a suicide-monitoring board. Depression and awareness were assessed at baseline and weeks 24, 52, and 104 (or at endpoint, which was defined as a suicide event). Depression was assessed by using the total score (range=0–25) of the Calgary Depression Scale (13), and one item of this instrument was used to specifically assess hopelessness. Awareness of disorder was examined by using the Scale of Functioning (item 12, range=1–4), which assesses the level of awareness that the patient has concerning his or her psychiatric illness over the preceding 7 days (14). Statistical analyses of suicide monitoring with board-determined endpoints were conducted by using Cox’s proportional hazards regression models. Covariates included treatment medication, pooled country, sex, and age.

For the clozapine treatment group, mean awareness levels were 2.33 (SD=0.87) for schizophrenia and 2.69 (SD=0.87) for schizoaffective disorder. For the olanzapine group, mean scores were 2.53 (SD=0.88) and 2.67 (SD=0.84), respectively. Both treatment groups demonstrated improvement over the follow-up concerning awareness of disorder and depression, and the patients who made a suicide attempt had less improvement in depression (2.49 [SD=5.54] versus 5.58 [SD=6.02]; z=7.373, p<0.0001).

The results of a Cox proportional hazards analysis for main effects of baseline variables demonstrated that greater awareness significantly predicted time to suicide events (hazards ratio=1.17; χ²=4.38, df=1, p<0.04). As demonstrated by Table 1, the inclusion of depression at baseline in the model rendered awareness nonsignificant as a predictor of suicide behavior. However, along with male gender and younger age, depression remained a highly significant predictor of suicide events in the model (hazards ratio=1.16; χ²=175.59, df=1, p<0.0001). Parallel results were found when these analyses were limited to the hopelessness component of depression. No differences were observed by medication group concerning the relationships of depression or insight with suicidal behavior, and no differential effects were observed by diagnosis type. However, although baseline awareness was an absolute risk factor for suicide events, increases in awareness as a function of treatment (over the follow-up) were associated with a decreased risk of suicide events (hazards ratio=0.75; χ²=21.9, df=1, p<0.0001).

The present findings are consistent with past research indicating that baseline awareness levels are positively related to suicide risk (8, 9) but suggest that such effects are mediated by depression or hopelessness levels. Furthermore, important differences were found for the role of baseline awareness compared to treatment-related changes in awareness. Specifically, this latter variable was found to reduce the risk of suicide events over the 2-year observation period.

While the consequences of acknowledging one’s disorder may increase feelings of hopelessness or depression for many patients, increases in awareness related to treatment itself may be an important indicator of positive medication response. In this way, the present findings underscore the complexity of awareness in schizophrenia and schizoaffective disorder and indicate the need for careful clinical monitoring of both baseline and treatment-related changes in patient insight concerning his or her condition.

Received May 1, 2003; revision received Dec. 30, 2003; accepted Jan. 12, 2004. From the Psychopathology Research Laboratory, University of Bordeaux. Address reprint requests to Dr. Swendsen, Psychopathology Research Laboratory, University of Bordeaux, 3 ter, Place de la Victoire, 33076 Bordeaux, France; [email protected] (e-mail).