Is good school performance a risk factor for suicide in psychoses?

Alaräisänen et al. (1) have analyzed the relationship between good school performance and risk of suicide in patients with psychoses. Long-term prospective cohort studies, with standardized diagnostic assessment and reliable data collection, are indeed few in this area. We have a few observations. First, selection of the exact cut-off point demarcating good school performance (defined here as mean scores 8.5 and over ) may have influence on results. Indeed, a previous publication from the same cohort (2) had used the cut-off mean score as 9. When numbers are small, even a single extra subject being counted as having good school performance can appreciably influence relative risk statistics. Instead, plotting of individual mean school scores of subjects who committed suicide (among psychotic and non-psychotic subjects) on a scatter diagram could have given a fair idea of the link between the two variables (suicide and school performance) without any dispute over a cut-off point of 8.5 as being possibly arbitrary. As mentioned under results (1), the group other psychoses had the highest hazard ratio. This group included 14 subjects with a primary mood disorder and four with schizoaffective disorder. Mood disorder is a well-known risk factor for suicide. So, we also wonder if this might have contributed to the high hazard ratio in the other psychoses group and, inter alia, in all psychoses group as well. A recent important meta-analysis of 29 studies has identified several risk factors for suicide in schizophrenia such as previous depressive disorders or affective symptoms, fear of mental disintegration, previous suicide attempts, drug misuse, hopelessness, agitation, poor adherence to treatment, and recent loss (3). It might be a good idea to look at some of these parameters in detail from the case notes retrospectively, and see if some of these known risk factors interact with school performance as a risk factor. This issue may be important because good school performance , in reality, might be a proxy variable in the pathway of risk and causality. For example, in an early important nested case–control study, although college education appeared as a significant risk factor for suicide in schizophrenics in bivariate analysis, it no longer remained significant in a multivariate logistic regression analysis (4). In the end, we must thank the authors for the thoughtprovoking paper based on a long-term prospective cohort design. We hope our suggestions might help to further analyze and interpret the data critically.