Suicide is a fatal outcome of psychiatric illness. It is a serious and demanding task for psychiatry to prevent it. There is no consensus, however, either regarding what causes suicide, or how to prevent it. This may depend on the particular severities, or even impossibilities, associated with suicide research. The cases for study can only be identified post mortem, and ‘prevented suicides’ cannot be identified at all. Furthermore, suicide is a rare event, which makes conclusive randomized clinical trials (RCTs) impossible. Even meta-analyses of RCTs have far from the sufficient statistical power to demonstrate a possible 50% difference in suicide rate between the intervention group and the controls. High-risk groups are often studied to improve the statistical power, but placebo-controlled trials with suicide as the outcome are still impossible by ethical reasons. Individuals who have ‘attempted suicide’ form a heterogeneous group, but on average they are at high risk for repeated attempts and suicide. Studies of various interventions have, however, hardly been able to demonstrate any differences in repetition of attempts, and of course much less what regards differences in suicide (1).
There are, however, other ways than RCTs to gain knowledge of suicide, i.e. naturalistic epidemiological studies, with suicide either as an outcome measure, or as defining the study group. Depending on design such studies may be ‘almost conclusive’, but as a rule, the results should be interpreted with caution. Naturalistic studies of good quality tend to be large, prospective, controlled, individual based, and testing a priori hypotheses. Studies utilizing retrospective or uncontrolled or aggregated data may not necessarily be inferior, however.
To date the most important single suicide study may be the retrospective uncontrolled descriptive investigation of 134 suicides in St Louis by Eli Robins and co-workers (2). This investigation, which has been replicated over and over again in other countries and time periods, clarified that suicide practically does not occur without the presence of mental illness, most commonly depression, and thereafter alcoholism. It may be a logical continuation to study the treatment received by depressed or alcohol dependent and potentially suicidal individuals. Studies by our group and others, have demonstrated that depressed individuals who have committed suicide are seldom treated with antidepressants, leaving an open area for interventions (3). The rapid increase in the use of antidepressants during the last 15 years may be considered as a major intervention in the depressed population. I presented in this journal in 2000, based on previous studies, the hypothesis that the radical decline in suicide (−25%) seen during the same period actually was the consequence of this intervention (4). This hypothesis has since then been supported by nine subsequent studies, and not supported, nor falsified, by two (4, 5). Treatment of depression is the suicide preventive intervention, at the population level, that to date has by far the strongest evidence base (3, 6–8). In patients with bipolar disorder, there may be even stronger evidence that lithium prophylaxis prevents suicide (9).
The Gotland Study was a project where primary care physicians were intensively educated about depression. This was followed by increased use of antidepressants, and a decreased number of suicides. The subsequent year, however, these possible effects of the project had disappeared (10). In this issue of the journal Henriksson and Isacsson report a similar project from another Swedish county, Jämtland (11). This project has been going on continuously since it started in 1995. Since then, the use of antidepressants has increased more in Jämtland than in Sweden in general, and, the number of suicides has decreased more steeply than in the rest of Sweden. This result supports the validity of the Gotland study, and adds to the evidence for that the treatment of depression is a key to suicide prevention.
Further, Fialko et al., in this issue, report an analysis of suicidal ideation in 290 individuals with schizophrenic, schizo-affective, or delusional disorders (12). The cases were originally enrolled in an RCT of interventions aimed at preventing relapse in psychosis. One of the scales used was the Beck Depression Inventory. Its item 9 regards ‘suicidal ideation’. The authors found that suicidal ideation was related to low mood and anxiety, to negative self-esteem and beliefs, as well as to daily alcohol consumption, but not to hallucinations or delusions, except auditory hallucinations with negative content, and delusions of guilt. This result is not surprising considering our knowledge about suicides mentioned above. It gives support to the hypothesis that depression is a necessary factor for suicide (13, 14). Obviously, depression is not, however, a sufficient factor, but additional risk factors are required too (15, 16).
Oquendo et al. have reviewed prospective studies of suicide and attempted suicide in patients diagnosed with major depression or bipolar disorder, and found that the most robust risk factor was a previous suicide attempt (17). Interestingly, with increasing number of previous attempts, the risk for repeated attempts increases, while the risk of suicide seem to decrease. The authors interpret this as ‘suicide completers and attempters are distinct but overlapping cohorts’.
One useful method of obtaining data for epidemiological studies is to utilize large registers by record linkage. Søndergård et al. have combined data from Danish health registers with a large prescription database of prospectively entered records (18). They used this database for defining two cohorts of individuals as being exposed, or unexposed, to antidepressants. Among other results, they found that in parallel with the decrease in suicide in Denmark, the proportion of suicides who had been prescribed antidepressants has grown. A minimal number of suicides who all have had the chance to respond to treatment is of course the ideal.
These papers in this issue of Acta Psychiatrica Scandinavica form a major contribution to our knowledge about suicide. In the identification, treatment and monitoring of depressed individuals lies the key to suicide prevention.