Evaluating suicide prevention: various approaches needed

Existing literature on the effects of suicide prevention is almost entirely confined to non-controlled, non-randomised and mainly retrospective studies. These studies do not take into consideration potential biases in selection of patients for treatment: for example, the impact of non-compliance with medication or psychotherapy, multiple diagnosis, or multiple medication (1).

Another problem in these evaluations is the short follow-up period (2). This is especially relevant in individually oriented pharmacological studies, but also in other studies, if a placebo effect is suspected. The longer the follow-up period, the better it can be ascertained whether any favourable development during the intervention is due to the placebo effect or to the persistent, if declining, effects of the intervention itself.

In evaluations of suicide-preventive programmes, an experimental or naturalistic approach can be used. The classic randomised experimental design is a benchmark that is often hard to apply, due to practical and ethical reasons, in studies of the effects of suicide prevention. Unfortunately, the merits of other designs, such as quasi-experimental ones, are often judged in terms of how far they resemble experimental design, although it is known that establishing and maintaining controlled conditions is difficult in population-oriented suicide-preventive studies conducted in natural field settings, such as schools (3), workplaces (4), etc.

Granted, outcome evaluation is crucial. But process assessment is important as well. This kind of evaluation allows us to identify elements that succeed in producing the expected outcomes. Thus, we can also improve tailoring of future interventions.

In both naturalistic and experimental suicide-preventive studies, insight into the personal and professional characteristics of the persons performing interventions or providing treatment is vital (5). These characteristics have a major bearing on the outcome of preventive intervention. Unfortunately, variables of this kind are not systematically evaluated. Moreover, management of control groups in pharmacological, psychotherapeutic and population-based interventions alike is often described not at all or very inadequately.

Insufficient statistical power in studies of suicide-preventive effects is a widely acknowledged problem, but one that has been poorly tackled, owing to the various obstacles to large-scale studies. We therefore need to use results from meta-analyses based on small, non-experimental sets of material, with non-significant differences between various treatments. This is because meta-analyses of an aggregate set of material have considerably greater statistical power for demonstrating significance and, perhaps above all, because they allow more convincing 95% confidence intervals (CI) to be reported than for individual studies (6).

However, meta-analytical calculations require several small sets of studies using comparable methodology. The World Health Organization, European Office (WHO EURO) Network on Suicide Research and Prevention (7, 8) and the WHO's worldwide SUPRE (SUicide PREvention) initiative (9), with network members on the five continents, constitute a stable basis for tailoring multicentre studies of suicide-preventive activities on both the individual and the population level. In the long term, only studies performed in several countries and at several centres, applying the same methodology and knowledge in the interpretation of results, can provide answers as to which strategies and methods are effective means of preventing suicide.

Studies of natural experiments also help to document the effectiveness of suicide-preventive measures. In the former Soviet Union, for example, the greatest suicide-preventive impact for males in the 20th century was demonstrated during perestroika: the fall in alcohol consumption brought about by Gorbachev's anti-alcohol policy was followed by a 40% decline in suicide among men in the labour force in all 15 Soviet republics. In Europe, during the same period, the fall in the male suicide rate was as small as 3% (10).

However, the dominant problem in the evaluation of various suicide-preventive programmes, as I see it, is neither researchers' reluctance to tailor randomised, controlled, prospective studies with a sufficient number of subjects in intervention and control groups, nor the lack of integration of public-health measures and clinical practice. The main problem is the shortage of financial resources. This is because suicide has been a neglected problem (11), although it is among the foremost single causes of death in Europe in the 15-44 age group and the one involving the most years of life lost, and despite rising suicide rates among males aged 15-19 in Europe.

Obstacles to funding of suicide research and preventive research are rooted in the powerful taboo on suicide and the uncomfortable feelings aroused by the topic not only in laymen, but also among professionals. Suicide is associated with shame and guilt. Silence, ambivalence and neglect impede an open scientific approach to the problem (1). Ambivalent and negative attitudes towards suicide and its prevention – with prevention regarded as impossible and suicide as predestined – are widespread among politicians, health policy makers and administrative bodies responsible for approving grants. Philosophers', anthropologists' and historians' inputs, which may assist us in understanding the taboos and attitudes in our society that hinder suicide prevention, are therefore highly welcome.