While the ability to prevent suicide is far less advanced than the prevention of heart disease, in my editorial the analogy highlighted the need for a multifaceted approach to anti-suicide strategies. I made the point that a single preventive measure would not be effective in reducing suicide mortality, as evidenced through the prevention of other types of death such as ischaemic heart disease. In the case of suicide, for example, the worldwide optimal treatment of depression would bring only a minimal reduction in suicide rates (further details available from the author upon request). None the less, fighting depression is generally perceived as the K constant of suicide prevention in existing national strategies. This happens despite growing evidence substantiating a much reduced life-span risk for suicide in depression than that reported in earlier investigations (Reference Bostwick and PankratzBostwick & Pankratz, 2000). Given the complexity of its pathways, the prevention of suicide, like the prevention of many types of death, requires a combination of approaches, such as public and medical education, promoting community connectedness, controlling access to means, early identification and intervention, etc.

It is certainly true that risk factors for suicide are unstable and may change over time (Reference De LeoDe Leo, 2002), but probably more important is the (mostly unexplored) interaction between risk and protective factors. This is the really crucial issue in suicide prevention (by the way, protective conditions of course counteract also the risk of ischaemic heart disease: the Mediterranean diet and omega-3-fatty acids have already convincingly underlined the role of local differences in mortality rates). And this recalls another important point raised by Dr Ravi Shankar, which refers to the local (cultural/traditional) specificity of suicidal behaviour. In countries such as China, risk factors for suicide are not dissimilar from those of Western countries — what varies is their ranking in terms of importance and expressivity (Reference Phillips, Yang and ZhangPhillips et al, 2002). Furthermore, it is well-known that within the same country there may be contiguous areas with largely differing suicide rates and that the same risk factors may operate differently in different social contexts.

To identify the exact components of a multifaceted prevention programme, tailored to local characteristics, greater knowledge of risk and protective factors is needed for both the psychiatric and general populations. Prevention of suicide is currently based on scant evidence. Therefore, I fully agree with Dr Ravi Shankar's view that more sound research is required. Prevention must be grounded in evidence if it is likely to have an effect on suicide mortality.