Ecological studies and the big puzzle of falling suicide rates

In many industrialised countries, suicide rates have declined substantially in recent years (1). Numerous studies have identified an inverse relationship between antidepressant sales and suicide rates in these countries (2). The many papers have differed on their interpretation of the results, ranging from a firm belief that the association is based on certain causality to the denial of any relationship and trying to prove that the findings are purely spurious. The latter group has had their problems when there are so many replicates of this finding. This does not, however, mean that the first group has proven their case. We do not know whether the antidepressants caused the fall in the suicide rates. The decrease could have been related to other phenomena connected to the introduction of SSRIs in the beginning of the 1990s. These could include increased treatment optimism, doctor and patient awareness through information from the pharmaceutical industry and the availability of diagnostic tools such as Beck s depression scale and the Montgomery–Åsberg depression rating scale. In the current issue of Acta Psychiatrica Scandinavica, Kapusta et al. present an ecological study on the inverse relationship between the density of psychotherapists and the suicide rate in Austria (3). It puts forward an alternative explanation for the fall in suicide rates; or rather a complementary explanation to that offered by the many publications on antidepressant sales and suicide rates. Maybe a rise in the number of practising psychotherapists could have contributed to a decline in suicide rates? Kapusta et al. also suggest that the density of psychotherapists might be a proxy for the measurement of improved psychiatric health services rather than the direct cause of lower suicide rates. The beneficial trend may be a result not only of an increase in practising psychotherapists or antidepressant drugs, but of a broader spectrum of health care services and interventions working together. Suicide is an important public health issue, but prevention efforts have been plagued by the lack of a firm evidence base due to several challenges in research. First, we have the complex relationship between completed and attempted suicide. These two closely related phenomena have, partly, different risk factors. Another challenge is that suicide is, fortunately, a rare event even in high-risk populations. Randomized trials will always have problems with power in the statistical analysis, together with ethical constraints, and have proved to be of little value in research into completed suicides. Finally, we have the problem of confounding by indication. Those receiving drugs and psychotherapy are the sickest groups. This makes it difficult to establish a protective effect even in case–control studies. So, ecological studies are needed in this field. However, ecological studies also pose several problems for our interpretation of any association shown. Researchers will naturally look for causality, but may have difficulties with this in ecological studies. In an earlier study (J. G. Bramness, unpublished data), we found that the suicide rates in Norway fell simultaneously with an increase in the sales of angiotensin II receptor blockers. Using a simple regression analysis, more than 80% of the variance in suicide rate was explained by sales of angiotensin II receptor blockers! In ecological studies we need to explore the theoretical basis of a relationship, if we want to suggest causality. Our knowledge of the field does not support a claim of a causal relationship between sales of angiotensin II receptor blockers and suicide rates. In the case of Kapusta et al., we are on much safer ground. Psychotherapy is known to be beneficial in the treatment of depression (4). That treatment of depression may prevent suicide can be considered a very well founded suicidological hypothesis because of the high prevalence of depression among completers (5). This has been suggested prior to the many ecological studies (6). This theoretical foundation is shared by Kapusta et al. s work and earlier research on antidepressants and suicide (2). One problem often overlooked in ecological studies is the need to use the adequate statistics. Many earlier studies have fallen into some of the obvious pitfalls. It is not enough to point out a Acta Psychiatr Scand 2009: 119: 169–170 All rights reserved DOI: 10.1111/j.1600-0447.2008.01318.x Copyright 2009 The Authors Journal Compilation 2009 Blackwell Munksgaard