Strickland et al (Reference Strickland, Deakin and Percival2002) are to be congratulated for their Herculean task of testing the elegant hypothesis that cortisol is the biological link between stressful life events and the onset of depression. In a large community sample they found that cortisol was not elevated in currently depressed or vulnerable subjects, but was increased after recent life events, and hence concluded that ‘the hypothalamic—pituitary—adrenal [HPA] axis is sensitive to social stress but does not mediate vulnerability to depression’ (italics added). However, it is possibly premature to bury the ‘HPA-hypothesis’ under this evidence alone. The rationale for a community study is that it captures the effects of life events, but this exclusive approach misses the other naturalistic event in depression, namely admission to hospital. Thus, the proportion of patients in the ‘depressed’ group showing severe illness was less than 10% (9 out of 94), leaving more than 90% with mild (51%, n=48) to moderate (41%, n=37) illness. Given the well-replicated finding of elevated cortisol in major depression (Reference Harris, Borsanyi and MessariHarris et al, 2000) it would be helpful to provide a subgroup analysis for the severely depressed sample. Nevertheless, it is particularly interesting that there was significantly elevated cortisol in a subgroup of women who had experienced recent life events, whether or not they were currently depressed. It is possible that exposure to life events is a watershed in the evolution of depression, some people showing enhanced serotonin activity and avoiding depression, and others showing reduced serotonin function and becoming depressed, as suggested by animal studies (Reference McAllister-Williams, Anderson and YoungMcAllister-Williams et al, 2001).

Moreover, the investigation of state and trait abnormalities in major depression by Bhagwagar et al (Reference Bhagwagar, Whale and Cowen2002) shows that recovery from depression is accompanied by a restoration of HPA-axis function, suggesting that cortisol is a trait marker for depression. One inference of these combined findings is that life events preceding depression lead to raised cortisol and lowered serotonin, whereas life events not leading to depression are associated with raised cortisol and normal or even enhanced serotonin function (in what might be described as the ‘depression-resistant’ subgroup). In order to disentangle these apparently disparate findings it seems that a longitudinal study including a subgroup analysis of people with major depression is inevitable, and I am sure that Strickland's group will not disappoint us.