We agree that although observational studies have shown that depression is associated with subsequent impairment in health-related quality of life in coronary heart disease, intervention studies have failed to provide convincing proof that treating depression improves this outcome. Previous intervention studies have not addressed this question satisfactorily because the SADHART study (Glassman et al, 2002) was not sufficiently powered to demonstrate the efficacy of antidepressants in coronary heart disease and the ENRICHD study (Berkman et al, 2003) did not anticipate very high rates of spontaneous remission of depression or unplanned prescription of antidepressants in the control group. The results of these trials, however, together with our own results are valuable for planning future treatment trials.

We also agree that there are many unanswered questions relating to the nature of the association between depression and negative outcomes in coronary disease. As mentioned by de Jonge & Ormel, the timing of the onset of depression (Reference Dickens, Percival and McGowanDickens et al, 2004a ), the specific aspects of depression or anxiety that are associated with poor outcome and the possibility of vulnerable sub-populations of patients (such as those without social support) (Reference Dickens, McGowan and PercivalDickens et al, 2004b ) require further investigation. Furthermore, whether the association between depression and negative outcomes in coronary disease is the result of residual confounding by severity of heart disease (Reference Dickens, McGowan and PercivalDickens et al, 2005) remains unsolved. Further research is required to address these questions, although it is likely that most will only be convincingly resolved through intervention studies.