Authors' reply

As highlighted by Mahendran, depression in late-life is often accompanied by medical comorbidity. ¹ Owing to the lack of information on medical comorbidity in the studies included in our meta-analysis, we could not evaluate to what extent the differences found could be explained by an overlap in somatic symptoms of depression and medical comorbidity. Of the 11 included studies, only 4 reported on medical comorbidity. As the study sample of Koenig et al consisted of a medical in-patient population, medical comorbidity was present in both younger and older people with depression. ² Moreover, age-related differences in the phenomenology of depression persisted after adjustment was made for medical comorbidity. In the studies of Brodaty et al, Gournellis et al and Tan et al, the levels of somatic comorbidity were indeed higher in older compared with younger people with depression. ^3–5 We did acknowledge that age-related somatic comorbidity may have caused some overlap with somatic symptoms of depression, explaining part of the age-related differences in the phenomenology of major depression. On the other hand, somatic comorbidity may also have an impact on the phenomenology of late-life depression, apart from the overlap of symptoms. Unfortunately, in our meta-analysis it was impossible to unravel potential mediating effects.

As noted in the introduction section, we agree with Mahendran that sociocultural and psychological factors related to ageing may influence the clinical presentation of depression in late life. In this meta-analysis, however, we aimed to investigate whether age-related differences in the phenomenology of depression exist at all. The question as to which of the biological, psychological or sociocultural factors may cause age-related differences, and how they might modify the phenomenology of depression in late life, needs further examination.

An important issue raised by Mahendran concerns the distinction between clinimetrics and psychometrics. Of course clinicians cannot rely on existing psychometric rating scales alone when making clinical decisions. However, this distinction does not affect the overall results of our meta-analysis. Age-related differences in the clinical manifestation of major depression were investigated to start with. Going one step further, phenomenological differences corresponding to differences in prognosis, treatment and determinants, need to be investigated in future research, all of them important for clinical reasoning. Furthermore, this may not be so much an issue of clinimetrics as opposed to psychometrics, but a consequence of the inadequacy of the categorical DSM-IV classification system, leading to extensive comorbidity and diagnostic heterogeneity which impedes the search for determinants. ⁶ As depression is a highly heterogeneous disorder, we focused on major depression to enable the search for age-related differences. Moreover, because no commonly used clinimetrically based model exists, we chose to use the most appropriate instrument currently available.