In their recent paper, Paterniti et al (Reference Rosenman2002) reported that depressive symptoms predict cognitive decline over a 4-year period. This is a well-designed and well-written study that replicates a previous finding from similarly well-designed studies. Negative findings on this question, however, are also common in the literature, including a report from the same French group a few years ago (Reference La Vecchia, Lucchini and LeviDufouil et al, 1996). I would like to point out some relevant issues overlooked by Paterniti et al.

First, I find it unfortunate that the paper cites few negative reports, with no mention in the discussion of the many longitudinal studies that have reported no association between depressive states and subsequent cognitive decline (Reference La Vecchia, Lucchini and LeviDufouil et al, 1996; Reference Prince, Lewis and BirdPrince et al, 1996; Reference De LeoCervilla et al, 2000). It is particularly surprising that Paterniti et al quote the study by Chen et al (Reference Durkheim, Spaulding, Simpson and Simpson1999) as reporting that ‘depressive symptoms are predictive of cognitive decline’, when in fact they found that dementia predicted the onset of depressive symptoms but not the other way round.

Second, it is regrettable that Paterniti et al overlooked the only study to date addressing the very same question but for a considerably longer follow-up period (Reference De LeoCervilla et al, 2000). Longer follow-up periods could help to distinguish between psychopathology shared by depression and dementia (e.g. difficulties with memory and concentration, or apathy), as pointed out by a previous study by some co-authors of Paterniti's paper (Reference La Vecchia, Lucchini and LeviDufouil et al, 1996). Indeed it could be argued that if depressive symptoms have a real capacity to predict cognitive decline, the latter should be expected to become more apparent as the study's follow-up period lengthens. This, in fact, has not happened in our cohort (Reference Prince, Lewis and BirdPrince et al, 1996; Reference De LeoCervilla et al, 2000) and I believe this adds potentially unique information to Paterniti et al's discussion.

Finally, it is also regrettable that Paterniti's group did not explore the repeatedly reported interaction with gender in considering whether depressive symptoms predict cognitive decline (Reference Prince, Lewis and BirdPrince et al, 1996; Reference De LeoCervilla et al, 2000). Indeed, the latter studies have reported that if an association exists between depressive symptoms and cognitive decline, this seems to be the case in men only (Reference De LeoCervilla et al, 2000), or in men of above-median premorbid IQ (Reference Prince, Lewis and BirdPrince et al, 1996).