We read with interest the meta-analysis by Mezuk et al. (1) on the bidirectional relationship between diabetes and depression. In their report, the authors examined whether having depression increases risk of type 2 diabetes and the inverse relationship, namely, whether having type 2 diabetes increases risk of depression. The authors concluded that depression was associated with a 60% increased risk of developing type 2 diabetes but that there was only a modest (15%) increased risk of depression in individuals with type 2 diabetes. While we welcome this report, we would argue that there are a number of important methodological and conceptual issues that qualify the conclusions that can be drawn from this meta- analysis.
First, when considering type 2 diabetes as a risk factor for depression, Mezuk et al. only included studies that had excluded prevalent cases of depression at the first measurement (exclusion criterion 4). Although the use of this criterion was not explicitly justified, we can assume that this was meant to allow conclusions to be drawn about the number of new cases of depression. However, excluding prevalent cases of depression at first measurement does not guarantee that participants did not suffer from depression at some point in the past. Depression often is a highly recurrent condition, and past depression is the most important risk factor for future depression (2). Therefore, interpretation of the results of this meta-analysis should include the acknowledgment that participants may have an increased (but unknown/unmeasured) vulnerability to depression. This is likely to be stronger in people with diabetes because the prevalence of depression in diabetes is increased in this population (3). With the exception of that by Brown et al. (4), none of the other studies in this meta-analysis excluded people with a history of depression. Therefore, conclusions concerning the role of diabetes as a risk factor for new cases of depression seem premature.
On the other hand, excluding cases of current depression at first measurement may well limit any conclusions that can be made with regard to the role of diabetes as a risk factor for recurrent depression. Because there is some evidence that recurrence of depression in people with diabetes occurs mainly in those who have had depressive disorder in the past (5), excluding people with a vulnerability to depression may underestimate the effect of type 2 diabetes as a risk factor of depression.
Given the uncertainties in the current literature, we cannot agree with the conclusion of Mezuk et al. (1) that research should “move from epidemiologic investigations that have established this [between diabetes and depression] association and begin the process of empirically testing causal hypotheses.” Rather, we would argue that better designed epidemiological studies are still required in order to establish the true incidence of depression in people with diabetes and the factors related to its recurrence in people with and without past depression.
Acknowledgments
No potential conflicts of interest relevant to this article were reported.
