Response

Kawada^{(Reference Kawada1)} commented on our article on bidirectional associations between intake of food groups and depressive symptoms in the InCHIANTI study^{(Reference Elstgeest, Visser and Penninx2)}. Specifically, Kawada contemplated on our non-significant association between coffee and tea intake and subsequent depressive symptoms, whereas previous studies have shown associations.

In the direction from diet to depressive symptoms, we only found significant associations for fish and shellfish and for sweet foods, not for the eleven other food groups, including the group coffee and tea. We referred to the meta-analysis of Grosso et al.^{(Reference Grosso, Micek and Castellano3)} that showed that coffee consumption was significantly associated with decreased risk of depression, also when only prospective studies (n 3) were examined (relative risk highest v. lowest category: 0·88, 95 % CI 0·79, 0·99). The meta-analysis of Wang et al.^{(Reference Wang, Shen and Wu4)}, which included fewer but mainly the same individual studies as Grosso et al.^{(Reference Grosso, Micek and Castellano3)}, logically showed an inverse association as well; however, when stratified by region, the pooled association of two European studies was not significant. In our discussion, we also referred to one of the largest prospective studies on this link that showed a weak association^{(Reference Guo, Park and Freedman5)}. A more recent prospective study showed a significant inverse association for the consumption of at least four cups of coffee/d compared with less than one cup of coffee/d; yet, they found no linear dose–response association between coffee consumption and the risk of depression^{(Reference Navarro, Abasheva and Martínez-González6)}.

Regarding tea consumption, Kawada^{(Reference Kawada1)} referred to two older and cross-sectional studies on tea consumption and depressive symptoms^{(Reference Niu, Hozawa and Kuriyama7,Reference Hintikka, Tolmunen and Honkalampi8)} but did not refer to more recent meta-analyses. One study of Grosso et al.^{(Reference Grosso, Micek and Castellano3)} showed that evidence for an association between tea consumption and depression risk is less strong compared with coffee. In contrast, the meta-analysis of Dong et al.^{(Reference Dong, Yang and Cao9)} found a significant inverse association; however, they did not include the large study of Guo et al.^{(Reference Guo, Park and Freedman5)} that did not observe a tea–depression association.

There are several potential reasons for our null association. First, the coffee and tea intake of our study population was moderate (mean coffee + tea intake: about 112 (sd 78) g/d or about one cup/d). This is much less than the highest categories (three or four cups/d) that were associated with depression risk^{(Reference Grosso, Micek and Castellano3,Reference Guo, Park and Freedman5,Reference Navarro, Abasheva and Martínez-González6)} . Second, we summed coffee intake and tea intake and did not examine them separately. A third reason might be that for our Italian study population of mainly older adults, other factors might be more important in relation to depressive symptoms than the intake of most food groups.

In our study, we used residuals of coffee and tea intake, both as continuous variables as well as in quartiles. The latter was chosen to examine potential non-linear associations, as Kawada^{(Reference Kawada1)} also suggested. For both fish and sweet foods, the association was strongest for the highest quartile, while the middle quartiles had comparable regression coefficients. We observed this non-linearity also for coffee and tea but in the unexpected direction (higher intake associated with more depressive symptoms); however, the association was not statistically significant, nor was the P _{for trend}.

Future studies on the association between coffee/tea and depression are needed. Well-designed, prospective studies can provide more insight, if they include a large consumption range, distinguish the different types of coffee and tea and include different ethnicities. To ultimately show causality, randomised controlled trials are needed to study the potential beneficial effects of coffee and tea consumption on depressive mood.