Association between dietary patterns and adult depression symptoms based on principal component analysis, reduced-rank regression and partial least-squares.

Abstract

There have been inconsistent findings on the association between dietary patterns and depressive symptoms (DepS). In addition, studies have used single analysis methods to identify dietary patterns. In the current study, we aimed to determine the association between dietary patterns, derived by principal component analysis (PCA), reduced-rank regressions (RRR) and partial least-squares (PLS), and DepS among adults using a cohort study in Australia. We examined a total of 1743 study participants (≥24 years, 48.9% males) using cross-sectional and longitudinal data from the North West Adelaide Health Study (NWAHS). The Center for Epidemiological Studies-Depression (CES-D) scale was used to assess DepS and a score ≥16 was considered as having depression. Dietary data were collected using a food frequency questionnaire. Eicosapentaenoic acid (EPA)/Docosahexaenoic acid (DHA), folate, magnesium (Mg) and zinc (Zn) densities were chosen as the response variables for RRR and PLS analyses. Dietary patterns were identified by PCA, RRR and PLS. Odds ratios (OR) and 95% confidence intervals (95% CI) were estimated across quartiles (Q) using log-binomial logistic regression to assess the association between dietary patterns and DepS. Sensitivity analyses, including a longitudinal association between dietary patterns and DepS among 859 participants, were performed. Multiple imputation was performed to investigate the effect of missing data on the estimates. In this study, 16.9% (14.2% in men and 20.8% in women) of the participants had DepS. We retained two, four and four dietary patterns captured by PCA, RRR and PLS respectively. The 'prudent' pattern determined by PCA [ORQ4VsQ1=0.57; 95% CI: 0.35, 0.92] and PLS [ORQ4VsQ1=0.66; 95% CI: 0.43, 1.00] together with the 'typical Australian' pattern determined by RRR [ORQ4VsQ1=0.60; 95% CI: 0.40, 0.90] were inversely associated with DepS whereas the 'western' pattern derived by PCA [ORQ4VsQ1=2.04; 95% CI: 1.12, 3.68] and PLS [ORQ4VsQ1=1.62; 95% CI: 1.05, 2.50] was positively associated with DepS. In the longitudinal analysis, the 'prudent' pattern determined by PCA [ORQ4VsQ1=0.52; 95% CI: 0.25, 1.09] tended to be inversely associated with DepS whereas 'western' patterns determined by PCA [ORQ4VsQ1=3.47; 95% CI: 1.37, 8.78] and PLS [ORQ4VsQ1=2.47; 95% CI: 1.24, 4.91] were positively associated with DepS. We found that a dietary pattern characterized by high intakes of fruits, vegetables, medium fat dairy, nuts, legumes, and fish was inversely associated with DepS in this population-based study. Contrary to this, a dietary pattern characterized by high intakes of processed and red meat, fast foods (snacks and takeaway foods), soft drinks, white bread and high-fat dairy products were significantly associated with DepS. Multiple imputation and sensitivity analysis identified similar patterns of association between dietary pattern and DepS. The findings indicate that the 'western' pattern was consistently associated with an increased risk, and the 'prudent' pattern tended to be associated with a reduced risk of DepS. This suggests that dietary interventions may assist with the treatment of DepS. However, current evidence on the impact of diet on DepS should be supported using further longitudinal studies with extended follow up, larger sample sizes and repeated measures.