Data-driven dietary patterns in relation to psychological distress and psychological quality of life among community-dwelling older adults in Shanghai, China: a cross-sectional study
Evidence linking overall dietary patterns with psychological health in older adults remains inconsistent, particularly in urban China. We derived dietary patterns among community-dwelling older adults in Shanghai and examined their associations with multiple psychological outcomes. We analyzed cross-sectional baseline data from adults aged ≥60 years recruited at two community health service centers in Shanghai. Previous-month diet was assessed using a modified quantitative 25-item food-frequency questionnaire. Parallel analysis guided principal component analysis of 21 non-alcohol food groups; bootstrap resampling assessed pattern stability. The prespecified primary outcome was continuous K10 psychological distress; secondary outcomes were K10 ≥16 and the WHOQOL-BREF psychological-domain score. A reconstructed depressive-symptom composite and CD-RISC-10 were exploratory. Associations per 1-SD were estimated with HC3 linear or modified Poisson regression. Model 2, selected as the main parsimonious model in the revised analysis, adjusted for age, sex, district, and education. FDR correction was applied within four-pattern outcome families and globally across 12 main-model comparisons spanning primary and secondary outcomes. Among 246 respondents, 233 were included in pattern derivation and the main regression analyses. Four components explained 45.28% of total variance; the fourth, characterized by sugar-sweetened beverages, showed limited bootstrap reproducibility and was treated as exploratory. No dietary pattern was associated with continuous K10 (all within-family FDR-adjusted P ≥0.208) or K10 ≥16 (all within-family FDR-adjusted P ≥0.459). In a secondary, hypothesis-generating analysis, each 1-SD higher sweet–fried–processed pattern score was associated with a 0.392-point lower WHOQOL psychological-domain score (β = −0.392, 95% CI −0.671 to −0.113; within-WHOQOL-family FDR-adjusted P = 0.024; global FDR-adjusted P = 0.073). No exploratory association with the depressive-symptom composite or CD-RISC-10 survived within-outcome-family FDR correction. The first three patterns and the secondary WHOQOL association were similar in three-component varimax and four-component promax analyses. The analysis of the prespecified primary outcome, continuous K10 psychological distress, was null. The association between the sweet–fried–processed pattern and poorer psychological quality of life was secondary and hypothesis-generating. Given the cross-sectional design, small sample, and data-driven dietary assessment, prospective replication using validated dietary measures and external pattern validation is required.