Dietary Patterns and Health Outcomes (Cardiovascular, Metabolic, Endocrine, Neurological, Skeletal Muscular, Cancer)

Health status and lifestyle information was self-reported on a questionnaire at baseline, and on questionnaires distributed to participants biennially thereafter. Dietary intake data was collected in form of a comprehensive, 131-item food frequency questionnaire (FFQ), distributed among participants every four years. The response rate remained over 90%. Dietary data was validated using biomarkers, health status data using medical records. Dietary pattern scores were derived from FFQs using cumulative average whenever possible from years preceding the outcomes. Multivariable Cox proportional hazards models were used to evaluate associations between dietary pattern scores and health outcomes, except in case of pregnancy complications (such as gestational diabetes mellitus/GDM and hypertensive disorders of pregnancy/HDPs) where multivariable logistic regression models with generalized estimating equations, with an exchangeable working correlation structure to account for correlated outcomes between pregnancies.