The association between dietary patterns derived by three statistical methods and type 2 diabetes risk: YaHS-TAMYZ and Shahedieh cohort studies

Abstract

AbstractFindings were inconsistent regarding the superiority of using recently introduced hybrid methods to derive DPs compared to widely used statistical methods like principal component analysis (PCA) in assessing dietary patterns and their association with type 2 diabetes mellitus (T2DM). We aimed to investigate the association between DPs extracted using principal component analysis (PCA), partial least-squares (PLS), and reduced-rank regressions (RRR) in identifying DPs associated with T2DM risk. The study was conducted in the context of two cohort studies accomplished in central Iran. Dietary intake data were collected by food frequency questionnaires (FFQs). DPs were derived by using PCA, PLS, and RRR methods considering. The association between DPs with the risk of T2DM was assessed using log-binomial logistic regression test. A total of 8667 participants aged 20–70 years were included in this study. In the multivariate-adjusted models, RRR-DP3 characterized by high intake of fruits, tomatoes, vegetable oils, and refined grains and low intake of processed meats, organ meats, margarine, and hydrogenated fats was significantly associated with a reduced T2DM risk (Q5 vs Q1: RR 0.540, 95% CI 0.33–0.87, P-trend = 0.020). No significant highest-lowest or trend association was observed between DPs derived using PCA or PLS and T2DM. The findings indicate that RRR method was more promising in identifying DPs that are related to T2DM risk compared to PCA and PLS methods.