Comparative analysis of two analytical methodologies for deriving dietary patterns associated with vitamin D insufficiency and anemia among expectant mothers

Abstract

Objectives: Dietary patterns play a role associated with acute or chronic diseases. This study compared the correlation between dietary patterns and vitamin D status, using two methods for dietary pattern identification, related to gestational anemia among expectant mothers. Methods: In this cross-sectional study a total of 1502 expectant mothers aged > 15 years were recruited from the Nutrition and Health Survey data. Dietary patterns were discerned through principal component analysis (PCA) and reduced rank regression (RRR). Associations between dietary pattern scores, vitamin D, and anemia-related biomarkers were validated using linear and binomial logistic regression adjusted for sociodemographic and economic factors. Results: Two dietary patterns were identified: a plant and marine-based dietary pattern (PMDP) via PCA and a convenience food dietary pattern (CFDP) via RRR. PMDP was characterized by high intakes of plant and marine foods, while CFDP showed high intakes of canned and pickled foods. Risk estimation indicated that high PMDP intake was associated with 30% reduction in odds of vitamin D insufficiency, while moderate CFDP intake was linked to an increased risk of vitamin D insufficiency by 46%. Logistic regression analysis showed a positive association between CFDP and serum iron (β = 0.08; 95% CI: 0.05, 0.24), but a negative association with total iron binding capacity (β = -0.05; 95% CI: -0.44, -0.07). Conclusions: The PCA-derived dietary pattern effectively identifies eating patterns, while the RRR-derived dietary pattern has a better estimate for disease associations in a specific population. The RRR-derived dietary pattern provides valuable insights for dietary guidelines targeting vitamin D insufficiency.

Keywords: anemia-related biomarkers; principal component analysis; reduced rank regression analysis; serum vitamin D.