| 摘要: | Objectives: Gestational anemia, also known as GA, has emerged as a significant public health concern among women of childbearing age. Concurrently, vitamin D, recognized as a prohormone, has garnered scientific interest due to its regulatory role in hepcidin. However, the precise relationship between vitamin D and the modulation of anemia-related blood biomarkers remains controversial. Therefore, each current investigation aimed to elucidate the following objectives-study 1: aimed to investigate the associations between dietary patterns derived from principal component analysis (PCA) and anemia-related blood biomarkers alongside vitamin D. Study 2: undertook to examine the dietary associations using reduced rank regression (RRR) analysis-derived dietary patterns and serum anemia-related blood parameters in conjunction with vitamin D. Study 3: was employed to reveal the associations between dietary patterns derived from k-means cluster analysis, vitamin D, and gestational anemia, incorporating risk predictions utilizing machine learning algorithms (MLA) support vector machine (SVM), k-nearest neighbor (KNN), na?ve bayes (NB), random forest (RF), decision tree (DT)]. Study 4: conducted a comparative analysis between two statistical methods aiming to identify the optimal predictive method for determining the association between dietary patterns, vitamin D insufficiency, and anemia risk.
Methods: A total of 1502 adult pregnant women (> 15 years old) were selected from the National Nutrition Health Survey in Taiwan conducted from 2017 to 2019 (NNHSIT -2017 to 2019). Data collection occurred during the initial trimester, encompassing anthropometric, socioeconomic, and dietary data, including food frequency questionnaires and 24-hour dietary recalls. Dietary patterns were derived using principal component analysis (PCA), reduced rank regression (RRR), and k-means cluster analysis. Linear regression (β coefficient, 95% confidence interval), binomial logistic regression (odds ratio [OR], confidence interval [CI]), relative risk analysis, and machine learning algorithms were employed to explore the associations between serum anemia-related blood parameters, vitamin D, and GA. Additionally, covariate adjustments were made using sociodemographic, anthropometric, and dietary components to ascertain the most accurate predictive associations.
Results: Study 1: revealed significant associations between expectant mothers (EMs) adhering to plant-based dietary patterns (PbDP) and carnivore dietary patterns (CDP) with serum 25-hydroxy cholecalciferol levels. Covariate adjustments indicated that EMs with moderate PbDP consumption exhibited reduced risks for serum folate (OR = 0.60, 95% CI: 0.41, 0.87) and 25-hydroxy cholecalciferol (OR = 0.69, 95% CI: 0.52, 0.93) deficiencies. EMs with the highest CDP consumption demonstrated decreased serum iron (OR = 1.33, 94% CI: 1.02, 1.75), vitamin B12 (OR = 0.25, 95% CI: 0.17, 0.37) and 25-hydroxy cholecalciferol (OR = 0.59, 95% CI: 0.44, 0.80) levels. Similarly, high consumption of dairy and non-dairy alternative dietary pattern (DnDADP) correlated with decreased tendencies for serum folate (OR = 0.67, 95% CI: 0.46, 0.98) and vitamin B12 (OR = 0.66, 95% CI: 0.48, 0.90) concentrations. In Study 2: after accounting for all pertinent factors, linear regression analysis demonstrated a positive correlation between the ferritin related dietary pattern (FrDP) and serum iron levels, along with a tendency towards a negative correlation with serum 25(OH) vitamin D. Pregnant women in the highest FrDP tertile exhibited reduced odds of low serum iron (OR = 0.65, 95% CI: 0.50, 0.85) but increased odds of low 25(OH) vitamin D (OR = 1.79, 95% CI: 1.32, 2.43) levels. Study 3: findings from binomial analysis indicated that individuals following the moderate plant + low animal (MP+LA) dietary pattern exhibited decreased probabilities of low serum iron (OR = 0.45, 95% CI: 0.34, 0.60) and ferritin (OR = 0.27, 95% CI: 0.21, 0.36) but an elevated probability of low serum 25-(OH) vitamin D (OR = 1.47, 95% CI: 1.14, 1.88) levels. The MLA model's accuracy in identifying dietary patterns ranged from 70% to 76%, with sociodemographic and dietary variables being the most influential predictors. In study 4: the final model of logistic regression analysis showed a positive correlation between serum iron levels and the convenience food dietary patter (CFDP), while a negative correlation was noted with total iron binding capacity. Risk assessment indicated a 0.41% reduction in the odds of vitamin D insufficiency among pregnant women with high intake of plant and marine-based dietary pattern (PMDP). Conversely, moderate consumption of RRR-derived DP (CFDP) was associated with a 0.95% increased risk of vitamin D insufficiency.
Conclusions: Overall, the findings of this study elucidate the intricate interplay between dietary patterns, vitamin D status, and gestational anemia risk among pregnant women. Among all the models, RRR stands out as a promising approach for measuring vitamin D deficiency and insufficiency-related anemia risk associations. Serum vitamin D exhibits positive associations with anemia-related blood biomarkers among pregnant women, highlighting the importance of maintaining appropriate serum vitamin D and iron status at the onset of pregnancy. |
