Maternal dietary patterns during early pregnancy and birth weight in Japan

Abstract

Abstract Background The average birth weight in Japan has decreased by 200 g in the last 40 years. Only three studies were reported for the association between maternal dietary patterns and birth weight in East Asia, whose results were inconsistent. We examined what maternal dietary patterns were associated with the birth weight in Japan. Methods Totally 22,493 pregnant women were recruited between July 2013 and September 2016 into the Tohoku Medical Megabank Project Birth and Three-Generation Cohort Study. We included 17,287 women who had a full-term single healthy baby into the analysis. Consumption of food and beverage items was evaluated based on food frequency questionnaire at the first-trimester. Dietary patterns were analyzed using a machine learning method of k-means clustering algorithm. Birth weight was obtained from the medical record. The association between dietary patterns and birth weight was analyzed using multiple liner regression model adjusted for potential confounders with multiple imputation method for missing values. Results Dietary patterns were classified into seven groups by cluster analysis: “high in rice (reference) (n = 8046)”, “middle in vegetables, beans, mushrooms, seaweeds and miso-soup (n = 4842)”, “high in fruits (n = 1176)”, “high in bread, dairy and alcohol (n = 1091)”, “high in meat and fish (n = 1049)”, “high in coffee, black tea, soft drinks and confections (high in coffee) (n = 585)”, and “high in vegetables, beans, mushrooms, seaweeds and miso-soup (n = 498)” groups. In multiple liner regression models, birth weight was 22.6 g (95%CI: 0.1 to 45.2 g) heavier in “high in fruits” group than in reference group and 39.4 g (95%CI: 8.6 to 70.3 g) lighter in “high in coffee” group than in reference group. The other groups were not statistically significant. Conclusions High consumption of fruits and high consumption of coffee, black tea, soft drinks and confections during early pregnancy were associated with increased and decreased birth weight, respectively. Key messages We found maternal dietary patterns during early pregnancy using a machine learning method of k-means clustering algorithm. We found maternal dietary patterns which associated with the birth weight in Japan.