Association between maternal dietary quality and gestational diabetes 1 mellitus based on the Diet Balance Index for Pregnancy

1 Background: Maternal diet is critical to the development of gestational diabetes mellitus 2 (GDM), but sparse studies have applied the Chinese Dietary Balance Index for Pregnancy 3 (DBI-P) to assess the maternal dietary quality and its relationship with GDM. We aimed to 4 examine the maternal dietary quality and its relationship with GDM risk using the newly 5 developed DBI-P. 6 Methods: We included 1122 pregnant women from the Tongji Birth Cohort (TJBC) in 7 Wuhan, China. The semi-quantitative food frequency questionnaire (FFQ) was used to obtain 8 the dietary information during pregnancy. The newly developed DBI-P, including DBI-P 9 components and DBI-P dietary patterns, was applied to comprehensively evaluate the 10 maternal dietary quality. GDM was diagnosed by the 75-g, 2-h oral glucose tolerance test at 11 24-28 weeks gestation. We used a generalized linear regression model to examine the 12 relationship between DBI-P components and blood glucose levels, and a logistic regression 13 model to examine the relationship between dietary patterns and GDM. Results: A total of 179 participants (16.0%) were diagnosed with GDM. There is a widespread 15 phenomenon of dietary imbalance among pregnant women in Wuhan. With per score increase 16 in nuts and eggs, fasting blood glucose increase by 0.03 mmol/l (95% CI: 0.01, 0.05) and 0.01 17 mmol/l (95 %CI: 0.01, 0.03), respectively, while per score increase in fruits, 1-h post-load blood 18 glucose decreased by 0.05 mmol/l (95% CI: -0.10, -0.01). Besides, compared with pattern E 19 characterized by higher intake of total energy, total fat, fruits, and cooking oil, pattern B (OR: 20 0.51, 95% CI: 0.26, 0.99) and pattern C (OR: 0.27, 95% CI: 0.09, 0.81) were associated with a 21 lower GDM risk. The associations between dietary patterns and GDM risk may be partly 22 attributed to the intakes of dietary total fat, carbohydrate, cholesterol, GDM. The newly established DBI-P can provide an easy-to-implement tool to assess maternal 1 dietary quality. These findings will provide new insights for the exploration of preferable 2 dietary evaluation methods and prevention and control of GDM.

≥28.0 kg/m 2 ) according to the BMI classification criteria for Chinese [28]. Smoking or drinking 23 defined as the habit of smoking or drinking at least once a week before pregnancy was divided 24 into two levels (Yes/No). The other variables, including primiparity, physical activity, family 1 history of diabetes and hypertension, and the history of abnormal pregnancy (including ectopic 2 pregnancy, induced labor, spontaneous abortion et al.) were treated as dichotomized variables 3 (Yes/No). The continuous variables (e.g., age, Pre-pregnancy BMI, and the average daily intake of 7 nutrients and foods) were presented as mean and standard deviation (SD), and categorical 8 variables (e.g., ethnic, education levels, and smoking) were expressed as frequencies 9 (percentages). One-way analysis of variance (ANOVA) and χ 2 test were used to detect the 10 differences in percentages and means, respectively. We also performed a generalized linear 11 regression model to examine the relationship between DBI-P components and blood glucose 12 levels. The results were presented as coefficients (β) with 95% confidence intervals (95% CIs).

13
Logistic regression models were used to evaluate the relationship between DBI-P dietary 14 patterns and GDM risk, further explored the contribution of nutrient and food intake to the 15 association between dietary patterns and GDM by calculating the odds ratios (ORs) and 95% Maternal diet quality based on DBI-P 8 Table 2 reports the scores for DBI-P components and the percentages of each score in the GDM 9 and non-GDM group. Insufficient intake of iron-rich foods, soybean and its products, aquatic 10 products, milk and milk products, eggs, and meat were common in both two groups, with 96.1%, 11 94.4%, 79.9%, 78.2%, 68.2%, 61.4% of the GDM women getting negative scores, respectively.  Association between maternal dietary quality and GDM based on DBI-P 3 as shown in Table 4. After adjusting for maternal age, ethnicity, pre-pregnancy BMI, and all 4 variables shown in the table, with per score increase in nuts and eggs, FPG increased by 0.03 5 mmol/l (95% CI: 0.01, 0.05) and 0.01 mmol/l (95% CI: 0.01, 0.03), respectively, while per 6 score increase in fruits, 1-h PBG decreased by 0.05mmol/l (95% CI: -0.10, -0.01). No 7 significant relationships were found between DBI-P components and 2-h PBG levels.
8 Table 5 shows the average daily intake of nutrients and foods in various DBI-P dietary 9 patterns. Only five dietary patterns found in this study, including pattern A to pattern E. The 10 optimal pattern A accounted for 18.4%. A total of 669 participants had a pattern B, in which the 11 average daily intakes of most dietary nutrients were below the recommended intakes, such as 12 dietary protein, vitamin A, magnesium, calcium, and iron et al. The results suggested that from 13 patterns A to pattern B to pattern C, the average daily intake of most nutrients and foods 14 decreased significantly (except for dietary vitamin C, magnesium, and total iron intake), and 15 pregnant women with pattern D consumed more nutrients and foods than those with pattern E 16 (except for dietary vitamin C). Besides, pregnant women with pattern D and pattern E consumed 17 excessive intake of total energy, total fat, cereals, fruits, and cooking oil. The average daily 18 intake of nutrients and foods in various dietary patterns were significantly different (all P<0.05).
19 Table 6 shows the relationship between DBI-P dietary patterns and GDM. The adjusted 20 models showed that pregnant women with pattern B and pattern C were related to a lower risk 21 of GDM compared to those with pattern E, with an adjusted OR of 0.51 (95% CI: 0.26, 0.99) 22 and 0.27 (95% CI: 0.09, 0.81), respectively. We further investigated the contributions of energy-23 yielding nutrients and food intake on the association between dietary patterns and GDM risk 1 and found that the association of pattern B with GDM risk was no longer significant after extra 2 adjustment for dietary total fat (OR: 0.53; 95% CI: 0.27, 1.05), or carbohydrate (OR: 0.53; 95% 3 CI: 0.27, 1.03), or cholesterol (OR: 0.52; 95% CI: 0.27, 1.03), or eggs (OR: 0.52; 95% CI: 0.26, 4 1.02), or cooking oil (OR: 0.52; 95% CI: 0.27, 1.03). The association between pattern C and 5 GDM risk was attenuated slightly but maintained statistical significance after adjusting the 6 above factors. These findings suggested that dietary total fat, carbohydrate, cholesterol, eggs, 7 and cooking oil may partly contributed to the difference in GDM risk. Besides, no large or 8 statistically significant changes in OR (95% CI) for GDM risk were observed after adjusting 9 for other dietary nutrients and food.

11
In this prospective cohort study, we newly developed a DBI-P to investigate the maternal dietary 12 quality and its relationship with GDM. Firstly, there is a widespread phenomenon of dietary 13 imbalance among pregnant women in Wuhan, China, which is mainly manifested as dietary 14 insufficiency. Secondly, we found nut and egg scores were positively related to FPG, while fruit 15 scores were inversely related to 1-h PBG. Thirdly, five dietary patterns were identified in this 16 study. Compared with pattern E characterized by high total energy, fruit and cooking oil intake, 17 pattern B and pattern C were related to a lower GDM risk. The associations between dietary 18 patterns and GDM risk may be partly attributed to the intakes of dietary total fat, carbohydrate,  Almost all pregnant women (96.1%) in the study had insufficient iron intake and 61.4% of 21 pregnant women had insufficient meat intake. During the survey, we were somewhat astonished 22 to note that multiple pregnant women never prefer to consume animal entrails before pregnancy, 23 which may be one of the reasons why the participants had such a severe iron-rich food deficiency during pregnancy. Since iron-rich food deficiency is related to premature, 1 postpartum hemorrhage, and low birth weight, it is indispensable to augment the intake of iron-2 rich foods during pregnancy [29,30]. Besides, soybean and its products, aquatic products, milk 3 and milk products, eggs, and meats are generally considered favorable sources of high-quality 4 protein and minerals, yet more than 60% of participants did not meet the recommended intake.

5
Adequate dietary protein intake during pregnancy is crucial to ensure a healthy outcome. Bao 6 et al.'s studies found that higher intake of vegetable protein was associated with a lower blood 7 glucose [23], while protein deficiency is associated with small-for-gestational-age infants and 8 premature birth [31]. As for cereals and vegetables, we found the intake of cereals (267.7 g/d) 9 and vegetables (327.6 g/d) basically met or exceeded the recommended intake, however, about 10 half of the participants still had insufficient intake, suggesting a possible polarization of dietary 11 intakes of cereals and vegetables. Meanwhile, the average fruit intake in our study was 477.1 12 g/d and nearly 55.0-75.0% of participants consumed excessive fruits, cooking oil and salt. We 13 speculate that the reason for excessive intake of fruit might be due to poor appetite during 14 pregnancy and the preference for sweet and sour foods, while excessive intake of cooking oils 15 and salt may be due to the inherent dietary characteristics of Wuhan people.

16
Diet during pregnancy has been described as a potentially modifiable factor for GDM [7]. In 17 this study, we found higher nut scores were associated with an increase in FPG. This finding is 18 contrary to what has been reported previously. For instance, Kendall et al. observed higher nut 19 intake was associated with a lower postprandial glycemia, whereas no significant association 20 was found in other studies [32][33][34]. Considering that the average nuts intake in our study was 21 15.3 g/d, exceeding the recommended intake of dietary guidelines for pregnant women, the 22 energy and fat or carbohydrate content of nuts and the types of nuts may be the primary cause 23 of the diverse effect. Additionally, we observed a positive correlation between egg scores and 24 FPG, which was consistent with two cohort studies in China and a prospective epidemiological 1 study in Polish [35,36]. One possible mechanism may be that increased egg consumption is 2 correlated with higher cholesterol levels, which is related to systemic inflammation and has 3 recently been linked to the development of GDM [35]. Furthermore, we also found that fruit 4 score was negatively correlated with 1-h PBG. Evidence showed that fruits are rich in 5 antioxidants as well as vitamins, minerals, polyphenols, and flavonoids, and the mechanism of 6 lowering blood glucose level in pregnant women may be related to the dietary total antioxidant 7 capacity [37]. Consistent with our study, the Tongji Maternal and Child Health Cohort observed 8 higher fresh fruit consumptions in mid-pregnancy were associated with a lower plasma1-h PBG 9 and 2-h PBG [6], whereas the Nurses' Health Study II conducted in US female nurses found no 10 significant association between higher whole fruit intake and GDM risk, but moderate 11 consumption of fruit juice appeared to lower GDM risk [38]. The accessibility of fruits in 12 various regions, and the differences in nutritional value and glycemic index of fruits may partly 13 explain these inconsistent findings. Besides, we were somewhat surprised to notice that the  In summary, the newly established DBI-P can provide an easy-to-implement tool to assess 23 maternal dietary quality, and the results showed that unfavorable dietary quality during pregnancy is associated with a higher risk of GDM. These findings will provide new insights 1 for the exploration of preferable dietary evaluation methods and prevention and control of GDM.  Competing interests 5 We have no competing interests.          Table 6 The contribution of nutrients and food intake to the association between dietary patterns and GDM (n=1122) * .