Normal pregnancy is characterised as a ‘diabetogenic state’ due to the increasing postprandial glucose and decreasing insulin sensitivity(Reference Catalano1). Considerable evidence shows that abnormal glucose metabolism during pregnancy, including hyperglycaemia and gestational diabetes mellitus (GDM), is associated with adverse health outcomes for both mother and offspring(Reference Bianco and Josefson2,Reference Farahvar, Walfisch and Sheiner3) . The incidence of GDM has risen in recent decades, posing a global public health burden(Reference Lende and Rijhsinghani4).
While there are multiple predictors of GDM (e.g. later age at childbearing, maternal obesity and family history of type 2 diabetes mellitus)(Reference Mcintyre, Catalano and Zhang5), maternal diet could play an important role. Most studies have examined the effect of individual food or nutrient on the risk of GDM(Reference Mijatovic-Vukas, Capling and Cheng6,Reference Misra, Yew and Shin7) . However, the ‘single nutrient’ approach may be not enough, considering the complicated interplay among nutrients and the difficulty to assess the independent effect of each nutrient or food precisely(Reference Akbaraly, Singh-Manoux and Dugravot8). Therefore, research increasingly focuses on overall diet quality to examine diet–diseases associations(Reference Hu9). Two approaches are usually applied: a priori dietary index based on dietary guidelines or diets known to be healthy, and a posteriori dietary pattern derived from factor or cluster analysis(Reference Hu9). Compared with data-driven dietary patterns, dietary index is able to evaluate the association of adherence to the current dietary guidance and disease prevention and is conducive for comparisons across different populations(Reference Miller, Lazarus and Lesko10). Associations between posteriori dietary patterns and GDM risk have been widely investigated. However, little work has been done on dietary index and maternal glucose metabolism. The few available data suggested that adherence to the Healthy Food Intake Index (HFII)(Reference Meinila, Valkama and Koivusalo11), the Dietary Approaches to Stop Hypertension (DASH)(Reference Izadi, Tehrani and Haghighatdoost12), the Mediterranean Diet Index (MDI)(Reference Izadi, Tehrani and Haghighatdoost12,Reference Karamanos, Thanopoulou and Anastasiou13) and the Healthy Eating Index (HEI)(Reference Tryggvadottir, Medek and Birgisdottir14) may lower maternal glucose levels or GDM risk. Notably, these limited studies were mostly conducted in the Caucasian population. More attention should be paid to Asian women, since this population is at greater risk of GDM than Caucasian women(Reference Hedderson, Ehrlich and Sridhar15,Reference Deputy, Kim and Conrey16) and has distinct dietary habits(Reference He, Yuan and Chen17).
The Chinese Diet Balance Index (DBI) was designed to assess adherence to the Dietary Guidelines for Chinese(Reference He, Fang and Xia18). It has been widely used and verified among subgroups in China. The Diet Balance Index for Pregnancy (DBI-P) was adapted from the latest version of DBI (DBI_16) and modified according to recommendations from the Chinese dietary guidelines for pregnant women(Reference Pan, Wu and Chen19). The DBI-P can evaluate the adherence to dietary guidelines and reflects overall excessive and inadequate nutrition among pregnant women(Reference Pan, Wu and Chen19).
To our knowledge, no study has investigated the association between adherence to dietary guidelines for pregnant women and maternal glucose metabolism by using DBI-P. Thus, we examined the association between adherence to the dietary guideline assessed by DBI-P and glucose metabolism in Chinese women.
Materials and methods
Study design and participants
Data were derived from the baseline survey of the Yuexiu birth cohort (ClinicalTrial.gov number: NCT03023293) in Guangzhou, China. We recruited pregnant women at 20–28 weeks at a maternal and child health hospital between March 2017 and September 2018. Eligible women were those aged 20–45 years, with a singleton pregnancy and accepted FFQ. Women with pre-existing metabolic, endocrine diseases (e.g. diabetes mellitus, CVD and polycystic ovary syndrome), pregnancy infection or mental disorder were excluded. Furthermore, those pregnant women who had missing records on the oral glucose tolerance test (OGTT) and had missing data on core food items were also excluded. In total, 942 pregnant women were included in this study.
The study protocol was approved by the Ethics Committee of the School of Public Health of Sun Yat-Sen University and adhered to the guidelines of the Declaration of Helsinki. All participants were carefully instructed and signed informed consent at initial enrolment.
Dietary data collection
Dietary intake during the past month before OGTT was assessed at 20–28 weeks of gestation via a face-to-face interview, using a semi-quantitative FFQ. The FFQ consisted of eighty-one food items and has been previously shown to be valid and reproducible for use among Chinese women(Reference Zhang and Ho20). The participants were asked to report the frequency (never, daily, weekly and monthly) and the number of servings per frequency for each of the food item. With food picture aids, trained interviewers recorded the portion size of the food consumption. Daily food intakes in grams were calculated using the product of daily frequency intake and amount of food intake per day in standard portions. In addition, dietary data were summed up by food group classifications corresponding with the Chinese Food Pagoda(Reference Wang, Lay and Yu21) for the calculation of DBI-P scores. The average daily intake of total energy was then computed based on the Chinese Food Composition Table(Reference Yang, Wang and Pan22).
Calculation of Diet Balance Index for Pregnancy
The Chinese DBI-P aims to assess the adherence to the dietary guideline among Chinese pregnant women. Lower absolute scores of the DBI-P denote greater compliance with the Chinese dietary guides for pregnancy. Food components of DBI-P include (1) cereals, (2) vegetables and fruits, (3) dairy products, soyabeans and nuts, (4) animal food (including meat, poultry, fish, shrimp and egg), (5) empty energy food (including cooking oil and alcoholic beverage), (6) condiments (including addible sugar and salt), (7) diet variety, and (8) drinking water. For each component, a score of 0 demonstrates meeting the recommended intake amounts. Positive score denotes excessive intake, while negative score indicates insufficient intake. The diet variety included twelve categories of food: rice and products; wheat and products; maize, coarse grains and products, starchy roots and products; dark-coloured vegetables; light-coloured vegetables; fruit; soyabeans and nuts; milk and dairy products; red meat and products; poultry and game; egg; and fish and shellfish. Scoring details of DBI-P can be found in Supplementary Table S1.
By summing scores for each DBI-P component, three indicators were calculated. The high bound score (HBS) indicates excessive food intake by summing all the positive scores. The low bound score (LBS) indicates inadequate food intake by summing all the absolute value of negative scores. The diet quality distance (DQD) indicates imbalanced food intake by summing the absolute values of both positive and negative scores. The ranges of scores for HBS, LBS and DQD were: 0–44, 0–72 and 0–96, respectively(Reference He, Fang and Xia18,Reference Pan, Wu and Chen19) .
Glucose tolerance test
At 20–28 weeks, pregnant women were routinely screened for GDM with a 75-g, 2-h OGTT test. All participants had fasted overnight for at least 8 h before OGTT. OGTT-0 h glucose (fasting plasma glucose), OGTT-1 h and OGTT-2 h glucose (postprandial glucose) were measured with the glucose oxidase method. According to the International Association of Diabetes and Pregnancy Study Group, the diagnosis of GDM was made when any of the following blood glucose values was met or exceeded: OGTT-0 h, 5·1 mmol/l; OGTT-1 h, 10·0 mmol/l; and OGTT-2 h, 8·5 mmol/l(Reference Metzger, Gabbe and Persson23).
Covariates
Information on socio-demographic and health characteristics was collected during the baseline survey. Maternal age (in years) was treated as a continuous variable. The occupation was categorised into four groups (housewives, administrators and technicians, commerce and services, and others). Monthly household income was divided into four groups (≤ 4000, 4001–6000, > 6001–10 000 and > 10 000 RMB). The history of GDM was categorised into three groups (primiparae, yes and no). Family history of diabetes, smoking and alcohol use were treated as dichotomised variables (yes or no). Physical activity was assessed using the International Physical Activity Questionnaire (IPAQ) and was expressed as metabolic equivalent tasks (MET)(Reference Craig, Marshall and Sjöström24). Data on height and pre-pregnancy weight measured by trained clinical nurses were obtained from medical records. Pre-pregnancy BMI was calculated subsequently as pre-pregnancy weight (kg) divided by height squared (m2).
Statistical analysis
Continuous variables which normally distributed were reported as mean ± standard deviation (sd). To evaluate the differences in maternal characteristics across two groups, t tests or χ 2 tests were applied. Multiple linear regression and logistic regression analyses were used to evaluate the associations between scores of DBI-P (for each component and three indicators) and maternal glucose metabolism. Model 1 was adjusted for maternal age and pre-pregnancy BMI. Models 2 was further adjusted for history of GDM, family history of diabetes, smoking, alcohol use, physical activities, daily energy intake, occupation and monthly household income. The Benjamini–Hochberg method(Reference Benjamini and Hochberg25) was used for the P-value correction upon multiple comparisons across DBI-P food components. All analyses were performed with SAS 9.4 (SAS Institute). All P-values were two-sided, and statistical significance was determined at the P-value less than 0·05 level.
Results
Characteristics of the participants
The socio-demographic characteristics of the participants are presented in Table 1. The incidence of GDM was 19 % (179/942). The mean age was 30·8 ± 4·89 years, and mean pre-pregnancy BMI was 20·54 ± 2·93 kg/m2. Compared with women without GDM, those with GDM had higher age, pre-pregnancy BMI, glucose levels, and higher percentage of history of GDM and had lower physical activity levels (P < 0·05).
Table 1. Characteristics of pregnant women by the development of GDM
GDM, gestational diabetes mellitus; MET, metabolic equivalent task; OGTT, oral glucose tolerance test.
Scores for the Diet Balance Index for Pregnancy food components and indicators of the participants
Table 2 provides the scores for DBI-P food components and indicators among the participants. The mean scores of cereals, vegetables, fruits, dairy products, soyabeans and nuts, fish and shrimp, egg, diet variety, and water and soup were negative, indicated insufficient intake and low variety. In contrast, the mean scores of meat and poultry, cooking oil, alcoholic beverage (scores near to 0), addible sugar, and salt were positive, indicating excessive intake. The mean scores for HBS, LBS and DQD were 6·29, 24·68 and 30·97, respectively. Further details of scores distribution of components and indicators were provided in Supplemental Table S2 and Supplemental Table S3. As shown in Table 2, participants with GDM had higher score of animal food intake than those without GDM (P < 0·05).
Table 2. Scores for DBI-P food components and indicators by the development of GDM
DBI-P, Diet Balance Index for Pregnancy; GDM, gestational diabetes mellitus; HBS, high bound score; LBS, low bound score; DQD, diet quality distance.
Scores for the Diet Balance Index for Pregnancy food components and indicators in relation to plasma glucose levels
Tables 3 and 4 present the multiple linear regression models for OGTT glucose levels by DBI-P food components and indicators. After adjustment for potential confounding factors and multiple comparisons, higher score of animal food intake (β: 0·045; se: 0·015; P = 0·045) was significantly associated with higher OGTT-2 h glucose levels. No significant association was observed between scores of other food components and maternal glucose levels. After adjustment for potential confounding factors, higher value of HBS (β: 0·037; se: 0·017; P = 0·029) was significantly associated with higher OGTT-2 h glucose levels. No significant associations were observed between the values of LBS, DQD and maternal glucose levels.
Table 3. Multiple linear regression of scores for DBI-P food components with maternal glucose levels
DBI-P, Diet Balance Index for Pregnancy; OGTT, oral glucose tolerance test; GDM, gestational diabetes mellitus.
Model 1 was adjusted for age and pre-pregnancy BMI.
Model 2 was adjusted for age, pre-pregnancy BMI, family history of diabetes, history of GDM, smoking, alcohol use, physical activities, daily energy intake, occupation and monthly household income.
For DBI-P food components, P-values were further adjusted for multiple comparisons using the Benjamini–Hochberg method.
Table 4. Multiple linear regression of scores for DBI-P indicators with maternal glucose levels
DBI-P, Diet Balance Index for Pregnancy; OGTT, oral glucose tolerance test; HBS, high bound score; LBS, low bound score; DQD, diet quality distance; GDM, gestational diabetes mellitus.
Model 1 was adjusted for age and pre-pregnancy BMI.
Model 2 was adjusted for age, pre-pregnancy BMI, family history of diabetes, history of GDM, smoking, alcohol use, physical activities, daily energy intake, occupation and monthly household income.
Scores for the Diet Balance Index for Pregnancy food components and indicators in relation to gestational diabetes mellitus
Table 5 shows the association of food components and indicators for DBI-P with risk of GDM. After adjustment for potential confounding factors and multiple comparisons, score of animal food intake (OR = 1·105, 95 % CI 1·038, 1·176) was positively associated with the risk of GDM. No significant relationships were observed between DBI-P indicators and GDM risks.
Table 5. Association between food components and indicators of DBI-P and risk of GDM
DBI-P, Diet Balance Index for Pregnancy; GDM, gestational diabetes mellitus; HBS, high bound score; LBS, low bound score; DQD, diet quality distance.
Model 1 was adjusted for age and pre-pregnancy BMI.
Model 2 was adjusted for age, pre-pregnancy BMI, family history of diabetes, history of GDM, smoking, alcohol use, physical activities, daily energy intake, occupation and monthly household income.
For DBI-P food components, P-values were further adjusted for multiple comparisons using the Benjamini–Hochberg method.
Discussion
This is the first study that has investigated the association between adherence to dietary guidelines during pregnancy and maternal glucose metabolism based on DBI-P. The current study showed that overall excessive food intake was positively associated with OGTT-2 h glucose levels. Of the DBI-P food components, excessive intake of animal food was associated with higher postprandial glucose levels and an increased risk of GDM.
Higher score of HBS in DBI-P reflects higher degree of overnutrition. Overnutrition may occur due to excessive intake of certain foods such as meat and poultry, cooking oil, addible sugar and salt, which dietary guidelines suggest consuming moderately or less(Reference He, Fang and Xia18). In our study, higher HBS score was associated with higher OGTT-2 h glucose levels. Consistent with this result, a Finnish study showed that higher adherence to the Nordic Nutrition Recommendations (NNR) evaluated by HFII was associated with lower OGTT-2 h glucose load(Reference Meinila, Valkama and Koivusalo11). Data from a non-interventional, multi-centre study indicated that adherence to the healthy Mediterranean diet was associated with better glucose tolerance and lower GDM risk in ten Mediterranean countries(Reference Karamanos, Thanopoulou and Anastasiou13). Further, a study conducted in Iceland found the HEI, which is based on the dietary recommendations for Americans, was associated with decreased risk of GDM(Reference Tryggvadottir, Medek and Birgisdottir14). However, null associations were found in Australian women, whose diet quality was assessed by the Australian Recommended Food Score (ARFS)(Reference Gresham, Collins and Mishra26). Only 4 % of GDM cases were identified by self-report in this Australian study(Reference Gresham, Collins and Mishra26). The prevalence of GDM determined from self-report may be underreported, which probably results in a loss of statistical power(Reference Comino, Tran and Haas27). Vajihe et al. reported that the Mediterranean and DASH diets were associated with decreased risk of GDM in Iranian pregnant women(Reference Izadi, Tehrani and Haghighatdoost12).
The indicator DQD and LBS were used to evaluate the imbalanced and insufficient dietary intake, respectively(Reference He, Fang and Xia18). The imbalance of dietary intake of our study participants was mainly due to insufficient intake. The score of water and soup accounted for the largest proportion of LBS score. In contrast, other healthy food components such as fruits and vegetables occupied a relatively small proportion. This may explain the null association between LBS and maternal glucose metabolism. Few studies have evaluated the diet quality of pregnant women with glucose metabolism, using priori dietary indices. Furthermore, most existing studies have focused on Western population. Due to differences in dietary intakes between Western and Eastern countries and the higher prevalence of GDM within the Asian population(Reference Deputy, Kim and Conrey16), more epidemiological studies are needed in Asian population.
Score for the total animal food is the sum of scores for meat, poultry, fish, shrimp and egg. In this study, total animal food intake was positively associated with postprandial glucose levels and GDM risk. However, when meat, poultry, fish, shrimp and egg were analysed separately, we found no significant association. This may suggest that the effect of combined dietary factors may be more easily detectable than that for isolated nutrients and foods(Reference Hu9). Excessive intake of animal fat during pregnancy was observed in previous studies(Reference Elvebakk, Mostad and Mørkved28,Reference Caut, Leach and Steel29) . Participants with GDM showed excessive intake of animal food and had a higher consumption of animal food than women without GDM in our study. Many ‘single food’ studies have found that higher consumption of red meat, processed meat and egg were associated with higher maternal glucose levels and an increased risk of GDM(Reference Schoenaker, Mishra and Callaway30,Reference Wu, Sun and Zhou31) . Previous study also showed that seafood pattern was associated with a higher risk of GDM(Reference He, Yuan and Chen17,Reference Hu, Oken and Aris32) . The potential mechanism by which the intakes of animal food may influence the GDM risk is complex. Animal food are rich in saturated fat, which may advance obesity, a leading risk factor for GDM(Reference Mcintyre, Catalano and Zhang5). In addition, animal food such as egg, meat and poultry are sources of cholesterol, heme iron, advanced glycation end products, and amino acids, which can cause β-cell damage, decreased insulin secretion or insulin resistance(Reference Wu, Sun and Zhou31,Reference Fretts, Follis and Nettleton33–Reference Newsholme, Bender and Kiely37) . Furthermore, fish and shrimp are the main contributors to arsenic, which have been shown to be associated with increasing insulin resistance, decreasing insulin sensitivity and impairing insulin production(Reference Filippini, Malavolti and Cilloni38–Reference Xia, Liang and Sheng42).
In the present study, no significant association was observed between vegetables or fruits and maternal glucose metabolism. Dietary intake enriched with plant-derived foods, such as vegetables and fruits, presents a low glycaemic pattern and may has a favourable impact on the incidence of GDM(Reference Schiattarella, Lombardo and Morlando43). We speculate that the null association in our study may be due to the little variation of vegetables and fruits intakes among study participants. Analyses with alcoholic beverage also showed no significant association with maternal glucose metabolism. Supplementary Table S2 showed that only 2·44 % of the participants reported alcohol consumption. In contrast, a Norwegian study found that not all dietary recommendations were followed during pregnancy, with 35 % of the women reported alcohol consumption causing for concern(Reference Elvebakk, Mostad and Mørkved28). The low intake of alcohol in our study may make it hard to discover the harmful influence of alcohol.
This study has several limitations. First, owing to the observational nature of our study, we cannot rule out all the residual confoundings and mediators such as gestational weight gain(Reference Zhang, Wang and Xu44). However, our analysis has accounted for many confounding factors including physiological characteristics, socio-demographic characteristics, physical activity and energy intake. Second, the association of certain healthy nutrients (PUFA, Ca, vitamin D, etc)(Reference Pan, Huang and Li45,Reference Chen, Feng and Yang46) and glucose metabolism could not be evaluated by DBI-P. However, the DBI-P consist of three dietary indicators and diverse food components covering all aspects of a healthy diet, which allows us to observe the potential diet–disease association resulting from cumulative intakes of food groups. Third, causation could not be established from this observational study. However, the causal inversion can be ruled out in our study, since habitual dietary intake was collected before the OGTT test and any diet intervention.
Conclusions
We found that the excessive total food intake, particularly animal food intake, was associated with higher postprandial glucose in pregnant women. High animal food intake during pregnancy was also associated with increased risk of GDM.
Acknowledgements
This study was supported by the Key-Area Research and Development Program of Guangdong Province (grant 2019B030335001) and the Leadership Improvement Support Program by the Chinese Nutrition Society (CNS2020100B-5).
W. P.: Methodology, Formal analysis, and Writing – Original Draft; S. K.: Writing – Review & Editing; Q. L.: Project administration; L. X.: Investigation and Validation; S. W.: Investigation; J. J.: Funding acquisition; and L. C. (corresponding author): Conceptualisation, Methodology, Resources, Writing – Review & Editing, Supervision, and Funding acquisition. All authors have read and approved the final manuscript.
The authors declare no competing interests.
Supplementary material
For supplementary material/s referred to in this article, please visit ·https://doi.org/10.1017/S0007114523000107
