Clinical Nutrition 40 (2021) 2791e2799

Contents lists available at ScienceDirect

Clinical Nutrition
journal homepage: http://www.elsevier.com/locate/clnu

Original article

Relevance of dietary glycemic index, glycemic load and fiber intake

before and during pregnancy for the risk of gestational diabetes
mellitus and maternal glucose homeostasis
Xiao Zhang a, Yunhui Gong b, Karen Della Corte c, Dianke Yu d, Hongmei Xue e,
Shufang Shan a, Guo Tian a, Yi Liang f, Jieyi Zhang a, Fang He g, Dagang Yang f, Rong Zhou b,
Wei Bao h, Anette E. Buyken c, 1, Guo Cheng i, *, 1
a
Department of Pediatrics, West China Second University Hospital, Sichuan University, Chengdu, Sichuan, PR China
b
West China Second University Hospital and Key Laboratory of Birth Defects and Related Diseases of Women and Children (Sichuan University) of Ministry
of Education, Sichuan University, Chengdu, PR China
c
Nutrition, Consumption and Health, Faculty of Natural Sciences, Paderborn University, Germany
d
School of Public Health, Qingdao University, Qingdao, PR China
e
College of Public Health, Hebei University, Key Laboratory of Public Health Safety of HeBei University, Baoding, Hebei, China
f
Department of Clinical Nutrition, Affiliated Hospital of Guizhou Medical University, Guizhou Medical University, Guiyang, PR China
g
West China School of Public Health and Healthy Food Evaluation Research Center, Sichuan University, Chengdu, PR China
h
Department of Epidemiology, College of Public Health, University of Iowa, Iowa City, IA 52246, USA
i
Laboratory of Molecular Translational Medicine, Center for Translational Medicine, Key Laboratory of Birth Defects and Related Diseases of Women and
Children (Sichuan University), Ministry of Education, Department of Pediatrics, West China Second University Hospital, Sichuan University, Chengdu,
Sichuan, PR China

a r t i c l e i n f o s u m m a r y

Article history: Background & aims: To date, the prevalence of Gestational diabetes mellitus (GDM) in China was 17.5%.
Received 8 August 2019 Given the substantial relevance of GDM for medium- and long-term health of both mother and offspring
Accepted 29 March 2021 and the paucity of existing data on the link between maternal diet and glucose homeostasis during
pregnancy in Asian population, additional studies are needed. To examine the relevance of dietary gly-
Keywords: cemic index (GI), glycemic load (GL) and fiber intake before and during pregnancy for the development of
Glycemic index
GDM and glucose homeostasis over the course of pregnancy.
Glycemic load
Methods: Cox proportional hazards analysis and linear mixed effects regressions were performed on data
Fiber intake
Gestational diabetes
from 9317 women for whom three food frequency questionnaires (pre-pregnancy, 1st and 2nd trimesters)
Glucose homeostasis and biochemical measures during pregnancy were available. Investigated outcome variables included
Pregnancy GDM risk, fasting plasma glucose (FPG), glycated hemoglobin (HbA1C), and homeostasis model assess-
ment insulin resistance (HOMA-IR) in the 1st, 2nd and 3rd trimesters.
Results: Women in the highest tertile of dietary GI (or GL) before pregnancy, in the 1st, or the 2nd trimester
respectively had a 12% (15%), 25% (23%) or 29% (25%) higher risk of developing GDM than those in the
lowest tertile (all p for trend
0.02). Women with the highest dietary fiber intake before pregnancy, in the
1st or 2nd trimester had a 11%, 17% or 18% lower GDM risk (all p for trend
0.03). Moreover, increases in
GI or GL and decreases in fiber intake over the course of pregnancy (1st to 3rd trimesters) were inde-
pendently associated with adverse concurrent developments in FPG, HbA1C and HOMA-IR (p
0.03).
Conclusions: Our study indicates that dietary GI, GL and fiber intake before and during pregnancy affects
glucose homeostasis of pregnant Chinese women.
© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

* Corresponding author. Laboratory of Molecular Translational Medicine, Center for Translational Medicine, Key Laboratory of Birth Defects and Related Diseases of Women
and Children (Sichuan University), Ministry of Education, Department of Pediatrics, West China Second University Hospital, Sichuan University, Chengdu, Sichuan, 610041, PR
China.
E-mail address: gcheng@scu.edu.cn (G. Cheng).
1
Anette E Buyken and Guo Cheng contributed equally to this study and share the last authorship.

https://doi.org/10.1016/j.clnu.2021.03.041
0261-5614/© 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

1. Introduction from public hospitals with obstetric services in South China

(Sichuan Province, Yunnan Province and Guizhou Province). 2 and
Gestational diabetes mellitus (GDM) is a common complica- 2e3 public hospitals were randomly selected within each urban
tion of pregnancy characterized by glucose intolerance with and rural area, respectively using respective clinic registers. Until
onset or first recognition during pregnancy [1]. For mothers, 2017, 12 urban and 15 rural hospitals were included. Pregnant
GDM is not only associated with adverse perinatal outcomes [2], women were invited at their first routine ultrasound examination
it is also related to a higher risk of GDM during subsequent in gestational week 9e11. To facilitate follow-up, only women who
pregnancies [3], as well as a higher risk of developing type 2 had lived in their current residence for at least 2 years were eligible.
diabetes (T2DM) and premature cardiovascular disease (CVD) The overall response rate was 91.2%. Women who agreed to
over the long term [4]. For offspring, GDM pregnancy confers participate were representative of the general pregnant population
greater risks of developing obesity, T2DM, hypertension, and CVD in age economic and education status according to the regional
in youth and adult life [2]. Given the substantial relevance of statistic book [27e29].
GDM for medium- and long-term health of both mother and Data were assessed at the first routine ultrasound examination
offspring, it is crucial to identify modifiable risk factors which (phase 1 (P1)), in gestational weeks 20e22 (P2), and in gestational
contribute to GDM prevention. weeks 33e35 (P3), and 8 times in infancy and childhood At P1, each
Since carbohydrate intake and its quality are known to influ- woman was asked to complete a self-administered questionnaire
ence glucose homeostasis and have been linked to T2DM [5,6], on previous birth and demographic characteristics, medical history,
understanding the role of the quality and quantity of carbohy- lifestyle, employment, annual family income, and family history of
drate on GDM development may have important public health chronic diseases. Participants were also face to face interviewed by
implications. To date, prospective observational studies have trained investigators with respect to their diet (one food frequency
found that a diet during pre-pregnancy with a higher glycemic questionnaire (FFQ) covering consumption over the 12 months
load (GL) [7] is related to higher GDM risk among US women before pregnancy and one 24-h dietary recall) and physical activity
[7,8]. Moreover, beneficial impacts of a lower glycemic index (GI) (one questionnaire covering physical activity and sedentary
diet on obstetric or fetal outcomes were reported from inter- behavior over the 12 months before pregnancy and one covering
vention studies in Canada [9], Australia [10e12] and Malaysia physical activity and sedentary behavior since the beginning of the
[13]. According to the diagnostic criteria of 2011 issued by the pregnancy). A 24-h dietary recall, a FFQ addressing consumption
Ministry of Health (MOH) China [14], which endorsed the new during the previous 12 weeks, and a physical activity questionnaire
criteria by the IADPSG [15], the prevalence of GDM in China was were also administered at P2 and P3.
14.8% in 2013 [16], which is higher than in Europe/USA [17,18]. Anthropometric measures (body weight before pregnancy and
Due to the paucity of existing data on the link between maternal during pregnancy), clinical measures, as well as information on
diet and maternal glucose homeostasis during pregnancy in current and past pregnancy outcomes, complications, and infant
Asian populations, additional studies are needed. abnormalities recorded in the Medical Birth Registry were linked to
To date, all cohort studies analyzing the link between maternal the study database. The study was approved by the Ethics Com-
diet and GDM risk addressed dietary intake in only one time mittee of the Sichuan University. At P1, all participants provided
window, i.e. either before pregnancy [7,8,19e22] or during preg- written informed consent for all examinations as well as for linkage
nancy [23e25]. In terms of public health, it would however of their data from the Medical Birth Registry and the data of their
be intriguing to determine whether there is a critical time win- offspring from the health registries.
dow for dietary carbohydrate intake, i.e. whether pre-pregnancy From 2014 to 2017, 10,126 pregnant women were recruited into
intake or intake across the two pregnancy trimesters are the NPGSC Study. The Ministry of Health China recommends
most relevant in terms of a potential diagnosis of GDM. In addi- screening by fasting plasma glucose (FPG) test at the first prenatal
tion, it is of interest to elucidate whether changes in dietary car- visit to rule out previously undiagnosed preexisting diabetes. Of the
bohydrate quality over the course of pregnancy already translate 10,126 women, 109 women with T2DM mellitus in any previous
into concurrent changes in glucose homeostasis among those pregnancy were excluded so as to address relevance of carbohy-
women who remain free of GDM. Such an analysis requires drate quality in healthy (non-risk) pregnancies. To be included in
repeated simultaneous measurements of both diet and glucose this analysis, participants had to have delivered a live, singleton
homeostasis. baby and to have answered the first general questionnaire, which
Therefore, the aim of this study was to use data from the resulted in 9825 women. Of them, 9553 had provided three FFQs
Nutrition in Pregnancy and Growth in Southwest China (NPGSC) and three 24-h dietary recalls, while 9809 women had two FFQs
cohort study to prospectively investigate whether dietary GI, GL and two recalls. We further excluded women with an implausible
and fiber intakes during potentially critical periods, i.e. the year energy intake (
500 and 3500 kcal/day) (n ¼ 129), as well as
preceding pregnancy, the 1st or the early 2nd trimester were those with missing information on parity, maternal occupation or
associated with GDM onset, and to investigate the concurrent annual family income (n ¼ 107). In total, 9317 women with com-
relevance of dietary GI, GL or fiber intakes for glucose homeo- plete information were included to investigate whether dietary GI,
stasis over the course of pregnancy (1st to 3rd trimester). GL and fiber intakes during potentially critical periods were asso-
ciated with GDM onset (Supplemental Figure S1). To investigate the
2. Materials and methods concurrent relevance on glucose homeostasis over the course of
pregnancy, data from 7823 participants without GDM were
2.1. Study sample considered; 9585 women both with and without GDM were
included in the sensitivity analysis.
The Nutrition in Pregnancy and Growth in South China (NPGSC)
study is a prospective cohort study initiated in January 2014 to 2.2. Nutrition assessment
investigate the relevance of maternal nutrition before and during
pregnancy for health outcomes among mothers and offspring [26]. Since only FFQ data can be used to estimate dietary intake
Briefly, using a sampling design stratified by urban and rural lo- before pregnancy, the present analysis addressing the relevance of
cations, a representative sample of pregnant women was drawn dietary intake during three time windows is based on dietary data
2792
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

collected by FFQs only. In a validation study (n ¼ 89), total carbo- gestation. A diagnosis of GDM was made if any one of the following
hydrate, GI, GL and dietary fiber estimated by FFQ acceptably values was met or exceeded: 0 h (fasting), 5.1 mmol/L; 1 h,
correlated with intakes assessed by two 3-day weighed diet records 10.0 mmol/L; and 2 h, 8.5 mmol/L [14]. Mothers diagnosed with
(total carbohydrate, r ¼ 0.69; GI, r ¼ 0.72; GL, r ¼ 0.68; fiber, GDM received dietary treatment followed by insulin treatment for
r ¼ 0.70) (Supplemental Tables S2 and S3). At each P1, P2 and P3, a management if the diet alone was insufficient to restore glucose
modified validated 128-item FFQ [30] was used to inquire how homeostasis.
often on average (never to 5 times/d) the participants had
consumed the respective food groups. The assessment of white rice, 2.4.2. Quality control
brown rice, red rice, glutinous rice, wheat noodle, steamed bread, All study laboratories successfully completed a standardization
bread, whole grain foods, potatoes, cakes, vegetables, fruits, sub- and certification program. The coefficient of variation within and
tropical fruits, dairy and dairy products, soybeans and its products, between study laboratories was less than 5% for each marker. All
meat (pork, beef, chicken or lamb), eggs, fish and shrimp, and laboratory equipment was calibrated and blinded duplicate sam-
beverages (including drinking water, mineral water, tea and herbal ples were used. All data had double entries in the database. Par-
tea, lemonades, fruit drinks (diluted and sugar-sweetened fruit ticipants were informed about all clinical data within 36 h after
juices), ice teas, soft drinks and sports drinks) was performed using collection.
standard serving sizes. To this end, participants were offered a
range of different serving sizes. Visual aids such as standard serving 2.5. Other information and clinical examinations
bowls, plates and glasses were displayed to the participants to
improve the accuracy of estimated portion sizes. The frequency and Demographic characteristics, medical history, lifestyle (e.g.
amount of consumption of each food or beverage per unit of time smoking behavior, alcohol consumption, tea/coffee consumption)
were converted into food intake per day. Total energy and nutrient of each woman were collected at P1. Self-reported pregravid weight
intakes were calculated using the continuously updated in-house was recorded on the day of registration. Body weight was measured
nutrient database [31] reflecting the composition of Chinese with an ultrasonic meter (Dingheng, Zhengzhou, China) to the
Foods [32]. nearest 100 g at enrollment and at regular intervals (in 4-week
intervals from enrollment to week 25, every 2 weeks until week
2.3. Estimation of dietary fiber intake, the glycemic index and 33, and weekly thereafter until birth) (Supplemental Figure S1).
glycemic load Gestational age (GA) was assessed during the first ultrasound
scan (Eub 5500, Hitachi; Eub 7500, Hitachi; Logiq E9, GE) on the day
The mean daily fiber was the sum of the fiber content of all foods of registration. GA was estimated by combining ultrasonography
consumed. The GI values for single food items inquired by the FFQ data with self-reports on the last menstrual period: if both mea-
were assigned according to a standardized procedure [33]. Foods sures were available and there was agreement (±14 days), self-
were either assigned 1) a published GI, 2) a published GI of a close reported data were used, otherwise ultrasound data were used.
match, or 3) the dietary GI calculated from the GI values of the
food's ingredients using recipes available in the in-house database. 2.6. Statistical analysis
The carbohydrate (CHO) content of the food was the principle
consideration when matching a particular food with one listed in SAS® procedures (version 9.2, SAS Inc, Cary, NC) were used for all
the tables. In addition, preference was given to local GI values: data analyses. All analyses were performed with a significance level
36.9% of direct or close matches had been measured in China, 17.2% of p < 0.05.
in Hong Kong, 6.9% in Singapore, 5.2% in Japan and 4.6% in South Because we were interested in investigating critical time win-
Korea. Foods containing mainly fat or protein with a CHO content dows regarding the relevance of dietary GI, GL or fiber intakes for
below 5 g/100 g were assigned a GI of 0. Dietary GL for each food GDM onset, we conducted separate analyses using dietary data
was calculated by multiplying the carbohydrate content of each collected on pre-pregnancy intakes and intakes during the 1st and
food by its GI values and its frequency of consumption [34]. Dietary 2nd trimester. Dietary GI, GL, fiber intake and all nutrient intakes in
GL for each participant was estimated by summing the values of GL these 3 different time periods were expressed as residuals from
for all food items. Overall dietary GI for a participant was obtained their regression on energy intake. Energy-adjusted residuals of
by dividing GL by the amount of carbohydrates consumed. dietary GI, GL and fiber intake were grouped into tertiles to illus-
trate their associations with the risk for GDM using Cox propor-
2.4. Outcome ascertainment tional hazards analysis. In the basic models, dietary GI, GL or fiber
intake was the independent predictors. The following variables
Fasting venous blood samples were taken at each P1, P2 and P3 potentially affecting these associations as confounders were
after an overnight fast of at least 8 h. Venous blood specimens were considered: maternal age, parity, location, family history of dia-
collected using EDTA tubes, or tubes containing anticoagulant so- betes, maternal education level, maternal occupation, family in-
dium fluoride, then immediately centrifuged and stored below 4  C come and pre-pregnancy BMI, physical activity, smoking/passive
for subsequent measurements. FPG was measured within 2 h using smoking before or during pregnancy, alcohol consumption before
glucose oxidase methods. Fasting plasma insulin concentrations or during pregnancy, pregravid BMI adjusted residuals of gesta-
(mIU/L) were determined within 4 h using chemiluminescence tional weight gain during pregnancy, energy intake, animal protein,
enzyme immunoassay. HbA1c was measured within 12 h using saturated fat, polyunsaturated fat, magnesium, iron, dietary GI or
high-performance liquid chromatography (D-10™ Hemoglobin GL adjusted residuals of sugar-sweetened beverages (SSB), and
Analyzer, Bio-Rad, CA). Based on these values, insulin resistance dietary glycemic index and dietary fiber (vice versa). Each potential
(HOMA-IR) was calculated according to the Wallace formula [35]. confounder was initially considered separately and included if it
substantially modified the association of dietary GI/GL/fiber with
2.4.1. Diagnostic criteria for GDM GDM or significantly predicted the outcome variable. In model 1,
After FPG test was performed at the first prenatal visit to exclude we considered perinatal factors, in model 2 additional adjustment
pre-existing diabetes (7.00 mmol/L), participants underwent a 2- for socioeconomic factors was performed. In a final model, we
h 75-g oral-glucose-tolerance test (OGTT) at 24e28 weeks of controlled for nutritional confounding. Hazard ratios (HR) were
2793
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

calculated for the respective tertiles (reference: tertile 1) and a test least 12 years. 1751 women developed GDM during the study
for trend was performed using the respective continuous variables. period, they were 27.1 (4.7) years old, significantly older than
To investigate the concurrent relevance of dietary GI, GL or fiber women without GDM. In addition, 53% of them came from urban
intakes for glucose homeostasis over the course of pregnancy, areas, and 58% of them had a school education of at least 12 years
linear mixed effects regression models (PROC MIXED in SAS), (Table 1).
including both fixed and random effects, were used to construct Anthropometric, nutritional, and biochemical characteristics
longitudinal models of FPG, HbA1c or HOMA-IR trajectories over the before and during pregnancy are shown in Table 2. Women clearly
course of pregnancy among the participants. A repeated statement gained more body weight in the 2nd and 3rd trimesters and had a
was used to account for the lack of independence that exists be- higher physical activity level before pregnancy. The sample
tween repeated observations (P1, P2 or P3) on the same person. consumed more energy in the 2nd and the 3rd trimesters, of which
Dietary GI, GL, or fiber intake and time (time, time2) were the more energy stemmed from protein and fat and less from carbo-
principle fixed effects, and maternal age group (20e25 years, hydrate in the 2nd trimester. Additionally, their dietary GI was
26e30 years, 31e35 years and 36e40 years) was a random effect. notably lower before pregnancy than during pregnancy.
The first routine ultrasound examination was considered the A higher dietary GI each before pregnancy, during the 1st
baseline visit, i.e. time ¼ 0. These analyses yielded two regression trimester and during the 2nd trimester was associated with higher
coefficients representing the following: 1) prospective estimate e risks for GDM: participants in the highest GI tertiles before preg-
the slope of these dietary variables at P1 on the change in FPG, nancy, in the 1st trimester, or in the 2nd trimester had an
HbA1c or HOMA2-IR until the end of the study period, i.e. at P3, and approximately 12%, 25% or 29% higher risk respectively for GDM
2) concurrent estimates representing the slope of the change in than those in the lowest GI tertile (Table 3, p for trend
0.005).
these dietary variables over the study period (i.e. from P1 to P3) on Similarly, women in the highest GL tertile before pregnancy, in the
the concurrent change in FPG, HbA1c or HOMA-IR. The change in 1st trimester, or in the 2nd trimester had an approximately 15%,
dietary GI, GL or fiber intake over the study period was calculated 23% or 25% higher risk for GDM than those in the lowest GL tertile
by subtracting baseline intake from the intake at each trimester. (p for trend
0.02). In contrast, women with highest fiber intake
All dietary variables were expressed as standard deviation before pregnancy, in the 1st trimester, or in the 2nd trimester had
scores (mean ¼ 0, standard deviation ¼ 1). This approach allowed an approximately 11%, 17%, or 18% lower risk for GDM respectively
comparison of all unadjusted models with the respective energy- (p for trend
0.03). In addition, participants with the combination
adjusted models, i.e. standard deviation scores were used in of a high dietary GI and a low fiber intake before pregnancy, in the
models without energy adjustment and studentized (mean ¼ 0, 1st trimester, or in the 2nd trimester had an approximately 17%,
standard deviation ¼ 1) residuals [36] in models adjusting for en- 32%, or 36% higher risk for GDM than those with the lowest GI and
ergy. The fixed effects of age, parity, location, maternal education highest fiber intake (Supplemental Figure S2). Furthermore, SSB
level, maternal occupation, monthly personal income, pregravid intake during pre-pregnancy, in the 1st trimester or in the 2nd
BMI, physical activity, smoking, passive smoking during pregnancy, trimester was associated with higher GDM risk (Supplemental
alcohol consumption during pregnancy and gestational weight gain Table S4).
during pregnancy were also considered, as these variables could Table 4 presents the prospective association between the di-
potentially affect the association between the principle dietary etary GI, GL and fiber intake in the 1st trimester for glucose ho-
variables and FPG, HbA1c or HOMA-IR. Additionally, other nutri- meostasis in the 3rd trimester (1st column) as well as the relevance
tional variables including animal protein, SSB, dietary GI (in fiber of changes in the dietary GI, GL and fiber intake between 1st
model) or fiber intake (in GI or GL models) were included. The trimester and 3rd trimester for concurrent changes in glucose ho-
following three terms were considered for each potential dietary meostasis over the course of pregnancy (2nd column) among 7823
confounder: intake at baseline, intake at baseline x time, and participants without GDM. Higher dietary GI, GL and lower fiber
change in intake during the study period. Only those variables that intakes in the 1st trimester were prospectively associated with
substantially modified (a change in the estimate  10%) the asso- higher levels of FPG, HAb1C or HOMA-IR in the 3rd trimester, e.g.
ciation of the principle dietary variables with FPG, HbA1c or HOMA- per each unit increase in dietary GI in the 1st trimester FPG levels in
IR in the basic models, or significantly predicted the outcome var- the 3rd trimester were higher by approximately 0.281 mmol. In
iable or improved the fit statistic (Akaike's information criterion: addition, increases in dietary GI, GL and a decrease in fiber intake
AIC) were included in the subsequent multivariate models. over the course of the pregnancy were associated with concur-
To test the robustness of our results, we re-ran our analyses of rently increased FPG, HbA1C or HOMA-IR values (decreased in
the potential relevance of dietary GI, GL or fiber intake on glucose relation to fiber) during the pregnancy, e.g. for every SD increase in
homeostasis including participants both with and without GDM. To dietary GI from 1st trimester to 3rd trimester, FPG increased by
evaluate the joined effect of dietary GI and fiber intake before approximately 0.217 mmol during this same time period. Similar
pregnancy, in the 1st trimester, or in the 2nd trimester on GDM, we results were observed in the sensitivity analysis among all women
separately cross-classified our participants into tertiles of dietary GI with GDM (n ¼ 1762, Supplemental Table S1).
and tertiles of fiber intake. Furthermore, SSB intake (%energy)
before pregnancy, in the 1st trimester, or in the 2nd trimester was 4. Discussion
grouped into tertiles to illustrate their associations with the risk of
GDM using Cox proportional hazards analysis. In this prospective cohort study, consumption of a diet with
higher dietary GI, GL or lower fiber intake before or during preg-
3. Results nancy was related to higher risk for GDM among Chinese women.
In line with this, changes in dietary carbohydrate quality during
Participants who were excluded from the study sample pregnancy appear to affect the development of glucose
(n ¼ 809) did not differ in age, location and pregravid BMI from homeostasis.
those who were included (n ¼ 9317). Approximately 48% of women GDM is a disease of impaired glucose homeostasis [1]. A higher
included in the present analysis who did not develop GDM came dietary GI or GL is increasingly recognized as an important risk
from urban areas, they were 25.8 (2.8) years old at the time of the factor for disturbed glucose homeostasis also among Chinese in-
occurring pregnancy and 57% of them had a school education of at dividuals [5]. This notion builds on findings that the habitual
2794
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

Table 1
General characteristicsa of the study participants.

Characteristics GDM risk sample (n ¼ 9317)b Glucose homeostasis

sample (n ¼ 7823)b
Women without GDM (n ¼ 7566) Women with GDM (n ¼ 1751) Pc

Urban (n (%)) 3662 (48.4) 931 (53.2) 0.7 4052 (51.8)

Maternal age (years) 25.8 (2.8) 27.1 (4.7) 0.04 26.2 (3.5)
Pregravid BMI (kg/m2) 21.0 (2.6) 21.5 (2.8) 0.1 21.3 (2.5)
Gestational diabetes mellitus (n (%)) 0 (0) 1751 (100) <0.0001 0 (0)
Single mother (n (%)) 204 (2.7) 38 (2.2) 0.5 179 (2.3)
Self-reported passive smoking during pregnancy (n (%)) 1704 (22.5) 402 (22.9) 0.8 1807 (23.1)
Self-reported active smoking during pregnancy (n (%)) 67 (0.9) 16 (0.9) 0.3 63 (0.8)
Self-reported alcohol drinking during pregnancy (n (%)) 97 (1.3) 24 (1.4) 0.2 95 (1.2)
Multivitamin use (n (%)) 5439 (71.9) 1325 (75.7) 0.6 6018 (76.9)
High education leveld (n (%)) 4315 (57.0) 1023 (58.4) 0.9 4529 (57.9)
Continue to work until maternity leave (n (%)) 4344 (57.4) 1087 (62.1) 0.8 4655 (59.5)
Moderate personal monthly incomee (n (%)) 3871 (51.2) 937 (53.5) 0.9 3974 (50.8)
Previous medical history and family history
Gestational diabetes mellitus (n (%)) 206 (2.7) 173 (9.9) 0.3 305 (3.9)
Polycystic ovarian syndrome (n (%)) 116 (1.5) 33 (1.9) 0.2 132 (1.7)
Primiparous (n (%)) 5529 (73.1) 918 (52.4) 0.1 5374 (68.7)
Family history of diabetes (n (%)) 1840 (24.3) 498 (28.4) 0.6 1921 (24.6)
a
Values are means (SD) or frequencies.
b
9317 women were included in the analysis of GDM risk, 1751 women developed GDM during the study period. 7823 participants were included in the analysis of glucose
homeostasis development.
c
Test for difference between the groups was performed, using t-test for normal distributed continuous variables and Chi-square test for categorical variables.
d
School education at least 12 years.
e
Personal income per month at least 3000 CNY (Chinese Yuan), which is considered an average level among the general population in Southwest China.

Table 2
Anthropometric, nutritional and biochemical dataa during pre-pregnancy the pregnancy trimesters in the GDM sample (n ¼ 9317).

Pre-pregnancy 1st trimester 2nd trimester 3rd trimester

Anthropometric data
Body weight (kg) 54.6 (4.9) 55.7 (5.1) 60.5 (5.3) 67.6 (6.3)
BMI (kg/m2) 21.1 (2.8) 21.5 (3.2) 23.4 (2.9) 26.1 (2.8)
Gestational weight gain during each trimester (kg) e 1.0 (3.1) 5.0 (3.2) 7.2 (3.1)
Physical activityb (MET-h/wk) 18.5 (8.6) 10.9 (8.5) 15.2 (9.5) 12.5 (10.2)
Daily nutritional datac
Total Energy (kcal) 1662 (512) 1632 (542) 2012 (578) 2079 (582)
Carbohydrate (% of energy) 53.3 (10.7) 53.7 (9.6) 49.6 (10.5) 50.2 (8.9)
Protein (% of energy) 15.8 (4.2) 17.2 (3.9) 18.3 (4.5) 17.6 (3.5)
Animal protein (g) 38.9 (12.8) 41.3 (16.7) 55.2 (17.5) 56.3 (17.9)
Fat (% of energy) 30.9 (5.9) 29.1 (5.2) 32.1 (5.6) 32.2 (5.8)
Saturated fat (g) 11.5 (5.1) 12.6 (6.0) 14.0 (6.6) 14.3 (6.5)
Polyunsaturated fat (g) 7.5 (6.4) 10.5 (5.6) 11.5 (5.4) 11.8 (5.5)
Glycemic index 62.6 (5.7) 70.3 (6.5) 68.1 (5.5) 67.8 (5.7)
Glycemic load (g) 127 (28) 135 (33) 141 (17) 145 (25)
Total fiber (g) 13.9 (4.5) 8.9 (3.8) 12.8 (3.7) 14.1 (4.3)
Grain (including white rice) (g) 375 (48) 319 (42) 336 (42) 356 (48)
White rice (g) 182 (39) 98 (59) 228 (51) 232 (41)
Tropical and subtropical fruit (g) 131 (23) 168 (31) 221 (27) 165 (18)
Cookie and cake (g) 51 (11) 142 (32) 115 (22) 92 (17)
Sugar-sweetened beverage (ml) 365 (26) 386 (35) 262 (33) 229 (31)
Meat (g) 91 (17) 87 (29) 138 (19) 140 (21)
Eggs (g) 45 (7) 40 (5) 59 (5) 56 (6)
Dairy and dairy products (g) 218 (29) 260 (23) 378 (28) 369 (23)
Fish and shrimp (g) 27 (4) 20 (7) 76 (5) 62 (4)
Magnesium (mg) 224.9 (69.9) 233.6 (76.4) 276.7 (75.3) 268.7 (76.9)
Iron (mg) 12.7 (3.3) 12.8 (3.7) 15.2 (3.9) 16.7 (3.8)
Biochemical measure
FPG (mmol/L) e 4.6 (2.4) 4.7 (3.2) 4.8 (1.8)
HbA1C (%) e 5.3 (1.3) 5.1 (0.6) 5.4 (0.5)
Insulin (mIU/mL) e 11.6 (6.3) 14.8 (20.3) 24.8 (22.4)
HOMA-IRd e 1.5 (0.6) 1.7 (0.9) 2.2 (0.8)
a
Values are means (SD).
b
Metabolic equivalent hours of activity per week.
c
All nutritional data represent crude values.
d
Calculated according to the Wallace formula.

2795
 X. Zhang, Y. Gong, K. Della Corte et al.
Table 3
Hazard ratioa for GDM by tertiles of dietary glycemic index, dietary glycemic load and dietary fiber in different periods in the GDM sample (n ¼ 9317).

Tertiles of dietary glycemic index

Pre-pregnancy 1st trimester 2nd trimesterb

c c c d c c c d
1 (48.2e59.1) 2 (59.4e65.5) 3 (66.1e73.2) p for trend 1 (53.5e66.3) 2 (66.8e76.2) 3 (76.8e81.7) p for trend 1 (56.2e65.9)c 2 (66.3e73.8)c 3 (74.1e82.9)c p for trendd

model 1 1 1.06 (1.01, 1.12) 1.11 (1.03, 1.21) 0.01 1 1.14 (1.07, 1.30) 1.23 (1.12, 1.35) 0.01 1 1.23 (1.13, 1.36) 1.30 (1.08, 1.51) 0.01
model 2 1 1.07 (1.02, 1.13) 1.12 (1.02, 1.20) 0.02 1 1.15 (1.08, 1.31) 1.26 (1.17, 1.34) 0.01 1 1.24 (1.11, 1.35) 1.31 (1.12, 1.49) 0.008
model 3 1 1.08 (1.03, 1.14) 1.12 (1.03, 1.19) 0.01 1 1.17 (1.06, 1.29) 1.25 (1.20, 1.33) 0.008 1 1.24 (1.15, 1.34) 1.29 (1.21, 1.48) 0.005

Tertiles of dietary glycemic load

Pre-pregnancy 1st trimester 2nd trimesterb
1 (87.3e103.8) 2 (104.5e128.6) 3 (129.2e164.1) p for trend 1 (97.9e112.8)c 2 (113.1e160.7)c 3 (161.3e174.2)c
c c c d
p for trendd 1 (111.6e132.8)c 2 (133.6e150.3)c 3 (150.8e175.8)c p for trendd
2796

model 1 1 1.05 (1.02, 1.10) 1.15 (1.04, 1.26) 0.03 1 1.16 (1.05, 1.29) 1.22 (1.10, 1.39) 0.02 1 1.20 (1.05, 1.36) 1.26 (1.07, 1.43) 0.01
model 2 1 1.06 (1.02, 1.11) 1.14 (1.03, 1.23) 0.03 1 1.18 (1.08, 1.30) 1.25 (1.09, 1.40) 0.01 1 1.21 (1.04, 1.38) 1.27 (1.10, 1.45) 0.01
model 3 1 1.07 (1.02, 1.12) 1.15 (1.08, 1.23) 0.02 1 1.19 (1.10, 1.29) 1.23 (1.10, 1.45) 0.01 1 1.21 (1.06, 1.37) 1.25 (1.11, 1.40) 0.01

Tertiles of dietary fiber intake

Pre-pregnancy 1st trimester 2nd trimesterb
1 (4.6e11.8)c 2 (12.1e15.2) c
3 (15.3e19.8) c
p for trend d
1 (4.5e6.8)c 2 (7.0e11.5)c 3 (11.7e14.5)c p for trendd 1 (7.2e11.5)c 2 (11.7e15.9)c 3 (16.1e19.2)c p for trendd

model 1 1 0.93 (0.86, 0.98) 0.88 (0.80, 0.95) 0.05 1 0.90 (0.84, 0.95) 0.84 (0.76, 0.91) 0.04 1 0.87 (0.80, 0.95) 0.83 (0.76, 0.91) 0.03
model 2 1 0.92 (0.85, 0.97) 0.89 (0.81, 0.96) 0.04 1 0.91 (0.86, 0.94) 0.85 (0.79, 0.92) 0.02 1 0.86 (0.79, 0.94) 0.82 (0.73, 0.90) 0.02
model 3 1 0.92 (0.85, 0.98) 0.89 (0.83, 0.94) 0.03 1 0.89 (0.83, 0.95) 0.83 (0.75, 0.91) 0.01 1 0.87 (0.80, 0.93) 0.82 (0.73, 0.91) 0.01
a
Data are HR with 95% confidence interval, model 1 adjusted for maternal age, parity and pregravid BMI, pregravid BMI adjusted residuals of gestational weight gain until week 12 (for 1st trimester) or pregravid BMI adjusted
residuals of gestational weight gain between 13 and 23 weeks (for 2nd trimester); model 2 additionally adjusted for maternal occupation and family income; model 3 additionally adjusted for energy intake, dietary GI or GL
adjusted residuals of sugar-sweetened beverage, energy adjusted residuals of animal protein, and energy adjusted residuals of glycemic index (in the fiber model) and fiber (in the GI or GL models) pre-pregnancy, in the 1st
trimester and the 2nd trimester, separately.
b
Dietary intakes in the 2nd trimester were reported from gestational weeks 13e22.
c
Range values are minemax in tertiles.
d
P for trend refer to the p values obtained in linear regression models, using dietary glycemic index, dietary glycemic load or dietary fiber intake as continuous variable.

Clinical Nutrition 40 (2021) 2791e2799

X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

Table 4
Mixed modelsa on the prospective relation between dietary glycemic index, dietary glycemic load and dietary fiber intake in the 1st trimester and FPG (mmol/L), HbA1C (%) and
HOMA-IR in the 3rd trimester as well as the association between changes in carbohydrate quality and concurrent changes in glucose homeostasis in the glucose homeostasis
sample (n ¼ 7823).

Prospective relation: Nutritional intakeb Concurrent association: Change in FPG between AIC3
in the 1st trimester in relation to FPG in the 3rd trimester 1st trimester and 3rd trimester per increase in intake by 1 SDb

b (SE) p b (SE) p

Glycemic index
Model 1d 0.299 (0.051) 0.02 0.228 (0.052) 0.03 6821
Model 2e 0.287 (0.045) 0.02 0.220 (0.048) 0.02 6510
Model 3f 0.281 (0.046) 0.01 0.217 (0.046) 0.01 6215
Glycemic load (g/d)
Model 1d 0.249 (0.040) 0.01 0.201 (0.039) 0.02 6762
Model 2e 0.246 (0.041) 0.01 0.195 (0.043) 0.01 6513
f
Model 3 0.250 (0.038) 0.007 0.196 (0.042) 0.01 6061
Total fiber (g/d)
Model 1d 0.116 (0.029) 0.03 0.129 (0.028) 0.03 6781
Model 2e 0.115 (0.026) 0.02 0.126 (0.027) 0.02 6476
Model 3f 0.112 (0.025) 0.02 0.125 (0.024) 0.02 6102

Prospective relation: Nutritional intakeb Concurrent association: Change in HbA1C between

in the 1st trimester in relation to HAb1C in the 3rd trimester 1st trimester and 3rd trimester per increase in
intake by 1 SDb
b (SE) p b (SE) p AICc

Glycemic index
Model 1d 0.216 (0.050) 0.03 0.172 (0.049) 0.03 6737
Model 2e 0.212 (0.047) 0.02 0.167 (0.046) 0.02 6403
f
Model 3 0.213 (0.048) 0.02 0.161 (0.045) 0.02 6125
Glycemic load (g/d)
Model 1d 0.260 (0.046) 0.02 0.198 (0.044) 0.04 6831
Model 2e 0.257 (0.044) 0.02 0.202 (0.046) 0.02 6365
Model 3f 0.255 (0.045) 0.01 0.205 (0.047) 0.01 6016
Total fiber (g/d)
d
Model 1 0.109 (0.032) 0.06 0.091 (0.029) 0.04 6711
Model 2e 0.112 (0.030) 0.04 0.094 (0.027) 0.03 6517
f
Model 3 0.109 (0.029) 0.03 0.097 (0.026) 0.02 6206

Prospective relation: Nutritional intakeb Concurrent association: Change in HOMA-IR

in the 1st trimester in relation to HOMA-IR in the 3rd trimester between 1st trimester and 3rd trimester per
increase in intake by 1 SDb
b (SE) p b (SE) p AICc

Glycemic index
Model 1d 0.040 (0.032) 0.03 0.021 (0.027) 0.03 10,012
Model 2e 0.037 (0.034) 0.03 0.023 (0.030) 0.02 9831
Model 3f 0.035 (0.032) 0.02 0.025 (0.032) 0.02 9715
Glycemic load (g/d)
d
Model 1 0.049 (0.041) 0.03 0.043 (0.032) 0.03 10,027
Model 2e 0.047 (0.043) 0.02 0.039 (0.029) 0.01 9926
f
Model 3 0.043 (0.039) 0.01 0.037 (0.026) 0.01 9769
Total fiber (g/d)
Model 1d 0.030 (0.031) 0.05 0.027 (0.031) 0.04 10,273
Model 2e 0.032 (0.029) 0.04 0.029 (0.028) 0.03 9882
Model 3f 0.029 (0.030) 0.03 0.028 (0.030) 0.02 9747
a
Models contain a random statement for the age group level with an unstructured covariance and a repeated statement with a first-order autoregressive covariance.
b
Expressed as SDs. 1 SD in dietary glycemic index was equivalent to approximately 5.3e6.4; 1 SD in dietary glycemic load was equivalent to approximately 18.3e31.1 g; and
1 SD in fiber intake was equivalent to approximately 3.3e3.9 g.
c
AIC: Akaike's Information Criterion (smaller is better).
d
Model 1: unadjusted model, contains maternal age.
e
Model 2: additionally adjusted for parity, maternal occupation, family income, pregravid BMI and pregravid BMI adjusted residuals of gestational weight gain (by itself and
in interaction with time).
f
Model 3: additionally adjusted for energy intake, dietary GI or GL adjusted residuals of sugar-sweetened beverage, energy adjusted residuals of animal protein, and energy
adjusted residuals of glycemic index (in the fiber model) and fiber (in the GI or GL models) (each as residual intake at the 1st trimester, intake at the 1st trimester x time and
difference in intake between 1st trimester and 3rd trimester).

consumption of a diet with a high GI or GL elicits enhanced post- that intake during early and mid-pregnancy is relevant for GDM
prandial insulin responses increasing the demand on b-cells, sub- risk than pre-pregnancy intake. This may be linked to the increased
sequently contributing to b-cell exhaustion and failure [37]. With insulin secretion and decreased insulin sensitivity during preg-
respect to GDM risk, our findings extend previous findings [7] nancy which we also observed in our study and which has been
confirming a relevance of GL before pregnancy for GDM risk also for reported in the literature: glucose tolerance and peripheral sensi-
pregnant women from South China and stressing the role of dietary tivity to insulin and hepatic basal glucose production may be
GI and dietary fiber. Our results indicate that the benefits of a low normal in the 1st trimester [38], however, in later gestation,
dietary GI are additive with those of a high fiber intake among maternal adipose tissue depots decline to support fetal growth,
Chinese women for GDM prevention. Moreover, our data suggest hence postprandial free fatty acid levels increase and insulin

2797
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

mediated glucose disposal worsens by 50e70% compared with pre- age groups, i.e. our participants may have been more “health
pregnancy [38]. Hence, our findings suggest that avoidance of a conscious” than the general pregnant population despite compa-
high dietary GI/GL before and during pregnancy should be advo- rable economic and educational status. In addition, it would be of
cated. In view of the substantial relevance of GDM for medium- and interest to investigate whether higher dietary GI or lower fiber
long-term health of both mothers and offspring, the high preva- intake before and during pregnancy affects offspring health (e.g.
lence of GDM in the Chinese population and the high dietary GI and birth weight); this will be addressed during the ongoing follow-up
GL, incorporation of standard advice and monitoring of the dietary of NPGSC. Finally, residual confounding by factors beyond the di-
GI/GL during routine pregnancy visits is urgently needed. etary and lifestyle factors considered remains a possibility.
In addition, using the powerful technique of repeated-measures The strengths of our study include the repeated, detailed mea-
regression accounting for within-person correlations, we surement of both dietary and clinical data allowing a comparative
confirmed that changes of dietary GI or GL over the course of assessment of the relevance of dietary intake in three potentially
pregnancy relate to concurrent changes in glucose homeostasis, critical time windows. In particular, the analysis of associations
indicating that adverse diet-induced changes in glucose homeo- between change in dietary intake and concurrent changes in
stasis can already be observed among pregnant women not glucose homeostasis has been proposed to possess features of a
developing GDM. Of note, a temporal relation between concurrent quasi-experimental design, in which some participants are
changes in carbohydrate quality and HbA1c levels may not hold assigned to a change in dietary glycemic index, glycemic load, or
because HbA1c levels usually reflect an average glucose status over fiber intake and others are not [42].
the past 3 months. Moreover, relevance of carbohydrate quality for In conclusion, our study indicates that dietary GI, GL and fiber
glucose homeostasis was also seen for the smaller sample of intake before and during pregnancy affects glucose homeostasis of
women who developed GDM. pregnant Chinese women.
In the present study, dietary GI before pregnancy was in line
with values observed among other Chinese and Japanese pop- Disclaimer
ulations, yet at the upper end of ranges observed in European and
US populations [5], reflecting the fact that the main carbohydrate The interpretation and reporting of these data are the sole re-
source in South China is commonly white rice characterized by a sponsibility of the authors.
mediumehigh GI value. However, in our sample of pregnant
women, GI increased particularly in early-pregnancy when cakes,
Funding sources
cookies e both characterized by high GI values (72e91) e and
sweetened beverages were clearly consumed in higher frequencies
This study was funded by National Natural Science Foundation
and amounts, in part because these foods are thought to be effec-
of China (81673158).
tive against morning sickness in early-pregnancy in Chinese cul-
ture. The SSB contribution to GDM risk that was observed in our
study, especially during the first trimester, was notable and may Statement of authorship
have important public health implications.
Higher intakes of fiber have been reported to be beneficially G.C. and R.Z. conceived the project. G.C. and X.Z. performed the
associated with glucose homoeostasis in observational studies analyses. A.E.B. assisted with the statistical analysis and its inter-
conducted among adults [39]. In the only prospective study con- pretation. G.C. and A.E.B. wrote the manuscript. G.T., J.Z., L.Y., S.S.,
ducted in pregnant women, a higher fiber intake in pre-pregnancy D.Y. (Dagang Yang) and D.Z. coordinated the study centers, G.T. and
was associated with lower GDM risk [7]. Our data corroborate this J.Z. performed the initial data analyses. G.C. supervised the study.
and additionally suggest a role of dietary fiber for glucose All authors critically reviewed the manuscript for important intel-
homoeostasis particularly during pregnancy. lectual content.
In our population, there were 4.1% women diagnosed with GDM
during their previous pregnancies, which are clearly lower than the Conflict of interest
19% of women who developed GDM in the present pregnancy, or
other studies from China populations. The reasons for these AEB is a member of the International Carbohydrate Quality
incomparable values may lie in the different GDM diagnostic Consortium (ICQC). None of the authors have any personal or
criteria: until 2011, GDM was generally diagnosed based on the financial conflicts of interest.
National Diabetes Data Group recommendations [40], the MOH
China proposed to follow the IADPSG thresholds [15] since 2011.
Acknowledgements
Subsequent analyses [41] revealed that application of these resul-
ted in 18% of all women in pregnancy identified as having GDM.
All participates and their families in the NPGSC Study are
Our study has several limitations: GI values had to be calculated
gratefully acknowledged. We also thank all colleagues working in
for approximately 7% of the CHO-containing foods from GI values of
the NPGSC Study for their continuing valuable help.
their ingredients. While this procedure has been controversially
discussed, it has been shown that the dietary GI of a whole diet or
mixed meal can be accurately estimated from the GI values of the Appendix A. Supplementary data
individual foods it contains. Comparing the relevance of pre-
pregnancy intake with intake during pregnancy may be Supplementary data to this article can be found online at
hampered by differences in re-call bias due to the length of the https://doi.org/10.1016/j.clnu.2021.03.041.
recall period before pregnancy compared to that during pregnancy.
Post-prandial glycemia and fasting glycemia describe different References
pathways in the development of diabetes. Data on post-prandial
glycemia were not available in our study, which is also a limita- [1] Metzger BE, Buchanan TA, Coustan DR, de Leiva A, Dunger DB, Hadden DR,
et al. Summary and recommendations of the fifth international workshop-
tion. Moreover, the participation rate varied by areas and age conference on gestational diabetes mellitus. Diabetes Care 2007;30(Suppl
groups, with higher participation in urban areas and among older 2):S251e60.

2798
X. Zhang, Y. Gong, K. Della Corte et al. Clinical Nutrition 40 (2021) 2791e2799

[2] Group HSCR, Metzger BE, Lowe LP, Dyer AR, Trimble ER, Chaovarindr U, et al. [21] Bao W, Tobias DK, Olsen SF, Zhang C. Pre-pregnancy fried food consumption
Hyperglycemia and adverse pregnancy outcomes. N Engl J Med 2008;358: and the risk of gestational diabetes mellitus: a prospective cohort study.
1991e2002. Diabetologia 2014;57:2485e91.
[3] Getahun D, Fassett MJ, Jacobsen SJ. Gestational diabetes: risk of recurrence in [22] Bowers K, Tobias DK, Yeung E, Hu FB, Zhang C. A prospective study of pre-
subsequent pregnancies. Am J Obstet Gynecol 2010;203(467):e461e6. pregnancy dietary fat intake and risk of gestational diabetes. Am J Clin Nutr
[4] Farahvar S, Walfisch A, Sheiner E. Gestational diabetes risk factors and long- 2012;95:446e53.
term consequences for both mother and offspring: a literature review. [23] Osorio-Yanez C, Gelaye B, Qiu C, Bao W, Cardenas A, Enquobahrie DA, et al.
Expet Rev Endocrinol Metabol 2019;14:63e74. Maternal intake of fried foods and risk of gestational diabetes mellitus. Ann
[5] Cheng G, Xue H, Luo J, Jia H, Zhang L, Dai J, et al. Relevance of the dietary Epidemiol 2017;27:384e90. e381.
glycemic index, glycemic load and genetic predisposition for the glucose [24] Osorio-Yanez C, Qiu C, Gelaye B, Enquobahrie DA, Williams MA. Risk of
homeostasis of Chinese adults without diabetes. Sci Rep 2017;7:400. gestational diabetes mellitus in relation to maternal dietary calcium intake.
[6] Bhupathiraju SN, Tobias DK, Malik VS, Pan A, Hruby A, Manson JE, et al. Publ Health Nutr 2017;20:1082e9.
Glycemic index, glycemic load, and risk of type 2 diabetes: results from 3 large [25] Radesky JS, Oken E, Rifas-Shiman SL, Kleinman KP, Rich-Edwards JW,
US cohorts and an updated meta-analysis. Am J Clin Nutr 2014;100:218e32. Gillman MW. Diet during early pregnancy and development of gestational
[7] Zhang C, Liu S, Solomon CG, Hu FB. Dietary fiber intake, dietary glycemic load, diabetes. Paediatr Perinat Epidemiol 2008;22:47e59.
and the risk for gestational diabetes mellitus. Diabetes Care 2006;29: [26] Zhang X, Gong Y, Chen Y, Corte KD, Duan R, Tian G, et al. Nutrition in preg-
2223e30. nancy and growth in Southwest China (NPGSC) cohort: design, implementa-
[8] Bao W, Tobias DK, Hu FB, Chavarro JE, Zhang C. Pre-pregnancy potato con- tion, and characteristics. Paediatr Perinat Epidemiol 2020;34(6):724e33.
sumption and risk of gestational diabetes mellitus: prospective cohort study. [27] Statistical bureau of Guizhou: Guizhou statistical yearbook. Beijing: China
Br Med J 2016;352:h6898. Statistic Press; 2013-2016.
[9] Grant SM, Wolever TM, O'Connor DL, Nisenbaum R, Josse RG. Effect of a low [28] Statistical bureau of Sichuan: SiChuan statistical yearbook. Beijing: China
glycaemic index diet on blood glucose in women with gestational hyper- Statistic Press; 2013-2016.
glycaemia. Diabetes Res Clin Pract 2011;91:15e22. [29] Statistical bureau of Yunnan: Yunnan statistical yearbook. Beijing: China
[10] Louie JC, Markovic TP, Perera N, Foote D, Petocz P, Ross GP, et al. A randomized Statistic Press; 2013-2016.
controlled trial investigating the effects of a low-glycemic index diet on [30] Longde W. Report of China national nutrition and health survey 2002 (1):
pregnancy outcomes in gestational diabetes mellitus. Diabetes Care 2011;34: summary report. Beijing: People's Medical Publishing House; 2005.
2341e6. [31] Cheng G, Duan R, Kranz S, Libuda L, Zhang L. Development of a dietary index
[11] Markovic TP, Muirhead R, Overs S, Ross GP, Louie JC, Kizirian N, et al. Ran- to assess overall diet quality for Chinese school-aged children: the Chinese
domized controlled trial investigating the effects of a low-glycemic index diet children dietary index. J Acad Nutr Diet 2016;116:608e17.
on pregnancy outcomes in women at high risk of gestational diabetes mel- [32] Yue-xin Yang, Guang-ya Wang, Xing-chang Pan. China food composition.
litus: the GI baby 3 study. Diabetes Care 2016;39:31e8. Beijing: Peking University Medical Press; 2009.
[12] Moses RG, Barker M, Winter M, Petocz P, Brand-Miller JC. Can a low-glycemic [33] Buyken AE, Dettmann W, Kersting M, Kroke A. Glycaemic index and glycaemic
index diet reduce the need for insulin in gestational diabetes mellitus? A load in the diet of healthy schoolchildren: trends from 1990 to 2002,
randomized trial. Diabetes Care 2009;32:996e1000. contribution of different carbohydrate sources and relationships to dietary
[13] Shyam S, Arshad F, Abdul Ghani R, Wahab NA, Safii NS, Nisak MY, et al. Low quality. Br J Nutr 2005;94:796e803.
glycaemic index diets improve glucose tolerance and body weight in women [34] Wolever TM, Vorster HH, Bjorck I, Brand-Miller J, Brighenti F, Mann JI, et al.
with previous history of gestational diabetes: a six months randomized trial. Determination of the glycaemic index of foods: interlaboratory study. Eur J
Nutr J 2013;12:68. Clin Nutr 2003;57:475e82.
[14] Yang HX. Diagnostic criteria for gestational diabetes mellitus (WS 331-2011). [35] Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Dia-
Chin Med J 2012;125:1212e3. betes Care 2004;27:1487e95.
[15] International Association of D, Pregnancy Study Groups Consensus P, [36] Hu FB, Stampfer MJ, Rimm E, Ascherio A, Rosner BA, Spiegelman D, et al.
Metzger BE, Gabbe SG, Persson B, Buchanan TA, et al. International association Dietary fat and coronary heart disease: a comparison of approaches for
of diabetes and pregnancy study groups recommendations on the diagnosis adjusting for total energy intake and modeling repeated dietary measure-
and classification of hyperglycemia in pregnancy. Diabetes Care 2010;33: ments. Am J Epidemiol 1999;149:531e40.
676e82. [37] Buyken AE, Mitchell P, Ceriello A, Brand-Miller J. Optimal dietary approaches
[16] Zhu WW, Yang HX, Wei YM, Yan J, Wang ZL, Li XL, et al. Evaluation of the for prevention of type 2 diabetes: a life-course perspective. Diabetologia
value of fasting plasma glucose in the first prenatal visit to diagnose gesta- 2010;53:406e18.
tional diabetes mellitus in China. Diabetes Care 2013;36:586e90. [38] Catalano PM. Trying to understand gestational diabetes. Diabet Med: J Br Diab
[17] Brown FM, Wyckoff J. Application of one-step IADPSG versus two-step diag- Assoc 2014;31:273e81.
nostic criteria for gestational diabetes in the real world: impact on health [39] InterAct C. Dietary fibre and incidence of type 2 diabetes in eight European
services, clinical care, and outcomes. Curr Diabetes Rep 2017;17:85. countries: the EPIC-InterAct Study and a meta-analysis of prospective studies.
[18] Eades CE, Cameron DM, Evans JMM. Prevalence of gestational diabetes mel- Diabetologia 2015;58:1394e408.
litus in Europe: a meta-analysis. Diabetes Res Clin Pract 2017;129:173e81. [40] Carpenter MW, Coustan DR. Criteria for screening tests for gestational dia-
[19] Bao W, Bowers K, Tobias DK, Hu FB, Zhang C. Prepregnancy dietary protein betes. Am J Obstet Gynecol 1982;144:768e73. https://doi.org/10.1016/0002-
intake, major dietary protein sources, and the risk of gestational diabetes 9378(82)90349-0.
mellitus: a prospective cohort study. Diabetes Care 2013;36:2001e8. [41] Moses RG. New consensus criteria for GDM: problem solved or a pandora's
[20] Bao W, Bowers K, Tobias DK, Olsen SF, Chavarro J, Vaag A, et al. Prepregnancy box? Diabetes Care 2010;33:690e1. https://doi.org/10.2337/dc09-2306.
low-carbohydrate dietary pattern and risk of gestational diabetes mellitus: a [42] Pereira MA, Jacobs Jr DR. Sugar-sweetened beverages, weight gain and
prospective cohort study. Am J Clin Nutr 2014;99:1378e84. nutritional epidemiological study design. Br J Nutr 2008;99:1169e70.

2799