The aim, design and setting of the study
To investigate the impacts of maternal status on the health of mothers and their children, this prospective cohort study was conducted in China between 2014 and 2018. Details on the cohort design are described elsewhere [24]. In brief, singleton pregnant women who had lived without pre-gravid diabetes mellitus or hypertension from 27 hospitals and community health care centers were selected from Guizhou, Yunnan, and Sichuan provinces using a cluster randomized sampling method. Women were followed-up (prenatal care) in regular intervals until they gave birth, and a self-administered questionnaire on socio-demographics, marital and fertile histories was completed once they had been recruited. During prenatal cares, anthropometric measures and clinical examination were routinely conducted as described below and recorded in the medical birth registry. The study was approved by the Ethics Committee of Sichuan University. All participants provided written informed consent for all study content as well as for linkage of their data from the Medical Birth Registry.
Overall, 52,221 pregnant women were recruited from 27 study sites. After exclusion of 1096 participants with missing values of GDM, preeclampsia, stillbirth, and pre-gestational BMI, 51,125 women and their children were included for the final analysis.
Questionnaire
Detailed maternal information on maternal age, residence (urban/rural), occupation, school education, and personal income at enrollment (gestational week 9–11) were collected by trained investigators. Information on marital (unmarried/married) and fertile (counts of conception and childbirth) histories were also collected. Detailed instruction on collecting height and bodyweight with 3 months before pregnancy were carefully given to participants. Pre-gravid BMI was calculated as weight divided by height squared (kg/m2). Specific pre-gravid-BMIs were categorized according to the World Health Organization (WHO) BMI categorization criteria recommended for Asian population [25]: underweight (< 18.5 kg/m2), normal weight (18.5–22.9 kg/m2), overweight (23–24.9 kg/m2), and obese (≥25 kg/m2). Overweight and obese categories were combined into one category in this analysis, as limited number of obese participants (2.3%) were recruited [24].
Anthropometrical measurement
Pre-gravid bodyweight was self-reported, and maternal bodyweight during pregnancy was measured with an ultrasonic meter to the nearest 0.1 kg by medical professionals at registration and at regular intervals (prenatal cares were conducted monthly from recruitment to gestational week 25, every 2 weeks until gestational week 33, and weekly thereafter until birth). Maternal pre-gravid self-reported height was rechecked by medical professionals with electronic stadiometer to the nearest 0.1 cm at the first anthropometric measurement. All anthropometric measurements were performed twice for each participant, which dressed lightly and barefoot during measurements. The overall GWG was calculated by the difference between the bodyweight of a woman at childbirth (the latest measurement of bodyweight) and that within 3 months before the pregnancy. To explore the relevance of GWG on various outcomes in a more precise way, GWG was calculated for each trimester. The first two trimesters were combined as women usually gain a small amount of body weight in the first trimester [26]. Therefore, the weight gain during the first two trimesters was defined as the difference between the weight measured at gestational week 28 and the weight before pregnancy, and GWG in the last trimester was defined as the latest measurement of weight prior to delivery minus the last weight measured at gestational week 28. The weight difference between gestational week 23 and pre-gravid weight was used for GDM risk prediction, as GDM is diagnosed at gestational week 24–28. Pregnant women diagnosed with GDM received dietary treatment followed by insulin treatment for management if the diet alone was insufficient to restore glucose homeostasis. The weekly GWG rate for the first two trimester was calculated using the weight gain during the first two trimesters divided by 28, and the weight gain during the last trimester divided by number of weeks from the gestational week 28 to neonatal delivery was defined as the last trimester GWG rate. For neonates, bodyweight was measured to the nearest 100 g, and recumbent length was assessed to the nearest 0.1 cm with a stadiometer. Scales were checked and calibrated regularly.
Diagnostic criteria for adverse pregnancy outcomes
Measurements and diagnoses of adverse pregnancy outcomes were performed by professionals in each study center by standard procedure and acknowledged criteria.
Through a two-hour 75 g oral-glucose-tolerance test at gestational week 24–28, patients with GDM were identified if any of the following plasma glucose levels was met: 0 hour (fasting), ≥ 5.1 mmol/L; 1 hour, ≥ 10.0 mmol/L; and 2 hour, ≥ 8.5 mmol/L [27]. Blood pressure was measured by the mercurial blood pressure device. Women with preeclampsia were identified by the following criteria: hypertension (systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg) after 20 weeks of gestation in previously normotensive women, along with the presence of proteinuria [28]. Cesarean deliveries that were performed before or during labor, decided by either doctor or pregnant women, were both included.
Neonates who were delivered earlier than 37 weeks of gestation were defined as preterm delivery [29]. Stillbirth was identified as the death of the fetus after 20 weeks of gestation [30]. Newborn with birthweight of over 4000 g was considered as macrosomia neonate [31]. SGA and LGA were defined as sex- and gestational age-adjusted birth weight < 10th and > 90th percentile, respectively [32].
Statistical analysis
Mean and standard deviation were expressed for normally distributed continuous variables. Categorical variables were expressed as frequency and proportion. These summary statistics were calculated for all data and each subgroup (normal weight, underweight, and overweight/obese). Statistical models were constructed to estimate the lowest risk at which weight gain causing one of the following adverse events: GDM, preeclampsia, cesarean delivery, stillbirth, preterm delivery, SGA, LGA, and/or macrosomia. GAMs [17] with logit link were used to analyze the relationships between weight gain and adverse pregnancy outcomes, the following function was used for the model’s prediction.
$$\textrm{f}\left(\textrm{x}\right)=\textrm{s}\left(\textrm{Weight}\ \textrm{Gain}\right)+\textrm{s}\left(\textrm{Age}\right)+\left(\textrm{preBMI}\right)+\textrm{s}\left(\textrm{Height}\right)+\textrm{Gender},$$
where s() is smoothed spline function, and “Gender” is dummy function (0: girl, 1: boy). This model was selected based on Akaike’s Information Criteria. Smoothed parameters were calculated using REML method. To estimate the optimal GWG range for Chinese women by keeping a balance between maternal health and fetal growth, the least risky weight gain was calculated using the estimated model by GAMs. The GWG ranges, with respect to the predicted probability, that did not exceed a 2.5% increase from the lowest combined risk of any adverse pregnancy outcomes (the presence of at least one of them: GDM, preeclampsia, cesarean delivery, stillbirth, preterm delivery, SGA, LGA, and macrosomia) were defined as the optimal range in each subgroup of different pre-gravid BMIs or maternal age categories. There is no consensus on the increase criteria to determine the optimal GWG range, and studies used either 1% [7] or 5% [8, 9] increase criteria. Considering that our sample size was in the middle of these studies, 2.5% increases criteria was used. GDM and stillbirth were excluded when estimating the optimal GWG range as these outcomes were identified at gestational week 24–28 and gestational week 20, respectively, medical intervention might be introduced after the occurrence of these outcomes and the association between GWG and the adverse outcomes might not be accurate. Cesarean delivery was excluded since the rate was too high (58.1%). The growth curves for GWG was also constructed using a linear mixed-effects model (fixed effects: gestational weeks, variable effects: subjects). All analyses were performed using the R software, version 3.6.1 (R Development Core Team, Vienna, Austria).
Quality control
The questionnaire used in this study was reviewed by experts and was then pretested for comprehensibility, acceptability, and logic and flow. The questionnaire was revised on the basis of the results of the pretests. The investigators in each study sites were strictly trained before the formal investigations, including understanding the principles and methods of the investigation and standardizing the interview skills. Physical measurements and clinical examinations were performed by medical professionals using standard methods. All the data were entered by two persons and checked one by one to ensure the accuracy of the entered data.