Study Population and Design
The study included mothers and children who had participated in the Born in Shenyang Cohort Study, a prospective, observational cohort study of gestational factors, pregnancy outcomes, and offspring health. The study design has been described elsewhere[37]. Briefly, the BISCS was a prospective study that enrolled healthy women with single pregnancies at 21–24 weeks gestation at 54 hospitals or community healthcare centers in the urban area of Shenyang, northeast China between April and September 2017. Of a cohort of 2068 women who were invited to participate, 1338 (mean gestation of 22 ± 1.2 weeks) agreed, and 1290 of them had live singleton births. Research fellows conducted home visits with mothers and infants within 7 days of birth and at visits to child development clinics at 1, 3, 6, 8, and 12 months of age. Follow-up data were available from 72.4% of the 1290 mothers and their children at 1 month, 71.2% at 3 months, 66.7% at 6 months, 77.8% ant 8 months, and 77.4% at 12 months.
Trained research fellows conducted face-to-face interviews at enrollment and follow-up visits. Demographic, societal, environmental, behavioral, and clinical characteristics of the mothers and children were collected by structured questionnaires. Anthropometric measurements of the children were performed at each follow-up visit. Among 1290 mother–child pairs, 210 with missing oral glucose tolerance test (OGTT) results, 133 with missing anthropometric data, and two with pregestational diabetes mellitus were excluded. The remaining 945 mother–child pairs were included in the final analysis, and the results were compared with the observations in the 345 excluded participants. The characteristics of the two groups were similar.
Gestational Diabetes Mellitus
Women were given a 75 g, 2 hour OGTT at 24 ± 1.2 weeks of gestation after an overnight fast of at least 8 hours. GDM required a fasting plasma glucose ≥ 5.1 mmol/L, a 1-h plasma glucose ≥ 10.0 mmol/L, or a 2-h plasma glucose ≥ 8.5 mmol/L following the criteria of the International Association of Diabetes Pregnancy Study Group[38]. Pregnant women were stratified by glucose status to GDM and non-GDM groups.
Anthropometric Characteristics of Children
Neonatal weight and length were measured by trained research fellows at 1, 3, 6, 8, and 12 months of age with calibrated infant stadiometers (Seca 416; Seca Corporation, Hamburg, Germany) and weighing scales (Seca 376+). The mean values of two successive measurements were reported. Sex- and length-specific z-scores for weight (WFLZ) and sex- and age-specific z-scores for BMI (BMIZ), weight (WFAZ), and length (LFAZ) were calculated using the World Health Organization child growth references[39]. The data were retrieved from the children’s medical records.
Covariates
Age, race, educational attainment, income level, parity, gestational age, and pre-pregnancy weight were collected with standard questionnaire at the enrollment visit. Height was measured on enrollment with a calibrated stadiometer. Age in years and gestational age in weeks were analyzed as continuous variables. The mother’s race was recorded as either Han or other. Women were stratified by educational level to four groups (middle school or below, high school, college, graduate school or above) and by parity to either one or more than one pregnancies. Maternal pre-pregnancy and paternal BMIs were stratified to three groups, < 18.5, 18.5 to < 24.0, and ≥ 24.0 kg/m2) using Chinese reference values. Overweight and obese women were included in a single category because of the limited sample size.
Statistical Analysis
The longitudinal associations between GDM status, i.e., with or without GDM, and blood glucose concentration (mmol/L) and infant WFLZ, WFAZ, and LFAZ from birth to 12 months were determined using linear mixed effects (LME) models. LME takes into account within-subject correlation of repeated measurements and also compensates for incomplete outcome measurements. The models included an unstructured covariance matrix for random-effect variables (intercept and slope) and a maximum likelihood-estimation method. Crude and adjusted analyses were performed with two models. Model 1 was adjusted for linear, quadratic and, cubic terms for infant’s age to estimate the association between GDM and blood glucose concentration with z-scores growth estimates across infancy. Model 2 included the variables in model 1 plus pre-pregnancy BMI, maternal age, race, parity, education, household income, and paternal BMI. Interactions of GDM status and blood glucose level with pre-pregnancy weight and BMI category and their association with infant growth measurements were tested by including the corresponding interactions into the models. Stratified analyses were performed to investigate the association between GDM status and infant size among women of different pre-pregnancy weight status (underweight vs. normal weight vs. overweight/obese). The association between GDM status and blood glucose level and infant sex- and age-specific z-scores for WFL, weight and length at birth and at 1, 3, 6, 8, and 12 months of age was estimated by multivariable linear regression. The full model was adjusted for pre-pregnancy BMI, maternal age, race, parity, education, household income, and paternal BMI. Multiple imputation was performed to compensate for missing values in the linear regression models. In the sensitivity analysis, BMIZ was compared with WFLZ as outcomes of the LME analysis (Table S1). To assess the robustness of the study findings, the LME analysis was repeated in participants without missing covariate or outcomes data (n = 505, Table S2). The statistical analysis was performed with Stata/SE version 13 (StataCorp, College Station, TX, USA). Two-sided p-values < 0.05 were considered statistically significant.