Ethnic Differences in Effects of Maternal Pre-Pregnancy and Pregnancy Adiposity on Offspring Size and Adiposity

Abstract

Context: Maternal adiposity and overnutrition, both before and during pregnancy, plays a key role in the subsequent development of obesity and metabolic outcomes in offspring.

Objective: We explored the hypothesis that maternal adiposity (pre-pregnancy and at 26-28 weeks' gestation) and mid-pregnancy gestational weight gain (GWG) are independently associated with offspring size and adiposity in early childhood, and determined whether these effects are ethnicity dependent.

Design: In a prospective mother-offspring cohort study (N = 976, 56% Chinese, 26% Malay, and 18% Indian), we assessed the associations of offspring size (weight, length) and adiposity (subscapular and triceps skinfolds), measured at birth and age 6, 12, 18, and 24 mo, with maternal pre-pregnancy body mass index (ppBMI), mid-pregnancy GWG, and mid-pregnancy four-site skinfold thicknesses (triceps, biceps, subscapular, suprailiac).

Results: ppBMI and mid-pregnancy GWG were independently associated with postnatal weight up to 2 y and skinfold thickness at birth. Weight and subscapular and triceps skinfolds at birth increased by 2.56% (95% confidence interval, 1.68-3.45%), 3.85% (2.16-5.57%), and 2.14% (0.54-3.75%), respectively for every SD increase in ppBMI. Similarly, a one-SD increase in GWG increased weight and subscapular and triceps skinfolds at birth by 2.44% (1.66-3.23%), 3.28% (1.75-4.84%), and 3.23% (1.65-4.84%), respectively. ppBMI and mid-pregnancy suprailiac skinfold independently predicted postnatal skinfold adiposity up to 2 years of age, whereas only GWG predicted postnatal length. The associations of GWG with postnatal weight and length were present only among Chinese and Indians, but not Malays (P < .05 for interaction).

Conclusions: ppBMI and GWG are independent modifiable factors for child size and adiposity up to 2 years of age. The associations are ethnic-dependent, and underscore the importance of ethnic specific studies before generalizing the applicability of risk factors reported in other populations.

Trial registration: ClinicalTrials.gov NCT01174875.

Figures

Figure 1.

A–D, Associations of maternal prepregnancy BMI and gestational weight gain with child weight (A, B) and length (C, D) from birth to 24 mo. Point estimates and 95% CIs for associations of log-transformed child anthropometric outcome (child weight, length) with maternal prepregnancy BMI and gestational weight gain. Regression coefficients and CIs are reported as percentage change in child anthropometric outcome for one SD increase in maternal prepregnancy BMI or gestational weight gain. Each of the two rows gives results from two different linear regression models where both maternal prepregnancy BMI and gestational weight gain were included as predictors. Both models adjusted for temporal effects, child sex, ethnicity, gestational age, parity, maternal height, education, and age. Time was coded using a binary variable for each distinct time point and interaction terms of time with each variable in the model were included. Results are also reported in Supplemental Tables 1 and 2.

Figure 2.

A–D, Associations of maternal prepregnancy BMI and gestational weight gain with child subscapular (A, B) and triceps (C, D) skinfolds from birth to 24 months. Point estimates and 95% CIs for associations of log-transformed child anthropometric outcome (child subscapular and triceps skinfolds) with maternal prepregnancy BMI and gestational weight gain. Regression coefficients and confidence intervals are reported as percentage change in child anthropometric outcome for one SD increase in maternal prepregnancy BMI or gestational weight gain. Each of the two rows gives results from two different linear regression models where both maternal prepregnancy BMI and gestational weight gain were included as predictors. Both models adjusted for temporal effects, child sex, ethnicity, gestational age, parity, maternal height, education, and age. Time was coded using a binary variable for each distinct time point and interaction terms of time with each variable in the model were included. Results are also reported in Supplemental Tables 1 and 2.

Figure 3.

A–F, Ethnicity-stratified associations of gestational weight gain with child weight (A–C) and length (D–F) from birth to 24 months. Gestational weight gain was associated with child weight and length after birth only among Chinese and Indian children. Point estimates and 95% CIs for each ethnic group, for associations of log-transformed child anthropometric outcome (child weight, length) with gestational weight gain. Regression coefficients and CIs are reported as percentage change in child anthropometric outcome for one SD increase in gestational weight gain. Each of the six panels gives results from six different linear regression models, for each of the three ethnic groups and two child outcomes, and gestational weight gain as predictor variable. All models adjusted for temporal effects, child sex, gestational age, parity, maternal height, education, age, and ppBMI. Time was coded using a binary variable for each distinct time point and interaction terms of time with each variable in the model were included. Results are also reported in Supplemental Tables 8 and 9. Results without stratifying by ethnicity are given in Figures 1, B and D.

Figure 4.

A–H, Ethnicity-stratified associations of ppBMI (A–D) and gestational weight gain (E–H) with maternal skinfolds. Point estimates and 95% CIs for each ethnic group, for associations of maternal skinfolds (triceps, biceps, subscapular, and suprailiac), with ppBMI and gestational weight gain. Regression coefficients and CIs are reported as change in maternal skinfold (mm) for one SD increase in prepregnancy BMI or gestational weight gain. Each column gives results from three different linear regression models for each of the three ethnic groups in which both maternal prepregnancy BMI and gestational weight gain are included as predictors. All models adjusted for parity, maternal height, education and age.

Figure 5.

A–D, Associations of maternal suprailiac skinfold with child weight (A), length (B), subscapular (C), and triceps (D) skinfolds from birth to 24 months. Point estimates and 95% CIs for associations of log-transformed child anthropometric outcome with maternal suprailiac skinfold. Regression coefficients and CIs are reported as percentage change in child anthropometric outcome for one SD increase in maternal suprailiac skinfold. Each of the four panels gives results from four different linear regression models with maternal suprailiac skinfold as predictor. All four models adjusted for temporal effects, child sex, ethnicity, gestational age, parity, maternal height, education, age, ppBMI, and gestational weight gain. Time was coded using a binary variable for each distinct time point and interaction terms of time with each variable in the model were included. Results are also reported in Supplemental Tables 20, 22, 24, and 26.