shows the characteristics of mothers and offspring. shows the association of IOM categories with adiposity and cardiovascular risk factors. Offspring of women who gained more than IOM recommended GWG were more likely to have greater BMI, WC, fat mass, leptin, SBP, CRP and IL-6 levels. They were also more likely to have lower HDLc and Apolipoprotein A1 levels. Children of women who gained less than recommended amounts had lower levels of adiposity, but other cardiovascular risk factors tended to be similar in this group to offspring of women gaining recommended amounts. IOM categories were not associated with DBP, non-HDLc, apolipoprotein B or triglyceride levels. Associations remained with adjustment for confounders. IOM categories were associated with binary outcomes of offspring overweight/obesity. In confounder adjusted models offspring of women who gained less than recommended, compared to those gaining recommended, levels had odds ratios of overweight/obesity (based on BMI) of 0.80 (0.67, 0.96) and of central obesity (based on waist) of 0.79 (0.69, 0.90) and offspring of mothers who gained more than recommended, compared to those gaining recommended, had odd ratios of overweight/obesity and central obesity of 1.73 (1.45, 2.05) and 1.36 (1.19, 1.57), respectively.
Mean difference (95%CI) in offspring adiposity, blood pressure, lipids, apolipoproteins and inflammatory markers by IOM categories of maternal gestational weight gain (N=5154 or 3457 as indicated)
When we used multilevel models including repeat measures of gestational weight to estimate GWG in more detail, three distinct periods of GWG were identified – early-0-14 weeks; mid- >14-36 & late-pregnancy > 36 weeks (). In early pregnancy 20.0% of women either lost weight or remained stable. The majority of women in both mid- (99.9%) and late-pregnancy (95.7%) gained weight. Web-table3
shows the correlations between estimated pre-pregnancy weight, estimated GWG in early-, mid- and late-pregnancy, total absolute GWG over the whole pregnancy and birthweight. Most correlations were modest or weak. There was a strong inverse association of estimated GWG in early and late pregnancy, and a strong positive association of estimated GWG in mid and late pregnancy.
Weight (kg) by gestational age (weeks) for mothers of boys (dashed line) and girls (solid line)
shows the associations of estimated pre-pregnancy weight (per 1kg change) and estimated GWG (per 400 kg/wk) with offspring adiposity (BMI, WC, fat mass, leptin) and BP. Estimated pre-pregnancy weight was positively linearly associated with all four measurements of offspring adiposity and SBP and DBP, with these associations remaining after adjustment for confounders.
Mean difference (95%CI) in offspring measurements of adiposity and blood pressure per 1kg change in maternal estimated pre-pregnancy weight and 400g/week estimated gestational weight gain (N=5154)
For associations of estimated GWG with adiposity and BP there was evidence of non-linearity with knots (changes in the direction and/or magnitude of association) at 0 and 500 g/week for GWG in early pregnancy and at 250 and 500 g/week in both mid- and late-pregnancy. Estimated GWG in all three periods generally had ‘U’ shaped associations with offspring adiposity, with null or inverse associations in women gaining low levels of weight, then null associations in the middle range of estimated GWG and then positive associations (model1, ). However, with adjustment for confounding factors (model2) the inverse associations at low levels of estimated GWG attenuated. In the confounder adjusted model, women who lost weight or did not gain weight in early-pregnancy (i.e. low estimated GWG women) had no association between their average gestational weight change per week and offspring adiposity. However, for those women (i.e. medium or high estimated GWG women) gaining weight during this period there was a positive association of estimated GWG with measures of offspring adiposity, which strengthened in women gaining on average 500g/week or more.
For mid-pregnancy estimated GWG up to 500g/week (i.e. low or medium estimated GWG) was not associated with offspring adiposity but offspring adiposity increased linearly with estimated GWG in mid-pregnancy after this level (i.e. in women with high GWG). There was no clear association of estimated GWG in late-pregnancy (beyond 36 weeks) with offspring adiposity or of estimated GWG in any periods with SBP or DBP. Associations of pre-pregnancy weight and estimated GWG with binary outcomes of adiposity (Web-Table4
) were consistent with those seen for the continuously measured variables shown in .
shows the associations of estimated pre-pregnancy weight and estimated GWG with lipids, apolipoproteins and inflammatory markers. For these outcomes there was no strong evidence of non-linear associations. Estimated pre-pregnancy weight and GWG in mid-pregnancy were positively associated with triglyceride levels and IL-6 and inversely associated with HDLc and ApoA1, though for triglyceride and ApoA1 confidence intervals were wide and included the null value. Estimated pre-pregnancy weight was also positively associated with non-HDLc, apolipoprotein B and CRP, but not with adiponectin. GWG in early- and late-pregnancy were not associated with lipids, apolipoproteins or inflammatory markers, with point estimates all close to the null value.
Mean difference (95%CI) in offspring lipids, apo-lipoproteins and inflammatory markers per 1kg change in maternal estimated pre-pregnancy weight and 400g/week estimated gestational weight gain for blood assay results (N=3457)
Further adjustment for birthweight did not substantively alter any of the confounder adjusted models (Web-Tables5a-5c
). All associations of maternal exposures that were present in confounder adjusted models were attenuated to the null with further adjustment for offspring fat mass (Web-Tables6a-6b
). When these additional analyses were repeated with offspring BMI, WC or leptin instead of fat mass results were very similar to those presented.
We found no evidence that associations of estimated GWG with any of our outcomes were modified by pre-pregnancy BMI or weight, irrespective of whether this was estimated or observed, (all p-values for interaction > 0.2). When the analyses with estimated GWG were repeated with only those women who had at least 2, 4 and 3 measures in each time period respectively (i.e. total of at least 9 per woman across pregnancy) there was no substantial change to the results. Associations with estimated GWG in late-pregnancy did not differ substantively from those presented when we used absolute weight gain. Associations did not differ substantively with the removal of women whose first antenatal measurement was after 15 weeks or whose last measurement was before 35 weeks.