|Home | About | Journals | Submit | Contact Us | Français|
Intrauterine life may be a critical period for the programming of later obesity, but there is conflicting evidence about whether pregnancy weight gain is an important determinant of offspring adiposity.
The purpose of this study was to examine the relationship of pregnancy weight gain with neonatal and childhood body composition.
The participants (n=948) were children born to women in the Southampton Women’s Survey who had dual-energy x-ray absorptiometry measurements of body composition at birth, 4 or 6 years. Pregnancy weight gain was derived from the mothers’ measured weights before pregnancy and at 34 weeks gestation, analyzed using 2009 Institute of Medicine (IOM) categories (inadequate, adequate or excessive), and as a continuous measure.
Almost half (49%) the children were born to women who gained excessive weight in pregnancy. In comparison with children born to women with adequate weight gain, they had a greater fat mass in the neonatal period (0.17 SD (95% CI 0.02, 0.32), P=0.03), at 4 years (0.17 SD (0.00, 0.34), P=0.05) and at 6 years (0.30 SD (0.11, 0.49), P=0.002). Greater pregnancy weight gain, as a continuous measure, was associated with greater neonatal fat mass (0.10 SD per 5kg weight gain (0.04, 0.15), P=0.0004) and, weakly, with fat mass at 6 years (0.07 SD per 5kg (0.00, 0.14), P=0.05), but not at 4 years (0.02 SD per 5kg (−0.04, 0.08), P=0.55).
Appropriate pregnancy weight gain, as defined by 2009 IOM recommendations, is linked to lower levels of adiposity in the offspring.
The prevalence of obesity is rising rapidly in children (1,2) and is associated both with childhood ill-health (3) and an increased risk of subsequent adult obesity (4;5), with clear consequences for future disease burden. Understanding the causes of childhood obesity will be key in defining preventative strategies in the future (3).
The intrauterine environment is thought to impact on many aspects of human health throughout the lifecourse (6), and may be a critical period for the programming of obesity. High pregnancy weight gain is directly associated with large size at birth (7). By altering the intrauterine environment, pregnancy weight gain may not only influence fetal growth, but also alter body composition in childhood and later life. Experimental studies in animals have linked both maternal under- and over-nutrition with obesity in the offspring as adults (8;9). Consistent with these observations human studies have linked both maternal famine exposure during pregnancy and maternal obesity with greater adiposity in the offspring (3;10;11).
In 1990 the US Institute of Medicine (IOM) published recommended levels of pregnancy weight gain according to pre-pregnancy Body Mass Index (BMI) categories (12). These have recently been revised (7), and consider the available evidence regarding various consequences for the short- or long-term health of the mother and child that are related to pregnancy weight gain (13). Studies that have examined the association between pregnancy weight gain and childhood adiposity have demonstrated inconsistent results. Positive linear associations have been described between pregnancy weight gain and BMI at age 3 (14), 7 (15) and 6-12 years (16). Recent studies amongst older offspring have found U-shaped relationships between pregnancy weight gain and both adult BMI (17) and adolescent BMI (18); this latter association became linear when the analyses were adjusted for maternal pre-pregnant BMI. However, two further studies (11;19) found no relationship between pregnancy weight gain and childhood adiposity.
Some of the differences between studies may be due to the use of BMI, which depends on both fat and lean tissue mass, and can provide a misleading indication of adiposity (20). BMI is known to be particularly difficult to interpret in children (21;22). In a recent small study using direct measures of body composition as assessed by Dual X-ray Absorptiometry (DXA) there was no association between pregnancy weight gain and adiposity at age 9 years (11). A second issue is that all previous studies relied upon recalled measures of pregnancy weight gain; such measures are potentially biased since they risk being influenced by a variety of sociodemographic characteristics (23).
The aim of our study was to describe the associations between pregnancy weight gain and DXA measures of neonatal and childhood body composition using data from a large longitudinal cohort in which the mother’s weight was measured before pregnancy.
The Southampton Women’s Survey (SWS) has assessed the diet, body composition, physical activity and social circumstances of a large group of non-pregnant women aged 20 to 34 years living in the city of Southampton, UK. Full details of the study have been published previously (24). Women were recruited between April 1998 and December 2002 through General Practices across the city. Each woman was sent a letter inviting her to take part in the survey, followed by a telephone call when an interview date was arranged. In total 12,583 women agreed to take part, 75% of all women contacted. Trained research nurses visited the women at home and collected information about their health, diet and lifestyles, as well as taking anthropometric measurements in triplicate. Women who subsequently became pregnant were studied at 11, 19 and 34 weeks gestation and their children were followed up in infancy and childhood.
Details of mothers’ parity and educational attainment (defined in 6 groups according to highest academic qualification) were obtained at the pre-pregnant interview, and height and weight were measured. Amongst women who became pregnant, smoking status in pregnancy was ascertained at the 11 and 34 week interviews. At 34 weeks gestation (late pregnancy) research nurses weighed the women again. The infants were visited at the ages of 6 and 12 months, and the duration of breastfeeding was defined according to the date of the last breastfeed collected at these visits.
The neonate’s crown-heel length (using a neonatometer, CMS Ltd, UK) and child’s height (using a portable stadiometer, Leicester height measurer) were measured. At birth, 4 and 6 years of age subsets of children attended for an assessment of body composition by DXA, using a Lunar DPX-L instrument (GE Corp, Madison, WI) in neonates and a Hologic Discovery instrument (Hologic Inc., Bedford, MA, USA) in childhood. Fat mass and fat-free mass were derived from a whole body scan, using paediatric software. The total X-ray dose for the whole body scans was approximately 10.5 microsieverts (paediatric scan mode), equivalent to around 1-2 days background radiation.
1,981 women became pregnant and delivered liveborn singleton infants before the end of 2003. The initial interview for 109 of these women was after the date of their last menstrual period, so they were excluded from the analyses. 5 infants died in the perinatal period and 2 had major congenital growth abnormalities, leaving 1,865 mother-offspring pairs. Since patterns of maternal weight gain amongst pre-term infants may differ from those amongst term infants (25), 114 mothers who delivered before 37 weeks are not included in the analyses. Anthropometric data were missing for 14 women before pregnancy and for 42 in late pregnancy, leaving 1,695 mother-offspring pairs for whom we could calculate weight gain in pregnancy. Of these children, 948 had whole body DXA scans at any time point and form the analysis sample in this paper: 564 at birth, 543 at 4 years and 402 at 6 years. 106 children were measured at all three ages. Of those measured at 4 years, 252 were also measured at 6 years
Pre-pregnancy data on the analysis sample were collected at a median of 1.1 years before conception. Since women’s weight tended to increase with age (by approximately 0.4% each year), a regression adjustment was used to remove the average increases in weight between data collection and conception, so that all women’s weights before pregnancy were adjusted to their predicted weight on the day of conception. Pregnancy weight gain was calculated as the difference between pre-pregnancy weight and weight at 34 weeks gestation. Women were categorized as gaining inadequate, adequate or excessive weight, according to the 2009 recommendations of the Institute of Medicine (IOM) (7) (Table 1). We used the weekly gains to adjust the recommended ranges to 34 weeks gestation. Weight gains below the lower end of this adjusted range were defined as inadequate, those within the range were defined as adequate, and those above the higher end as excessive.
All fat mass variables were positively skewed, and so were log-transformed before analyses. For ease of interpretation these values were converted to internal z-scores, as were fat-free mass variables to allow comparison between measures. Pearson’s correlation coefficients were used to examine relationships between measurements of fat and fat-free mass at different ages. Linear regression models were fitted with body composition variables as outcomes and pregnancy weight gain (both IOM weight gain categories, and treated as a continuous variable) as a predictor. All models for neonates included sex, gestation, age at measurement, age squared and length. All models at 4 and 6 years included sex, age at measurement and childhood height; by including length or height as covariates the effects of pregnancy weight gain on body composition were independent of child’s stature. Analyses were conducted unadjusted and adjusted for pre-pregnancy BMI. Additional adjustments were made for potential confounding factors that have been linked both to pregnancy weight gain and child’s body composition (maternal smoking in pregnancy, age, height, parity, educational attainment and breastfeeding duration, and birthweight) (3;7). Results for the continuous measure of pregnancy weight gain are presented per 5 kg increase. Tests for U-shaped relationships were performed by including quadratic terms in the models.
Statistical analyses were performed using Stata 11.0 (26). The Southampton Women’s Survey was approved by the Southampton and South West Hampshire Local Research Ethics Committee. Written consent was obtained from all participants.
Pregnancy weight gain of the SWS women studied is given in Table 1, their characteristics and those of their children are shown according to the IOM weight gain categories in Table 2 Excessive weight gain was common, such that almost half the women (49%) gained more weight in pregnancy than recommended by the IOM. 21% of women had inadequate weight gain in pregnancy. Compared with the 803 women who were eligible for the analyses but who did not have maternal anthropometric data or did not have DXA scans at any time points, the 948 participants in this study were better educated (P < 0.001), less likely to smoke in pregnancy (P < 0.0001) and older (P = 0.007), but there were no differences in height or pre-pregnancy BMI.
There were moderate correlations between neonatal fat mass and childhood fat mass at ages 4 (r = 0.24) and 6 years (r = 0.19) years; 4 and 6 year fat mass were strongly correlated, r = 0.86. There were stronger correlations between neonatal fat-free mass and childhood fat-free mass at ages 4 (r = 0.45) and 6 years (r = 0.44); 4 and 6 year fat-free mass were strongly correlated, r = 0.89.
We examined associations between inadequate or excessive weight gain and neonatal and childhood body composition outcomes (Table 3), using adequate weight gain as the baseline category. The unadjusted results revealed associations between excessive pregnancy weight gain and greater fat mass at 4 and 6 years, and a borderline association with neonatal fat mass, but no associations with fat-free mass. These results were not adjusted for pre-pregnancy BMI because the IOM recommendations are dependent on pre-pregnancy BMI. After adjustment for maternal smoking in pregnancy, age, height, parity, educational attainment and breastfeeding duration (for childhood outcomes), excessive pregnancy weight gain had statistically significant associations with fat mass at all three time points. Compared to women with adequate weight gain, those with excessive weight gain had offspring with 0.17 SD (95% CI 0.02, 0.32), 0.17 SD (95% CI 0.00, 0.34) and 0.30 SD (95% CI 0.11, 0.49) greater fat mass at birth, age 4 years and age 6 years respectively, which is equivalent to increases in fat mass of 7% (95% CI 1,14), 4 % (95% CI 0, 9) and 10 % (95% CI 4, 17). Birthweight was considered separately as a factor that might be on the causal pathway between pregnancy weight gain and childhood body composition. When the results at 4 and 6 years were additionally adjusted for birthweight, the effect sizes were attenuated slightly and that for 4 year fat mass became statistically non-significant (Table 3). Interaction terms were fitted in the models with and without confounders to test whether the detrimental effect of being in the excessive weight gain category on subsequent adiposity was greater amongst those of a higher (or lower) pre-pregnancy BMI, but none were statistically significant (data not shown). Amongst the 88% of participants who had paternal BMI recorded, the models were refitted including this variable, but this made little difference to the effect sizes (data not shown).
The effect of pregnancy weight gain category on fat mass at birth, 4 and 6 years is illustrated in Figure 1. Tests for U-shaped relationships of the ordered weight gain category variable (inadequate-adequate-excessive weight gain) on fat mass were of borderline significance at 4 years (P=0.08), and statistically significant when adjusted for confounders at 6 years (P = 0.01).
We also examined linear associations between pregnancy weight gain and neonatal and childhood body composition (Table 4). Before adjustment for confounders, only neonatal fat mass had a borderline statistically significant association with pregnancy weight gain, and there were no associations for fat-free mass. Adjustment for pre-pregnancy BMI strengthened the associations between pregnancy weight gain and both neonatal and 6 year fat mass, although the latter was borderline statistically non-significant. Adjustment for maternal smoking in pregnancy, age, height, parity, educational attainment and breastfeeding duration did not markedly change the estimates of effect size. Unlike two previous studies (17;18), we found no evidence of a U-shaped association between pregnancy weight gain analyzed as a continuous variable and any of the neonatal or childhood outcome measures, either before or after adjusting for confounders (all P > 0.05).
Pregnancy weight gain was positively associated with birthweight such that for every 5 kg increase in pregnancy weight gain there was a 76g (95% CI 54, 98) increase in birthweight (P < 0.0001), adjusted for confounders. Birthweight was also positively associated with both fat and fat-free mass in childhood (Table 5). Therefore when birthweight was included in the models comparing pregnancy weight gain and childhood body composition at 4 and 6 years (Tables (Tables33 and and4),4), the estimates were slightly attenuated.
Pregnancy weight gain was greater amongst primiparous women (mean (SD) 13.0 (5.9) kg) than multiparous women (mean (SD) 11.3 (6.1) kg), whereas neonatal fat mass was greater amongst multiparous women (median (IQR) 0.57 (0.42 – 0.73) kg) than primiparous women (median (IQR) 0.50 (0.37 – 0.63) kg) (both P<0.001). However, mother’s parity was not associated with the offspring’s fat mass at either 4 or 6 years. Inadequate pregnancy weight gain was particularly associated with lower neonatal fat mass in primiparous mothers (−0.21 vs −0.04 SDs in infants of multiparous mothers), but multivariate analyses including or excluding the mother’s parity showed little difference in the magnitude of the associations between maternal weight gain category and offspring adiposity. We also repeated our analyses amongst the subgroup of women in the Southampton Women’s Survey who conceived within one year of their initial interview and found similar results (data not shown).
We have demonstrated an association between excessive pregnancy weight gain and fat mass measured by DXA in childhood. Almost half the women studied gained excessive weight according to the 2009 IOM guidelines. Their children had a greater fat mass at birth, 4 and 6 years (increases of 7%, 4% and 10% respectively), when compared to children whose mothers’ weight gain in pregnancy was adequate. Greater pregnancy weight gain, analyzed as a continuous variable, showed weaker associations with childhood adiposity. After adjustment for confounders, weight gain was not associated with fat mass at 4 years, although there were associations between pregnancy weight gain and fat mass at birth, and a weak association at 6 years. Using the IOM categories, we found some evidence that inadequate pregnancy weight gain may also be unfavourable in relation to childhood adiposity (Figure 1).
Previous studies using BMI as the outcome measure have found positive linear associations with pregnancy weight gain and BMI in childhood (14-16). Our findings using DXA-assessed measures of fat mass at 6 years are consistent with these studies, although we did not observe this relationship at 4 years. There was no evidence of a U-shaped relationship between continuous pregnancy weight gain and adiposity in our study, as has previously been described in relation to BMI in adolescence (18) and adulthood (17). Three previous studies have compared the 1990 IOM pregnancy weight gain categories with childhood body composition, two using BMI (14;15) and one using DXA-assessed fat mass (11) as outcomes. These studies found that BMI or fat mass index increased in a linear manner across inadequate, adequate and excessive IOM categories. By contrast, using the 2009 IOM weight gain recommendations, we observed that both inadequate and excessive weight gain were associated with greater childhood fat mass at 6 years, although this was not seen at younger ages. Differing findings may arise from our use of a direct measure of adiposity as assessed by DXA, rather than using BMI as a surrogate measure, or the differing ages of the subjects studied. Our analyses also considered the association between birthweight and childhood body composition. In agreement with previous studies (22), we found a strong association between birthweight and childhood lean mass after adjustment for confounders, and a positive but weaker association between birthweight and childhood fat mass (Table 5).
A particular strength of the SWS is that we interviewed a large sample of young women both before and during pregnancy, thus providing a valuable opportunity to assess pregnancy weight gain using measured pre-pregnancy weight. Data are from a contemporary cohort of women from a wide range of sociodemographic backgrounds with a good response rate. We were able to use the 2009 IOM weight gain recommendations in our analyses, which are based on the most recent evidence available. We used DXA to provide direct measures of fat mass and fat-free mass. We had data on length at birth and height at 4 and 6 years; by including these variables in all models the effects found were independent of length or height. We were also able to control for a large number of other potential confounding factors, including important predictors of childhood body composition such as pre-pregnancy BMI, smoking in pregnancy and duration of breastfeeding.
A weakness of our analyses is that the time from pre-pregnancy assessment to conception is variable. However, we were able to adjust for this, and analyses amongst the subgroup of women who conceived within one year of their pre-pregnancy weight measurement gave very similar results. Weight gain in our study was calculated from before pregnancy to 34 weeks gestation. Whilst in calculating the IOM categories it was possible to account for this by using the recommended gain per week in the 2nd and 3rd trimester, the continuous measure of pregnancy weight gain was limited to the 34-week time point. DXA measurements were available on a subset of 948 children, 106 of whom had measurements at all 3 ages. When the subset were compared with children with no DXA measurements, they had mothers who were better educated, less likely to smoke in pregnancy and older. However, unless the associations between pregnancy weight gain and offspring body composition are different in the remainder of the cohort, it is unlikely that selection bias could explain the relationships observed.
At 6 years of age, we found evidence of a U-shaped relationship, such that adiposity was greater both among children born to mothers whose pregnancy weight gain was inadequate, as well as those with excessive gains. Although this accords with the findings of 2 recent studies, (17,18), in our study we did not find this association when we used weight gain as a continuous measure. Experimental studies suggest that different maternal exposures can lead to offspring adiposity through different mechanistic pathways (27). In relation to our findings we speculate that inadequate maternal weight may lead to adiposity in the offspring primarily through a central effect on the regulation of appetite (28), while excessive maternal weight gain may result in a variety of metabolic changes including greater adipocyte number in fetal and postnatal life, together with altered hepatic metabolism (29;30). Further data from our continued follow-up of this cohort of children will enable us to address the importance of inadequate and excessive pregnancy weight gain in the aetiology of childhood adiposity.
It is not possible in this observational study to determine whether associations are causal. Whilst we adjusted for several potentially important confounders, it may be that there are other genetic or familial behavioral factors that are associated with both weight gain in pregnancy and childhood adiposity that we did not account for in our analyses. In particular, variations in diet and levels of physical activity may be important. For example, children of obese parents have been shown to be less active and to have a greater preference for fatty foods (31,32). In our analyses, we cannot exclude the possibility that excessive pregnancy weight gain is linked to a more obesogenic diet and lower levels of physical activity in childhood. However, an alternative explanation is that the impact of weight gain during pregnancy directly affects the intrauterine environment in a manner that programs later obesity in the fetus; the mechanisms for such developmental origins of health are an area of current research (33). The magnitude of the associations between excessive weight gain and childhood adiposity decreased slightly after birthweight was added to the regression models, suggesting that it may be on the causal pathway, though it does not explain a large proportion of the associations.
As in previous studies (11;14;15), we found that excessive weight gain in pregnancy was common, and that it was most marked amongst women who were overweight or obese before pregnancy (Table 1). This group may therefore represent those who would most benefit from counseling from health professionals to limit gestational weight gain. There is, however, limited evidence from randomized controlled trials on the most appropriate dietary and/or physical activity interventions to employ (34). The Institute of Medicine (IOM) recommended levels of pregnancy weight gain are used in clinical practice in the US, but women in the UK are mostly unaware of these guidelines. Promoting measures to ensure appropriate weight gain in pregnancy as defined by the IOM may have a positive impact on childhood adiposity, although long-term direct implications for adult obesity and health are currently unknown.
We are grateful to the women of Southampton who took part in these studies and the research nurses and other staff who collected and processed the data. The SWS was funded by the Medical Research Council, the University of Southampton and the Dunhill Medical Trust.
The authors’ responsibilities were as follows: SRC conducted the statistical analyses and wrote the first draft of the paper. SMR, HMI and KMG gave substantial advice on the analyses, whilst NCH, ZAC and CC were involved with collection and intrepretation of the DXA data. All authors contributed to the interpretation of the analyses and the writing of the manuscript.
None of the authors had any conflict of interest.