Impact of Maternal Glucose and Gestational Weight Gain on Child Obesity over the First Decade of Life in Normal Birth Weight Infants
- First Online:
- Cite this article as:
- Hillier, T.A., Pedula, K.L., Vesco, K.K. et al. Matern Child Health J (2016) 20: 1559. doi:10.1007/s10995-016-1955-7
- 881 Downloads
Objective To determine, among children with normal birth weight, if maternal hyperglycemia and weight gain independently increase childhood obesity risk in a very large diverse population. Methods Study population was 24,141 individuals (mothers and their normal birth weight offspring, born 1995–2003) among a diverse population with universal GDM screening [50-g glucose-challenge test (GCT); 3 h. 100 g oral glucose tolerance test (OGTT) if GCT+]. Among the 13,037 full-term offspring with normal birth weight (2500–4000 g), annual measured height/weight was ascertained between ages 2 and 10 years to calculate gender-specific BMI-for-age percentiles using USA norms (1960–1995 standard). Results Among children who began life with normal birth weight, we found a significant trend for developing both childhood overweight (>85 %ile) and obesity (>95 %ile) during the first decade of life with both maternal hyperglycemia (normal GCT, GCT+ but no GDM, GDM) and excessive gestational weight gain [>40 pounds (18.1 kg)]; p < 0.0001 for both trends. These maternal glucose and/or weight gain effects to imprint for childhood obesity in the first decade remained after adjustment for potential confounders including maternal age, parity, as well as pre-pregnancy BMI. The attributable risk (%) for childhood obesity was 28.5 % (95 % CI 15.9–41.1) for GDM and 16.4 % (95 % CI 9.4–23.2) for excessive gestational weight gain. Conclusions for Practice Both maternal hyperglycemia and excessive weight gain have independent effects to increase childhood obesity risk. Future research should focus on prevention efforts during pregnancy as a potential window of opportunity to reduce childhood obesity.
KeywordsChildhood obesity Metabolic imprinting Gestational weight gain Hyperglycemia GDM Normal birth weight
What is already known on this subject? Gestational diabetes (GDM) and excessive weight gain (GWG) increase macrosomia risk; evidence suggests they also metabolically imprint offspring for increased childhood obesity risk. How GDM and GWG impact childhood obesity risk in normal weight babies is unknown.
What this study adds? Among a diverse population of 24,141 individuals (mothers with universal GDM screening and their normal birth weight offspring, born 1995–2003) we found both GDM and GWG independently increased the risk of childhood overweight and obesity in the first decade of life.
Are all normal birth weight babies really “normal” metabolically, with equal risk for childhood obesity? Both maternal hyperglycemia and excessive maternal weight gain increase the risk of macrosomic (high birth weight or metabolically “fat”) babies [12, 32], and macrosomia is a risk factor for childhood, and eventually adult, obesity . Paradoxically, low birth weight also increases risk for later obesity and diabetes [21, 34]. In clinical practice, the tacit assumption is that normal birth weight babies are metabolically normal, given the demonstrated “U-shaped relationship” for both low and high birth weight to increase future obesity, partly because the majority of infants with normal birth weight are typically the referent group for diabetes risk [3, 14, 15]. Earlier work demonstrated that maternal hyperglycemia is a risk factor for childhood obesity , even among normal birth weight babies. They also found that excessive pregnancy weight gain [>40 pounds (18.1 kg)] nearly doubles the risk of fetal macrosomia with each increasing level of maternal glucose, even among women with GDM .
Metabolic imprinting is a general term signifying adaptive nutritional conditions early in life that permanently affect later life disease . Based on prior work [11, 12], we hypothesized that maternal hyperglycemia and/or excessive weight gain might also imprint “normal birth weight” babies for childhood obesity. Although other risk factors after birth clearly also contribute to childhood obesity, if the intrauterine environment is metabolically important even in normal birth weight babies, this would be compelling to consider maternal interventions in order to give all babies the best metabolic chance at birth for normal weight in childhood.
This study’s objective was to determine whether the intrauterine environment also imprints childhood obesity risk among the majority of infants with normal birth weight over the first decade of life, which could have important clinical practice and public health implications for prevention efforts during pregnancy. We tested our hypotheses in a very large multi-ethnic US population of pregnant women universally screened for GDM and their children who had childhood height and weight measured up to 10 years after birth.
The study population was drawn from a combined membership of over 650,000 at two health care systems, one in the Northwest (NW) and the other in Hawaii (HI). Both memberships are ~20 % of the areas’ general populations and reflect their demographic/sociographic characteristics [11, 12], including members of lower socioeconomic status (SES) enrolled through Medicaid. In Hawaii, low-income individuals enroll under the state’s insurance plan for Medicaid, about 10 % of the state and site’s population. During the study period, the NW site served ~8 % of Medicaid members through a state insurance plan—a population demographically similar to the area population [11, 12]. All members at both sites have access to medically necessary services from their systems or by referral from their primary care physician. Because both sites are integrated health maintenance organizations (HMO) and because members are predominantly enrolled under employer group plans, the population is highly stable over time. Members receive virtually all care and services in the HMOs. Information on these services is available in the electronic medical record (EMR), entered directly by providers at points of service and continuously drawn from both systems’ administrative and clinical electronic databases: inpatient admissions and deliveries, outpatient visits, laboratory tests, pharmacy dispenses, chronic-disease registries, outside claims/referrals, and others. All databases are linked through each member’s unique health record number. Validated diabetes registries  and provider diagnoses were used to exclude women with pre-existing diabetes from analyses. This research was conducted according to the 1964 Declaration of Helsinki’s ethical standards and laid down in the 1964 Declaration of Helsinki and its later amendments. The Institutional Review Boards of both study sites and the State of Hawaii Department of Health approved this study including relevant waivers.
To maximize our ability to include children with heights/weights through the average age of “adiposity rebound” risk, age 5–7 [11, 25, 37, 38], we required continuous membership from birth to at least age 7 years. We also required at least one childhood height and weight measured after age 2 years. We evaluated childhood growth from age 2 for several reasons: most children have established individual growth curve percentiles that best correlate with adult height (i.e., after infant catch-up or catch-down growth) , and childhood growth curves (using standing height) begin at age 2 . We evaluated all available measured height and weights from age 2 up to age 10 years in 13,037 children with normal birth weight (see Fig. 1).
Classification of Childhood Overweight and Obesity
Overweight (>85th percentile) and obesity (>95th percentile) were classified as age-gender-specific percentiles for BMI (kg/m2), based on CDC criteria (with the normative reference range 1960–1995, when US children were typically more lean) .
Glucose Testing and GDM Diagnosis
The screening for all pregnant women at both sites initially uses a 50 g, 1-h GCT. During the study period, our protocol was that women with a positive (+) GCT [1 h glucose >=140 mg/dl (7.8 mmol/l)] but no GDM then received the 100 g, 3-h OGTT; GDM treatment was based on National Diabetes Data Group (NDDG) criteria. For women whose providers ordered subsequent screens (n = 455), we analyzed the latest test. Among normal birth weight babies, GDM [either by NDDG or Carpenter Coustan (C&C) criteria] had a similar risk for increasing childhood obesity and no significant interaction based on GDM diagnosis treatment. Therefore, we defined GDM using C&C criteria [2 or more of 4 possible glucose concentrations measured with the 100 g OGTT are positive: Fasting ≥ 95 mg/dl (5.3 mmol/l); 1 h ≥ 180 mg/dl (10 mmol/l); 2 h ≥ 155 mg/dl (8.6 mmol/l); 3 h ≥ 140 mg/dl (7.8 mmol/l)] .
We stratified maternal glucose screening results into three categories: (1) Normal GCT (referent group); (2) +GCT, No GDM (zero or one abnormality on OGTT; two or more abnormalities are required to diagnose GDM); (3) GDM by C&C criteria .
Maternal Weight Gain
Maternal weight gain was obtained from measured weights in the EMR for most women, and when EMR data were not available, from state birth-certificates. Among the 8430 with both measured EMR and birth-certificate weight gain, measured weight gain was only 0.8 kgs greater (standard deviation 7.6) than self-reported gain on the birth certificate.
The Institute of Medicine of the National Academies (IOM)  recommends pregnancy weight gain ranging from 15 to 40 pounds (6.8–18.1 kgs), depending on pre-pregnancy BMI. We defined excessive weight gain as >40 pounds (>18.1 kg)—an excessive amount regardless of pre-pregnancy BMI by IOM guidelines —as measured pre-pregnancy weight was not available in the EMR for all women.
Classification of Ethnicity and Other Covariates
State birth certificate records were used for birth weight and validated against the inpatient EMR. Ethnicity classification was based on the mother’s reported race on the states’ official birth certificates. As per Hawaii state algorithms for classifying race, if the mother reported being any part Native Hawaiian, ethnicity was classified as Native Hawaiian. If she did not list this, but a non-Caucasian race, then we classified the mother into that non-Caucasian group. Race was classified as Caucasian only if no other race/ethnicity was reported. State birth certificates also provided mother’s reported parity. Maternal age, baby gender, and measured pre-pregnancy BMI were obtained from EMRs. Pre-pregnancy maternal BMI (kg/m2) was available in a subsample of 5926 women from both sites with measured adult height and pre-pregnancy weight available electronically (the HI outpatient EMR was just beginning in 2003 vs. mostly implemented at NW in 1995). Measured pre-pregnancy weight was defined as the last measurement prior to estimated date of conception (up to 6 months); if unavailable, we used the earliest measurement during pregnancy (up to 12 weeks gestation).
Breastfeeding status is not easily captured in the EMR. However, we surveyed a subsample of women in the Hawaii region about breastfeeding; 302 of these women had children that met inclusion criteria for this study and answered the breastfeeding question “Was this baby breast-fed? (yes/no)”.
We conducted all statistical analyses using SAS Statistical Analysis System® version 9.2 (SAS Institute, Cary, NC).
We used t-tests, Pearson Chi square test, and Mantel–Haenszel Chi square trend tests to evaluate univariate relationships of subject characteristics with childhood obesity. Individual covariates and pairwise interactions were evaluated and included in multivariate models as appropriate. No interaction terms were statistically significant. In addition to maternal glucose and weight gain, final multivariate models included these covariates: maternal age, parity, maternal non-white race, baby gender, and child- birth year. For final modeling, we conducted stratified analyses for all ethnic groups with >50 persons. Though there was insufficient power for formal testing, results were comparable for all non-white groups. Therefore we combined these groups in the final model. Overweight and obesity were evaluated both as repeated measures analysis (measurements at multiple age-periods per child) and as a time-to-obesity analysis (first measurement of overweight or obesity). Proportional hazards models were used to estimate hazards ratios for overweight and obesity, adjusting for covariates. Confidence intervals for the hazards ratios were estimated using the robust sandwich covariance estimate which adjusts for clustering due to multiple pregnancies per woman during the study period . Attributable risk percent (AR %) was used to estimate the percent of obesity and overweight cases among children born to women with the designated risk factor (e.g., excessive weight gain) in the population that are due to the risk factor . All the statistical tests reported are two-sided; the term statistically significant implies a p value < 0.05.
Characteristics of Study Sample with Normal Weight Births 1995–2003 (N = 24,141 mothers and babies; 13,037 normal weight births)a
Maternal age at screen in years, mean (SD)
Maternal weight gain in Kilograms,b n(%)
>18.1 (>40 pounds)
Maternal weight gain in Kilograms,a mean (SD)
Maternal ethnicity,b n(%)
Other Pacific Islander
Baby Birth Weight in Grams, mean (SD)
Baby Gender, n(%)
Maternal glucose,b, c n(%)
+GCT, no GDM
Maternal glucose challenge test in mmol/L [mg/dl], mean (SD)
6.4 (1.6) [115.5 (28.4)]
Overall, the prevalence of GDM was 5.5 %, and approximately 20 % of women had excessive maternal weight gain >40 pounds (Table 1). Among women with GDM, 10.9 % required insulin treatment. Among the 13,037 children, the mean number of height and weight measurements per child was 9.5 (median 5). The majority of children also had 4 or more different age-periods represented from 2 to 10 years.
Prevalence of Overweight and Obesity in Early Childhood
Risk of childhood overweight and obesity by maternal glucose & weight gain
Childhood BMI > 85 %ile
Childhood BMI > 95 %ile
AR % (95 %CI)b
AR % (95 %CI)b
GCT+, no GDM
Maternal weight gain
<=40 pounds (18.1 kg)
Excessive maternal weight gain was also associated with a higher prevalence of offspring becoming overweight (>85 %ile) or obese (>95 %ile) across the first decade of life (Fig. 2b, p < 0.0001 compared to normal weight gain). After multivariate adjustment, excessive gestational weight gain (>40 pounds) was associated with greater than 15 % increased risk of normal birth weight offspring developing childhood overweight and obesity (HR 1.17 and 1.19; Table 2). The attributable risk percent (AR %) for childhood obesity (BMI > 95 %ile) was 28.5 % (95 % CI 15.9–41.1) for GDM and 16.4 % (95 % CI 9.4–23.2) for excessive maternal weight gain (Table 2). In other words, among women with excessive weight gain, we estimate an average 16.4 % reduction in childhood obesity risk if these women had actually gained ≤40 pounds during pregnancy (assuming all other risk factors remained unchanged). To translate AR % in the population perspective, about 20 % of women gained excessive weight (~2600). If these 2600 women were instead to achieve a weight gain <=40 pounds in pregnancy, this would prevent childhood obesity in ~426 children.
Maternal Pre-pregnancy BMI
As pre-pregnancy maternal obesity is a known risk factor for childhood obesity [4, 9], we did analyses including pre-pregnancy measured BMI in the subsample of 5926 women where these data were available. Consistent with recent US trends , half (51 %) of these 5926 women were overweight or obese. We confirmed that pre-pregnancy BMI was also an independent risk factor for childhood obesity in our sample. Importantly, although the multivariate HRs were slightly attenuated after adjustment for pre-pregnancy maternal BMI, the risk of childhood obesity (BMI > 95 %ile) remained significantly associated with both maternal hyperglycemia (GDM: HR 1.31; 95 % CI 1.06–1.63) and excessive gestational weight gain (HR 1.15; 95 % CI 1.01–1.37). Multivariate adjusted risk of childhood overweight (>85 %ile) also remained significant for maternal GDM and weight gain, after additional adjustment for pre-pregnancy BMI (not shown).
Breastfeeding has many established benefits to both mother and baby, including as another potentially modifiable way to reduce childhood obesity risk [7, 26]. Among the small subsample of 302 women who answered questions about breastfeeding, the risk of their children developing obesity was dramatically reduced among mothers that breastfed versus those who did not. Specifically, after multivariate adjustment that included maternal glucose and weight gain, children of women who breastfed had an obesity prevalence of 37 %, compared to 51 % for children of women who did not breastfeed (p = 0.006). A reduced rate of childhood obesity among mothers that breastfed was also evident for both women with and without a history of GDM in this subsample (data not shown).
Among our population of 24,141 mothers and their offspring who began life with normal birth weight, both maternal hyperglycemia and excessive maternal weight gain were independent risk factors for childhood overweight and/or obesity during the first decade of life. The attributable risk percent for childhood obesity with these risk factors was as high as 28 % (GDM). Thus, our results suggest both maternal glucose and excessive weight gain are important risk factors that are potentially modifiable during pregnancy to prevent metabolic imprinting of offspring for later obesity. Metabolic imprinting is a general term signifying specific nutritional conditions early in life that permanently affect later life disease and are characterized by “(1) a susceptibility limited to a critical ontogenic window early in development, (2) a persistent effect lasting through adulthood, (3) a specific and measurable outcome that may differ quantitatively among individuals, and (4) a dose–response or threshold relation between a specific exposure and outcome” .
Multiple potential mechanisms for metabolic imprinting are emerging, including neuroendocrine effects on the hypothalamic melanocortin system controlling energy homeostasis . Genetic inheritance for excessive maternal weight gain is an unlikely explanation for our results given a recent large within-family comparison study (same mothers; 2 or more of their children) that found high maternal weight gain increased birth weight independent of genetics . Our results can be characterized by all four criteria above that define metabolic imprinting with the obesity effect demonstrated through early childhood (and obese children are more likely to remain obese adults). This suggests that metabolic imprinting from maternal hyperglycemia and/or excessive weight gain both independently increase childhood obesity risk. Even if part of the independent maternal effects for childhood obesity we found could be explained simply by behavioral maternal dietary changes that persisted for the family through childhood, the clinical implications are the same. That is, our results suggest pregnancy may be a critical window of prevention opportunity for childhood obesity by maternal dietary interventions that can both improve maternal glucose and optimize weight gain.
Animal, including primate, studies have also demonstrated that both maternal hyperglycemia and maternal obesity program the next generation for obesity and diabetes [1, 8]. Combined with our results, this suggests great potential for prevention strategies in pregnancy as a means to prevent childhood obesity. Many [11, 28], but not all , human studies have also found maternal hyperglycemia to be an independent risk factor for childhood obesity. We previously found in our large diverse HMO population that the children of women treated for GDM were less likely to become obese at age 5–7 . In contrast, in a subsample of 199 children from the ACHOIS randomized trial, treatment of mild GDM did not significantly reduce the risk of their children becoming obese at age 4-5 years old . A subsample of 500 children from the MFUMN randomized trial of treating even milder GDM (maternal normal fasting glucose < 95 mg/dl [5.3 mmol/l] for inclusion) also did not find a difference in childhood obesity at age 7 in treated vs. untreated GDM . Notably, there was a marked difference in macrosomia (> 4,000 g) rates in these children between the treated and untreated maternal groups (4.6 % vs. 13.6 %, p < 0.001) . Pham et al.  found that large for gestational age (LGA), but not GDM, was a risk factor for toddler obesity developing by age 2-4 years. Our current study in normal birth weight infants in a large diverse population with the full clinical spectrum of maternal hyperglycemia found the impact of maternal hyperglycemia on childhood obesity risk increases over the first decade, most evident at age 5 and older. Since Pettitt et al.  landmark study in Pima Indians over 25 years ago, most of the focus of childhood obesity risk with GDM has centered around LGA babies. Our results suggest metabolic imprinting occurs with maternal hyperglycemia even when birth weight is normal.
Excessive maternal weight gain is an independent risk factor for childhood obesity, but few studies have also evaluated maternal hyperglycemia . One recent within-family study found excessive weight gain to increase childhood obesity risk . Sridhar et al.  recently found that excessive weight gain increased childhood obesity risk at age 2–5 years, independent of GDM. Thus, our results that weight gain of >40 pounds independently increases childhood obesity risk across the first decade, even after adjusting for the spectrum of maternal hyperglycemia including GDM, strongly suggest that avoiding excessive maternal weight gain is an important potentially modifiable mechanism to reduce childhood obesity burden. Excessive maternal weight gain (>40 pounds, excessive by IOM guidelines  occurred in ~20 % of women. We used >40 pounds as our criteria as this is excessive weight gain regardless of pre-pregnancy BMI. In the 5926 women with measured pre-pregnancy BMI, we conducted additional analysis using BMI-specific IOM gestational weight gain criteria (replacing exceeding >40 pounds with exceeding IOM guidelines as the risk factor). We found consistent increased risk for both childhood overweight and obesity, though the HRs were smaller than those associated with > 40 pound weight gain (data not shown). Thus, finding ways to keep gestational weight gain within IOM guidelines offers an important strategy to reduce childhood obesity burden.
Both maternal hyperglycemia and excessive weight gain are, in a sense, “overfed” states. Our results suggest this “overfeeding” (whether via excess glucose or overall calories) may then imprint the child for an overfed metabolism. Although more research is needed to delineate potential mechanisms and lifecourse risk factors that lead to excessive weight gain, as we await this research, dietary interventions early in pregnancy can now potentially have a dual impact to prevent (or treat) gestational diabetes and excessive weight gain.
Dietary interventions in pregnancy could thus have a large population impact to not only optimize maternal health, but also reduce future childhood obesity. In clinical practice, it is well recognized that pregnant women are more likely to embrace health behavior change—e.g., smoking cessation, alcohol abstinence—to maximize their baby’s health and future . A dietary intervention to prevent excessive weight gain in this population would have a high likelihood of success and very little, if any, untoward risk, while giving babies a metabolically “fair chance” as they grow and mature through childhood.
This study’s many strengths include the enormous sample size of nearly 25,000 individuals (mothers and offspring) in a diverse HMO population with universal GDM screening, eliminating potential bias from non-participation. Moreover, GDM treatment was based on the same practice guidelines. Multiple objectively measured childhood heights and weights between ages 2-10 in their offspring is also an enormous strength. Another is the ability to determine the final included sample is reasonably representative of the entire population of normal birth weight babies 1995–2003 who were excluded due to disenrollment and/or missing childhood BMI data. Bias due to attrition is unlikely as maternal glycemia for included births was very similar after a 50 g GCT compared to the excluded births (mean GCT 115.5 mg/dl vs. 114.8 mg/dl; p = 0.1104), as was average maternal weight gain (14.39 kg for included vs. 14.42 kg for excluded; p = 0.4043). We propose our findings are generalizable to the entire spectrum of birth weights as we selected those babies least likely to be at risk for obesity (i.e., normal birth weight).
Every study also has potential limitations. Measured pre-pregnancy BMI, a strong predictor of childhood obesity, was available in the EMR for only 5926 mothers. At the time most women present to their OB provider (already pregnant), it is too late to modify pre-pregnancy BMI. Therefore, our results that both maternal hyperglycemia and excessive weight gain remained independent risk factors for childhood obesity after adjustment for pre-pregnancy BMI in the 5926 mothers, is fundamentally important as both of these risk factors are potentially modifiable during pregnancy. Although breastfeeding history was available in only a small sample of 302 mothers queried in Hawaii, our findings that multivariate adjusted childhood obesity reduced risk by almost half (51 % of children had obesity if not breastfed vs. 37 % if their mothers breastfed) provides another compelling reason to recommend breastfeeding to new mothers.
It is important to recognize that the Centers for Disease Control and Prevention (CDC) has targeted several lifestyle risk factors for childhood obesity needing intervention, including reduction in sugar-sweetened beverage and fast-food consumption, and screen time (computers, TV, video games) as well as increased breastfeeding, physical activity, and consumption of fruits and vegetables . Although these lifestyle behaviors are not well captured by the EMR, our results suggest that avoiding excessive maternal weight gain and hyperglycemia may be important modifiable mechanisms to provide every normal birth weight child with a normal metabolic start, and thus the best chance to remaining normal weight in childhood and beyond.
In summary, in our very large population of nearly 25,000 mothers and their normal birth weight offspring with objectively measured childhood BMI across the first decade of life, we found both maternal hyperglycemia and excessive weight gain independently increase childhood obesity risk. Although there are, of course, other important modifiable risk factors for childhood obesity after birth, it is desirable to start life metabolically with the best chance of remaining normal weight. And while obesity prevention and treatment—in childhood and the general population—require multiple strategies on several fronts, pregnancy clearly presents an important window of prevention opportunity to reduce childhood obesity risk in future generations. Future research should include clinical interventions to manage gestational weight gain and/or maternal glucose as a means to reducing childhood obesity.
This work is supported by a Research Award from the American Diabetes Association, and grant award 1R01HD058015 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (both awards to TH). We would also like to thank Martie Sucec for her editorial review and Robin Daily for her formatting assistance.
Compliance with Ethical Standards
Conflict of interest
The authors declare that they have no conflict of interest.