Associations between gestational weight gain and weight development of the offspring: Differences depending on maternal pre-pregnancy BMI

For the outlined project, we retrieved data collected within the LIFE Child study conducted at the Research Center for Civilization Diseases at Leipzig University, Germany. The LIFE Child study is a large population-based cohort study [24, 25], established in 2011, which examines factors (e.g., nutrition, physical activity) that contribute to the occurrence of civilization diseases such as obesity, diabetes, and cardiovascular diseases during infancy, childhood, and adolescence (0–21 years).

In order to explore how conditions during pregnancy impact the health of children, the LIFE Child study assesses pregnant women during their 24th and/or 36th week of gestation. Infants are examined at the age of 3, 6, and 12 months, and thereafter once per year until age 21. In general, parents and children who suffer from chromosomal or syndromic diseases are excluded from the LIFE child study.

Participants are mainly recruited via advertisement and word of mouth. All parents provided informed written consent before participation. This study was designed in accordance with the Declaration of Helsinki and the study program was approved by the local ethics committee (Reg. No. 477/19-ek).

Data from 804 mothers and their children, collected until 2020 (before the COVID-19 pandemic), were analyzed. From this dataset, 113 mother–child pairs were excluded due to conditions which might influence weight development during pregnancy (e.g., gestational diabetes) or weight development of the child (e.g., preterm birth). More specifically, we excluded twins and triplets, women and children with diabetes (Type I, Type II and gestational), and children who were born preterm (< 37 weeks of gestation; < 260 days) or post-term (> 42 weeks of gestation; > 293 days). Furthermore, outliers were removed (GWG > 50kg, n = 1, and weight loss during pregnancy, n = 2). The final dataset included 691 mother–child pairs.

In Germany, 10 preventive medical check-ups (“U-examinations”) screen children for developmental impairments or diseases. In this study, we assessed children’s BW (n = 691) and their BMI at nine time points documented in the preventive check-up booklet, namely, 3–10 days after birth (n = 452) and at 4–5 weeks (n = 688), 3–4 months (n = 654), 6–7 months (n = 573), 10–12 months (n = 458), 21–24 months (n = 323), 34–36 months (n = 238), 46–48 months (n = 169), and at 60–64 months (n = 92). The decrease in the number of available data with increasing child age can be explained by the decreasing (re-)participation of older children in the LIFE Child study.

The data on maternal age, height, and weight were obtained from the maternity log, collected by the attending gynecologist at all prenatal examinations. PpBMI was calculated as weight before pregnancy (kg) divided by the square of height (m²). PpBMI was analyzed as a continuous variable or as a categorical variable (ppBMI category), with the two subgroups n/u weight (< 25 kg/m²) and o/o (≥ 25 kg/m²). The GWG period was defined as weight development from the last reported weight before pregnancy to the last reported weight during pregnancy, if measured less than 21 days before the delivery date.

The BW and BMI of the children documented at the time points of the “U-examinations” were used for the analyses, after transforming the weight and BMI measurements into age- and gender-adjusted standard deviation scores (SDS) according to the German growth standard (Kromeyer-Hauschild) [26]. At birth, we decided for BW and against BMI as length measurement at birth can be inaccurate. BW was additionally grouped into three categories: large for gestational age (LGA; BW-SDS ≥ 1.28), appropriate for gestational age (AGA; BW-SDS ≥  − 1.28 to 1.28), or small for gestational age (SGA; BW-SDS <  − 1.28).

SES was measured as a composite score combining information on parents’ education, professional position, and family income [27]. The score (range 3–21) can be used to categorize a family’s SES as high, medium, or low. Given the low percentage of low SES families in the present sample (1%), we combined low and middle SES and compared this group with the high SES group.

All analyses were performed using R version 4.2.1. Data were described in terms of mean/SD (continuous measures) and numbers/percentages (categorical measures). Associations between the independent variables (age, ppBMI, GWG, and SES) and the dependent variables (weight [status] at birth and BMI-SDS at U-examinations) were assessed using univariate linear (for BW and BMI-SDS) and logistic (for weight status at birth) regression analyses. In all analyses, the effects of GWG were reported by five kg interval. Furthermore, all associations were checked for interactions between the mother’s ppBMI category (o/o versus n/u weight) and GWG. The ppBMI categories overweight and obesity as well as underweight and normal weight were combined as the effects did not differ between these categories. Models were checked for variance inflation using the generalized variance inflation factors (GVIF^(1/2*Df)) < 2). Given that the interactions were significant in most analyses, we reported associations between GWG and the dependent variables separately for each ppBMI category.

The total sample consisted of 691 mother–child pairs. Table 1 shows the demographic characteristics of the sample. The mean BW-SDS was 0.17. Five percent of the children were SGA, 82% were AGA, and 12% were LGA. The mean maternal ppBMI was 23.2 kg/m². Most mothers (72%) were normal weight, 6% underweight, 16% overweight, and 6% obese. The GWG of women varied considerably from 0.4 to 34 kg, with a mean of 15kg. Figure 1 depicts women’s adherence to the IOM guidelines regarding their GWG, differentiated by their ppBMI category: most women did not meet the IOM recommendations [15] for GWG: 10% of underweight, 49% of normal weight, 68% of overweight, and 66% of obese women exceeded the recommended limits (i.e., more than 18 kg, 16 kg, 11.5 kg, and 9 kg, respectively).

Bild vergrößern

Regarding SES, 29% of women had a high SES and 45% had a low or medium SES. In 26% of cases, SES information was unavailable. The mothers who participated in our study were aged from 20 to 46 years, with a mean of 31 years.

Maternal age was significantly associated with children’s BW-SDS (β_age = 0.02, p = 0.02, 95% CI 0.00–0.04), but not with the risk of LGA (OR = 1, p = 0.5, 95% CI 0.97–1.07). The associations between maternal age and children’s BMI-SDS at the different preventive check-ups only reached statistical significance at the time point of U9, i.e., at the age of 5 years (β_age = 0.04, p = 0.03, 95% CI 0.00–0.07). There were no significant associations between maternal age and BMI-SDS at time points U2 to U8 (β_age between −0.02 and 0.01, all p > 0.08).

There was no significant association between SES and BW of children (β_SES = −0.07, p = 0.4, 95% CI −0.24 to 0.14) or the risk of LGA (OR = 1.2, p = 0.5, 95% CI 0.69–2.08). Similarly, there was no significant association between SES and children’s BMI-SDS at time points U2–U9 (β_SES between −0.08 and 0.27, all p > 0.12).

In women with o/o, BW was significantly higher than in women with n/u (β_o/o=  + 0.43, p < 0.01). Significant interactions between GWG and ppBMI showed that the strengths of associations between GWG and BW differed by maternal ppBMI category. In women with n/u, GWG showed a strong association with BW-SDS (β_GWG = 0.05, p < 0.01, 95% CI 0.03–0.07). In these women, the estimated mean BW-SDS was −0.14 for a GWG of 10 kg and 0.36 for a GWG of 20 kg. In comparison, in women with o/o, GWG showed no significant association with BW-SDS (β_GWG = 0.0002, p = 0.99, 95% CI −0.03 to 0.03) and the estimated mean BW-SDS was 0.29, independently of GWG.

Logistic regressions revealed a higher risk of bearing a LGA child for women with o/o compared to women with n/u (OR = 3, p < 0.01, 95% CI 1.34–6.97). In women with n/u, the risk for LGA increased strongly with GWG (OR = 1.6, p < 0.01, 95% CI 1.23–2.25), as shown in Table 2 and Fig. 2. In women with n/u, the estimated risk of LGA was 6% for GWG of 10 kg, but 15% for GWG of 20 kg. In women with o/o, GWG was not significantly associated with the risk of LGA (OR = 1.01, p = 0.9, 95% CI 0.7–1.5). In these women, the estimated risk of LGA was 16%, irrespectively of GWG. The risk of LGA in mothers with n/u only approaches that of mothers with o/o if GWG is 20 kg or more (Table 2 and Fig. 2).

Bild vergrößern

As with BW, GWG in women with n/u before conception was positively associated with the BMI-SDS of the children at 1 week to 5 years of age, with an almost constant effect size (β_GWG between 0.06 and 0.2, Fig. 3a). Statistical significance was reached at the time points of U2, U6, U7 (each p < 0.01), U8 (p = 0.01), U5 (p = 0.05), and U9 (p = 0.04) (Table 3). On the contrary, in mothers with o/o, GWG was at no time point significantly associated with children’s BMI (Fig. 3a and Table 3). However, children of mothers with o/o had a generally higher BMI-SDS than children of mothers with n/u (β_GWG between 0.07 and 1.0) (Fig. 3b).

Bild vergrößern

Previous studies found that GWG was positively associated with infants’ BW [8, 28, 30] and with a higher risk of o/o throughout childhood [9‐11]. Furthermore, higher maternal BMI before pregnancy was associated with higher child weight at birth and in later development [9, 31].

In the current literature, there is no consensus on how the maternal BMI level might modify the association between GWG and BW [6, 14, 21, 22]. In line with previous findings [8, 20], we found that GWG has a stronger effect on BW in n/u mothers than in mothers with o/o. Zhao et al. suggested that in normal-weight mothers, GWG could be an independent predictor for adverse BW [29]. In women with o/o, different associations between GWG and BW were described [6, 14, 21, 22].

Concerning the further weight development of children after birth, prior studies observed significant associations between excessive GWG and children’s BMI-SDS only in normal-weight mothers [31], or only small additional effects of excessive GWG in mothers who were already o/o [9].

An explanation for why excessive GWG and maternal ppBMI are associated with children’s (birth) weight is provided by the development overnutrition hypothesis. Obesity, insulin resistance, and excessive GWG cause high glucose and triglyceride levels in pregnant women, which are transferred transplacentally, inducing higher levels of blood sugar and nutrients in the fetus [10, 32]. In response, the fetal pancreas starts to produce greater amounts of insulin, triggering fetal growth [29, 33‐36].

Excessive GWG and ppBMI, as described before, cause higher maternal and fetal glucose levels, which may shape the fetal metabolism and organ and tissue structure through epigenetic mechanisms in utero [4]. This fetal programming could increase susceptibility to obesity throughout the offspring’s lifespan [37]. These explanations could also explain our findings on the association between ppBMI, GWG, and weight of children up to 5 years of age. Likewise, Arroyo-Jousse et al. indicate that maternal overnutrition and obesity could lead to higher expression of the hormones leptin and adiponectin in maternal and fetal adipose tissue and could modulate the placental function and fetal physiology, which may lead to obesity in later life [38].

One possible reason why the effect of GWG is particularly strong in women with n/u is that, before pregnancy, women with n/u usually have a healthier insulin and glucose metabolism than women with o/o. In n/u mothers, high GWG can initiate an adverse metabolic situation, whereas in mothers with o/o, the metabolic condition might already be adverse at the beginning of pregnancy. Our study reveals that an n/u mother only approaches the same risk of LGA as women with o/o if she gains more than 20 kg during pregnancy. This might indicate that a GWG of 20 kg leads to a similar metabolic situation as being overweight or obese.

The strengths of our study are the large sample size, the consideration of the effect of ppBMI category on the association between GWG and BW, and the investigation of weight development until age five. Nevertheless, there are several limitations. There is a bias in our study population, since mother–child pairs with lower SES were under-represented. Most participants lived in a large city. In addition, the exclusion of women with (gestational) diabetes as well as children born pre- or post-term created a supernormal collective. Moreover, not all confounding factors that might influence the association between GWG or ppBMI and children’s weight development were considered, e.g., maternal energy intake, smoking, alcohol consumption, movement, parity, birth order, or paternal BMI [39, 40]. Also, we did not distinguish during which period of pregnancy the weight was gained [8].