Growth patterns from birth to overweight at age 5‐6 years of children with various backgrounds in socioeconomic status and country of origin: the ABCD study

Summary Introduction Children from minority groups are at increased risk of overweight. This study compared BMI growth patterns from birth onwards of boys and girls with overweight at 5‐6 years, according to socioeconomic status (SES) and country of origin, in order to gain more insight into the critical periods of growth to overweight. Methods A total of 3714 singletons of the multi‐ethnic ABCD study were included. Within children with overweight at age 5‐6 years (N = 487, prevalence boys: 11.6%, girls: 14.6%), BMI growth patterns from birth onwards (12.8 serial measurements; SD = 3.1) were compared between children from European (69.4%) and non‐European mothers (30.6%), and between children from low (20.8%), mid (37.0%) or high SES (42.2%), based on maternal educational level. Results BMI growth to overweight did not differ between children of European or non‐European mothers, but it did differ according to maternal SES. Children with overweight in the low and mid SES group had a lower BMI in the first 2 years of life, an earlier adiposity rebound and increased in BMI more rapidly after age 2, resulting in a higher BMI at age 7 years compared to children with overweight in the high SES group [∆BMI (kg/m2) between high and low SES: boys 1.43(95%CI:0.16;3.01) and girls 1.91(0.55;3.27)]. Conclusion Children with overweight from low SES have an early adiposity rebound and accelerated growth to a higher BMI at age 5‐6 years compared to children with overweight from the high SES group. These results imply that timing of critical periods for overweight development is earlier in children with a low socioeconomic background as compared to other children.


| INTRODUCTION
Over the past decades there has been a worldwide increase in the prevalence of overweight and obesity among children. 1 Children with overweight or obesity are at increased risk of adult overweight and its associated comorbidities. [2][3][4][5] Early prevention of childhood overweight is therefore of absolute importance and identifying critical periods in the development of childhood overweight is an essential step towards effective prevention.
BMI growth patterns differ from early age onwards between children who will develop overweight in late childhood compared to those with normal weight. [6][7][8] Children with overweight have higher birthweights, and faster rates of growth in the first years of life. [6][7][8][9][10][11] Moreover, their BMI peak tends to be higher and their adiposity rebound-a renewed rise in BMI after the decrease in BMI that followed the BMI peak-tends to be earlier with a steeper increase in BMI compared to those of normal weight. 6,7,[12][13][14] This general pattern to overweight is likely to reveal differences when comparing lower vs higher socioeconomic groups, or ethnic minority populations vs the host populations. Although this has not been studied directly, previous studies provide indications to suggest such differences in growth patterns. For instance, children of low socioeconomic background are born smaller, and are more likely to be overweight at age 5 years, 15,16 showing an increased growth in the first years compared to children of high socioeconomic background. In addition, children in ethnic minority populations are more often small at birth, but have more obesity in childhood compared to children from Dutch origin. [17][18][19][20] This shows that they grow faster after birth, compared to children of Dutch origin. There are indications that these differences in early postnatal growth are partly due to ethnic and socioeconomic differences in feeding practises. [21][22][23][24][25][26][27] Therefore, this study will investigate BMI growth patterns to overweight at age 5-6 years using longitudinal measurements of BMI from birth onwards stratified for maternal country of origin and maternal socioeconomic background. Country of origin will be indicated by country of birth (European origin vs non-European origin) and socioeconomic background by maternal educational level. Growth curves will be described for boys and girls separately, as growth patterns of boys and girls have been shown to differ. 28,29 Disparities in timing of the critical periods in BMI growth to overweight could help to better identify children in minority groups at risk for childhood overweight at an early stage of development.

| Study population
This study was part of the Amsterdam Born Children and their Development (ABCD) study. The aim of the prospective cohort ABCD study was to examine the association between maternal lifestyle, medical, psychosocial and environmental conditions during pregnancy and children's health at birth as well as in later life. 30

| BMI growth pattern
Growth data were obtained from the Youth Health Care registration of the Public Health Service in Amsterdam. Following a standard procedure, trained nurses measured at regular follow-up moments between birth and age 6 years weight and height and BMI was calculated (mean = 12.8; SD = 3.1). For each child an individual BMI growth curve was modelled. 31 Since the length of the child was first measured at 4 weeks of age, the BMI score at birth was extrapolated.

| Overweight at age 5-6 years
BMI was based on height and weight measures at 5-6 years old, obtained from the ABCD health check (69.4%) complemented by growth measurements of the Youth Health Care registration (30.6%).
Height was measured to the nearest millimetre using a Leicester portable height measure (Seca), and weight to the nearest 100 g using a Marsden weighing scale (model MS-4102). 32 The agreement between the two registration for height and weight could be estimated in 1531 children who were measured during the ABCD health check as well as by YHC professionals within 1 year. The intra class correlation (ICC) of height was 0.83 and for weight 0.90 which indicate good to excellent agreement. Children were divided in normal weight and overweight/ obese, by sex-and age specific BMI cut-off points defined by the International Obesity Task Force. 33

| Country of origin
Country of origin of the child was based on the country of birth of the mother and her mother, reported in the pregnancy questionnaire, to include first and second generation. 17 The children were divided in two groups: European origin (European countries, United States, Canada, Australia and New Zealand) and other.

| Socioeconomic status
SES was based on maternal education in years after primary school, reported in the pregnancy questionnaire and divided in three categories: low (0-5 years), middle (6-10 years) and high (>10 years).

| Characteristics of the study population
The following variables were used for demographic information: maternal age (years) and maternal smoking during pregnancy (yes/no) which were self-reported in the pregnancy questionnaire, sex (boy/girl), gestational age (days), parity (nulliparous/multiparous), and birth weight of the child (gram) were retrieved from the Youth Health Care registration.

| Analyses
Characteristics of the study sample were described according to weight status at age 5-6 years for boys and girls separately. Demographic differences between children with normal weight and children with overweight as well as differences between the response and nonresponse group (live-born singletons with permission to follow-up meeting the inclusion criteria, but without growth data and/or BMI status at 5 years of age) were tested with chi-square tests for categorical variables and two-sample t tests for continuous variables.
To model the overall longitudinal BMI growth patterns of normal weight and overweight, two separate sex-specific linear mixed models were fitted to the BMI growth data. 34 To capture the non-linearity of the BMI growth patterns, the fixed effect describing the relation between BMI and age were modelled with a natural spline function. The knot placement of the natural splines was based on quantiles of age.
For both models, the best number of knots was chosen based on the Bayesian Information Criterion (BIC). To quantify the difference between normal weight and overweight, interaction terms were added to the model. Finally, we used a random intercept and slope to capture the correlation between BMI growth measurements from the same child.
Firstly (Part 1), differences in BMI growth patterns (BMI peak, BMI adiposity rebound) between children with normal weight and overweight were described by visual inspection. Differences between BMI at certain ages (0, 3, 6, 9 months, 1, 2, 3, 4, 5, 6, 7 years) were checked by investigating the amount of overlap of the 95% confidence intervals. If the 95% confidence intervals are not overlapping, there is some evidence that the BMI growth patterns are different.
F I G U R E 1 Flowchart of the study population Secondly (Part 2), the population was restricted to only children with overweight (N = 487). The same modelling procedure was followed as in Part 1. The BMI growth curves were compared for children who had overweight at age 5-6 years of European origin and of other origin.
Thirdly, to adjust for the interrelation of SES and country of origin, the population was further restricted to children with overweight from only European origin (N = 206). BMI growth curves were compared of children with overweight from high SES, middle SES and low SES groups. We did not correct for any covariates as potentially important variables as maternal smoking, birth weight and maternal weight status are strong determinants of postnatal growth, both influenced by ethnicity 17 and SES 15,16 and are therefore part of the causal pathway. A secondary analysis with subdivision of the non-European group into children from Turkish, Moroccan and Surinamese decent was Prevalence of overweight in our study population

| Study population
In our study population 13.1% of the children were overweight.
Boys and girls with overweight had respectively a 3.4 and 3.6 kg/ m 2 higher BMI than children with normal weight at age 5-6 years.
Overweight was more prevalent in girls (14.6%) than in boys (11.6%) and in children with a non-European origin (24.8%) and low SES (15.6%) background (Table 1). Children with overweight more often had mothers who were younger, more often overweight themselves, multiparous and smoked more often than the mothers of children with normal weight ( Table 2).

| Part 1: Total population
Based on the BIC, the optimal number of degrees of freedom for the natural spline function describing the relation between BMI and age was 7 for both boys and girls. Overweight boys at age 5-6 years had a higher birth weight ( Table 2) and higher BMI at birth (Table 3) compared to normal weight boys.). Figure 2 presents the average BMI growth curves of children with normal weight and overweight at age 5-6 years. Visual inspection of this figure shows that in children with overweight, the BMI peak was higher (±1 kg/m 2 ) than for children with normal weight. Although timing of the BMI peak was earlier in overweight boys than in normal weight boys, no differences were seen in girls. On the other hand, timing of the adiposity rebound was earlier in children with overweight (at age 3 years) than in children with normal weight (at age 5 years). The magnitude of the drop in BMI after the BMI peak differed between boys with overweight (ΔBMI = ±1 kg/m 2 ) compared to boys with normal weight (ΔBMI = ±2 kg/m 2 ), similar findings were found for girls (ΔBMI = ±0.5 kg/m 2 for girls with overweight and ΔBMI = ±1.5 kg/m 2 for girls with normal weight.
3.3 | Part 2: Children with overweight at age 5-6 years 3.3.1 | BMI growth curves in children with overweight at age 5-6 years: European vs non-European origin children Based on the BIC, the optimal number of degrees of freedom for the natural spline function describing the relation between BMI and age was 5 for both boys and girls. Figure 3 (lower panel) presents the BMI growth curves to overweight of boys and girls from European origin and non-European origin. No differences were found in birth weight and subsequent growth patterns to overweight between both populations (Table 3 and Table S3). The secondary analysis indicated differences within the non-European group, which were most pronounced for children from Turkish decent. Girls with overweight at age 5-6 from Turkish decent showed the highest BMI peak, the earliest adiposity rebound and the highest BMI from age 2 onwards ( Figure SS1). (only statistically significant for the low group; Table 3) and an earlier adiposity rebound (Figure 3) than children from the high SES group. After age 3 years BMI increased rapidly in the low/middle SES group and at age 7 years the children with overweight in low/middle SES groups had a higher mean BMI than those in the high SES group (Table 3).

| DISCUSSION
This study investigated socioeconomic and country of origin inequalities in BMI growth patterns to overweight at age 5-6 years, for boys and girls separately. No differences in BMI growth patterns to overweight were found when comparing European origin children with non-European origin children. However, children with overweight in the low/middle SES group had a lower BMI during the first 2 years, an earlier adiposity rebound and a higher BMI at age 5-6 years compared to children with overweight in the high SES group.

| Strengths and Limitations
An important strength of this study was its longitudinal design with,

| Underlying mechanisms
There are different mechanisms that could underlie the SES differences found in our cohort, but infant feeding patterns can probably be ruled out. Although children with low SES in our cohort had shorter breastfeeding duration and earlier introduction of solid foods, 44 which is associated with the development of childhood overweight, 45,46 they had lower BMIs in the first 2 years.
Moreover, within our cohort, differences in feeding style did not mediate the association of low SES with weight-for-length gain from 1-5 years. 44 The lower BMI in the first 2 years might be explained by maternal smoking. Maternal smoking during pregnancy impairs foetal growth and causes low birth weight, [47][48][49] which often leads to catchup growth and an increased risk of overweight in childhood. 50 In this study, 35% of the mothers with a low SES smoked during pregnancy, this was even higher (51%) if their children were overweight at age 5-6 years (Table S5). A previous study within our cohort showed that maternal smoking during pregnancy largely attributed to the higher prevalence of infants born small-for-gestational age in the low SES group. 51 Therefore, the socioeconomic differences in growth patterns to overweight observed in infancy could be explained by maternal smoking during pregnancy. Furthermore, in the earlier mentioned study on the association between SES and weight-for-length gain from 1 to 5 years, the most important mediator was maternal prepregnancy BMI. 44 The increased growth after two years might therefore be a consequence of maternal overweight, as prevalence of maternal overweight is higher in women of low SES. 52-55 related to an unhealthier lifestyle. Important childhood mediators in the inverse association between SES and childhood overweight at age 5 years are maternal TV watching, consuming breakfast and TV watching by the child. 56 More in general, not participating in organized sport was found to be associated with increased BMI z-scores, 57 and children with overweight from a low socioeconomic background sleep shorter than their peers with normal weight. 58,59 In our study, children in the low SES group had more often mothers with overweight, they watched more TV and slept less, and were less often a member of a sports club (Table S5). Another explanation for the accelerated growth of children from the low SES group might be that mothers/parents perceive the child's overweight as normal and therefore take no action to change their lifestyle. Maternal underestimation of child's weight is more common in mother from low SES compared to mother from high SES. 60 The recognition of child's overweight is an important first step for the success of interventions aimed at prevention of further accelerated weight gain.

| Implications
This study showed that children with overweight from the low/middle SES group have lower BMIs in the first year compared to the high SES group, but after 2 years, these groups accelerated in growth and had a higher BMI at age 5-6 years. Therefore, it is important to closely monitor children with a low SES background as their weight might be in the normal range before the age of 2, but this does not prevent them from being at risk of developing overweight later on. An early adiposity rebound should be an indicator for extra care, but to identify the adiposity rebound, it is necessary to measure anthropometrics frequently, for instance every 3-4 months, in both infancy and early childhood. This is of foremost importance as BMI in childhood is a predictor of adult BMI [3][4][5] and childhood SES has a greater influence on adult BMI and the prevalence of overweight/obesity than adult SES. 61 A combination of both a low SES and high BMI in early childhood might therefore be most detrimental for adult BMI.

| CONCLUSION
To our knowledge, this is the first study that compares growth patterns to overweight between children from different socioeconomic and country of origin groups. Stratification by SES and country of origin background showed that the development of overweight is broadly similar for children from European origin and non-European origin. However, children with overweight from a low SES background had a lower BMI during the first 2 years, but an earlier adiposity rebound and a more rapid BMI development after 2 years compared to children in the high SES group. Our results imply that within the Youth Health Care a distinction should be made according to children's SES in determining the critical periods in childhood growth.