Association between birth weight and risk of abdominal obesity in children and adolescents: a school-based epidemiology survey in China

Data in this study comes from the baseline cross-sectional survey of a national multicenter cluster-controlled trial aimed to reduce the burden of obesity in children and adolescents (Registration number: NCT02343588). The survey was conducted in seven provinces or cities of China (Hunan, Ningxia, Tianjin, Chongqing, Liaoning, Shanghai, Guangzhou). Design and sampling procedure have been reported in detail elsewhere [25]. Briefly, a multi-stage cluster sampling method was used to select participants. At first, several regions were randomly selected from each province/city, and 12 to 16 schools were randomly chosen from each region. In each school, two classes were randomly selected in each grade and the whole class students were invited to participate in this survey. Those were recruited in this survey only after they and their parents signed informed consents voluntarily. All survey sites used the same protocol during the implementation process, and all processes of randomization were performed by a staff member who was not involved in the survey.

A total of 41,003 primary and middle school students aged 6–17 years old were enrolled in this study initially, of whom 30,486 remained in the analytical sample for the present work after excluding participants with missing records of urban/rural area (n = 234), weight (n = 1387), height (n = 1630), waist circumference (n = 1786), birthdate (n = 2780) or birth weight (n = 3902). The trial was approved by the Ethical Committee of Peking University (IRB0000105213034), and all participating children and their parents signed the informed consent forms for the physical examination and questionnaire survey.

Anthropometric measurement was conducted by trained professional staff according to the standard protocol. Participants were asked to wear their underwear when waist circumference (WC, cm) and height (cm) were measured. WC was measured at the midpoint between the iliac crest and lowest rib by Myotape scale (Accufitness, Green Villge, Colordo, USA) with 0.1 cm precision, and height was measured using the portable stadiometer (model TZG, Jiangyin Hongya Science and Education Equipment Co., Ltd.) with 0.1 cm precision. Both WC and height were measured twice and the average of two records was used in data analysis. Waist-to-height ratio (WHtR) was calculated as WC divided by the height. The cut-off value of 0.5 was used for WHtR to classify abdominal obesity [26].

Students’ questionnaire and parental questionnaire were used to collect the information from children and their parents, respectively.

The students’ questionnaire was completed by themselves in children of grade 4 to 12 (with age range from 10 to 17 years old) and was completed under the assistance of their parents in children of grade 1 to 3 (with age range from 6 to 9 years old).

The students’ questionnaire was used to collect information of dietary behavior, including the consumption of fruits, vegetables, total meat, and sugar-sweetened beverages (SSBs) over the past 7 days. A serving of fruit/vegetable was defined as the size of an adult fist (approximately 120 g) [27], a serving of SSBs was defined as a canned beverage (approximately 250 ml), and a portion of meat equals to the size of an adult’s palm (approximately 100 g) [28]. Students reported the frequency (days) and amount (servings) of food and drink intake per day, the daily dietary intake was estimated as follow: average daily intake = (days × servings in those days) /7.

Information of child’s physical activity was recorded by the International Physical Activity Questionnaire-Short (IPAQ-S) [29], which has been widely used in children and adolescents. All recruited students were asked to report the frequency (days) and duration (hours and minutes) of moderate to vigorous-intensity physical activities (MVPA) over the past 7 days, and the average daily time of MVPA was calculated as follow: average daily time = (days × duration in those days)/7.

The parental questionnaire was filled out by parents at home. Paternal and maternal height and weight were reported in the parental questionnaire and then used to calculate body mass index (BMI), a cut-off point of BMI ≥ 24 kg/m² was applied to define parental overweight/obesity. Parental educational level was classified into four categories (junior high school or below, senior high school, junior college, and college or above). Parents were required to report their children’s early life information, including birth weight, gestational age and number of fetus (singleton, twice or more) based on the birth certificate or health clinic record [30]. If they did not have it, parents were asked to recall the birth weight based on their own measurements. About 70.9% of parents report the information of their children’s early life based on the birth certificate or the health clinic card. Besides, information about feeding patterns (breastfeeding, not breastfeeding) and number of children (single child, two or more children) were additionally obtained. According to birth weight, participants were divided into five categories: < 2500 g, 2500-2999 g, 3000-3499 g, 3500-3999 g, and ≥ 4000 g [31].

Descriptive results were characterized as means ± standard deviations for continuous variables and percentages for categorical variables. One-way ANOVA, independent sample Student’s t-test and Chi-square test were used to compare the distribution of descriptive characteristics between boys and girls. The nonlinear association between birth weight and WHtR was performed by fractional polynomial regression model. Logistic regression models were performed to assess the odds ratios (ORs) and 95% confidence intervals (CIs) for abdominal obesity when participants with birth weight of 2500–2999 g were used as the reference group. The crude model was adjusted for sex (only in total group), residence area, and age, while the adjusted model was further adjusted for gestational age, delivery mode, fetus number, feeding pattern, single-child status, paternal overweight/obesity, maternal overweight/obesity, paternal educational level, maternal educational level, daily fruit and vegetable consumption, daily meat consumption, daily SSBs consumption, and daily moderate to vigorous physical activity. Since about 25.6% of the original study sample had to be excluded for a variety of reasons, a sensitive analysis was conducted in the initial sample of 41,003 children. In addition, we have also done a sensitive analysis by extracting participants with original birth weight records (n = 21,615). All analyses were conducted using SPSS 20.0 (IBM, NY, USA) and Stata 14 software (College Station, TX, USA) and a two-sided P value of < 0.05 was considered statistically significant.

A total of 30,486 students aged 6–17 years were included in the analysis. As showed in Table 1, 52.1% of the participants were boys, and 59.9% lived in urban areas. The mean birth weight was (3324.4 ± 502.9) g in total participants, which was higher in boys (3366.8 ± 513.9 g) than in girls (3278.5 ± 486.7 g) (P<0.01). The prevalence of abdominal obesity was 16.4% in total participants with a higher rate in boys (20.5%) than that in girls (12.0%) (P<0.05).

The nonlinear relationship of birth weight and WHtR is shown in Fig. 1. A J-shaped association was observed in the crude model. This association was slightly changed after further adjusting for confounding factors, with the range of birth weight associated with the lowest WHtR around 2500 g. In the stratified analysis, similar patterns were detected both in boys and girls. However, the birth weight associated with the lowest WHtR in boys was lower than that in girls.

Figure 2 shows the association between birth weight and risk of abdominal obesity in boys and girls. The crude model shows that compared with participants whose birth weight ranged from 2500 to 2999 g, those who had higher birth weight (3000–3499 g, 3500–3999 g, ≥ 4000 g) are associated with an elevated risk of abdominal obesity. Though lower ORs were detected in participants with birth weight < 2500 g, especially in girls, the difference was not statistically significant (P = 0.692). When further adjusted for potential confounding factors, only minor changes were observed for these estimations. The birth weight (≥ 3000 g) was positively associated with higher risk of abdominal obesity (for 3000–3499 g: OR[95% CI] = 1.13[1.01,1.25]; for 3500–3999 g: OR[95% CI] = 1.22[1.09,1.37]; for ≥4000 g: OR[95% CI] = 1.50[1.31,1.71]). In addition, the ORs of abdominal obesity were higher in boys than those in girls in the same birth weight group. Participants had an evelated risk for those with a birth weight ≥ 3000 g in boys and ≥ 4000 g in girls. Based on the original sample (n = 41,003) and original birth record sample (n = 21,615), we have done two sensitive analyses and have come to similar results with the present study (Table S1 and Table S3).

To investigate whether the association varied among age groups, participants were divided into three groups according to their ages. Table 2 shows that high birth weight was related to abdominal obesity regardless of the age groups. Participants with birth weight ≥ 4000 g had higher risk of abdominal obesity in all three age groups, with the ORs ranging from 1.27 (95% CI: 1.03–1.58) to 1.59 (95% CI: 1.29–1.97). Participants with birth weight between 3000 and 3999 g also showed an elevated risk, though the statistically significant result was only found in children aged 6–9 years. In addition, across the age groups, no significant association between low birth weight (< 2500 g) and abdominal obesity was found. Based on the original sample (n = 41,003) and original birth record sample (n = 21,615), we have done two sensitive analyses and have come to similar results with the present study. (Table S2 and Table S4).