Associations of Body Mass Index Growth Rates and Body Composition With Cardiometabolic Risks in Chinese Preschool Children

Abstract

Objective

To examine the relationship between body mass index (BMI) growth rates, body composition, and cardiometabolic markers in preschool children.

Methods

Three-year-old children were recruited for this cohort study. BMI and body composition measurements were obtained at enrollment, with multiple BMI measurements spanning ages 1 month to 3 years extracted from medical records. Levels of total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol, non-high-density lipoprotein cholesterol (non-HDL-C), remnant cholesterol (RC), uric acid (UA), and fasting plasma glucose were measured at 3 years. Data analyses employed piecewise linear mixed models and logistic regression models.

Results

Out of 3822 children recruited, 3015 were included in the analysis. The accelerated BMI z-score growth rate between 6 and 24 months was positively correlated with high TG and LDL-C levels, with sex, birthweight, and size for gestational age disparities. Obesity increased the risks of high TG level and the highest RC quartile in boys. Fat mass index and percentage of fat mass were linked with high UA level and dyslipidemia, particularly high TG and non-HDL-C levels, in boys. Fat-free mass index showed negative associations with high levels of TC and non-HDL-C in boys and high LDL-C level in girls (P < .05).

Conclusion

This study underscores the significant impact of BMI growth rates and body composition on cardiometabolic markers in 3-year-old children. The effects of BMI growth rates in specific periods varied by sex, birthweight, and size for gestational age, and boys exhibited a higher susceptibility to adverse outcomes.

Childhood obesity represents a significant public health challenge (1). Since 1980, its prevalence has been increasing, surpassing the rise in adult obesity rates in many countries (1). Globally, the number of children under 5 living with overweight reached 37 million in 2022, marking a concerning upward trend from previous years (2).

Obesity poses grave health implications, impacting metabolic processes and physiology and elevating the risk of various disorders, including metabolic disorders, cardiovascular diseases, depression, and certain cancers (3). Notably, the incidence of pediatric glucose and lipid metabolic disorders has surged worldwide, serving as established risk factors for cardiovascular diseases (4, 5). Children at higher risk are predisposed to experiencing cardiovascular events later in life due to sustained risk factors (6).

Previous studies predominantly relied on body mass index (BMI) as a sole anthropometric measure to predict metabolic disorders related to childhood obesity. However, BMI fails to adequately assess body composition or distinguish between fat and fat-free mass (7), which can significantly vary among individuals within the same BMI category. Childhood growth trajectories serve as crucial predictors of future cardiometabolic risk (8). Nonetheless, it is believed that the effect estimates generated by growth-modeling approaches were not easily interpretable (9). Yet research has primarily focused on the growth's association with obesity, leaving the relationships with cardiometabolic risk factors largely unexplored in preschoolers (10).

Conventional lipid markers have been studied in relation to cardiovascular risk factors. The emerging estimation of remnant cholesterol (RC) in children presents a novel avenue of exploration, which is the cholesterol content of triglyceride-rich lipoproteins, including very low-density lipoproteins, intermediate-density lipoproteins, and chylomicron remnants, offering valuable predictive insights into subsequent cardiovascular diseases without the need for fasting measurements (11, 12).

In this bidirectional cohort study, our aim is to investigate the association between BMI growth rates and body composition with cardiometabolic markers in 3-year-old children, to identify cardiometabolic risk profiles early and develop targeted interventions to mitigate future health risks.

Methods

Study Design and Population

The bidirectional cohort study was conducted in Tianjin municipality, China, from 2017 to 2020 (13-17). Using a stratified cluster sampling method, 11 out of the 16 districts in Tianjin were selected, with 42 kindergartens chosen subsequently. Children in their first year of kindergarten were recruited during 2017-2018 and followed up in their second and third years. Data preceding their enrollment in the cohort were obtained from maternal and child health (MCH) records, containing information on routine healthcare visits from maternal pregnancy until the children's sixth year. Additional data were collected through questionnaires.

Exclusion criteria included children unable to provide informed consent, those with chronic conditions or medication use affecting growth and development, those having acute illnesses hindering participation in physical examinations, those being part of a multiple birth, those lacking body mass and composition measurements in their first year of kindergarten, or those with unavailable or absent collected blood samples (14).

The study adhered to the Declaration of Helsinki and received approval from the Tianjin Women's and Children's Health Center Institutional Review Board (BGI-IRB 17116- 201711). Written informed consent was obtained from the participants’ parents.

Measurements

Anthropometry and body composition

Height measurements of children at age 3 years were obtained using wall-mounted stadiometers without shoes. Body composition analysis was performed using a body composition analyzer (Seehigher BAS-H, Beijing, China) employing bioelectrical impedance analysis (BIA) at frequencies ranging from 1 kHz to 1 MHz, which measured the impedances of the body and each limb while the participants were in a standing position. This device was suitable for children aged 3 years and older (13, 16). Measurements included body weight, fat mass, fat-free mass, muscle mass, and percentage of fat mass (FM%). We also extracted the multiple previous measurements of childrens’ weight and length or height from age 1 month to 3 years from MCH records, which were measured at their routine healthcare visits using standardized procedures (13, 14). The BMI, fat mass index (FMI), and fat-free mass index (FFMI) were computed by dividing the weight, fat mass, or fat-free mass in kilograms by the square of the height/length in meters, respectively. The BMI z score (zBMI) was calculated and age- and sex-standardized according to World Health Organization measures (18-20), and children with zBMI values above +2 and +3 SD were classified as overweight and obese, respectively, based on the World Health Organization reference (19-21). The criterion for defining severe obesity is 1.2 times the criterion for obesity (19-22).

Cardiometabolic markers

At age 3, children underwent fasting venous blood sample collection. Serum total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C), uric acid (UA), and fasting plasma glucose (FPG) were analyzed following the protocol. Abnormal cardiometabolic markers were defined as follows: according to pediatric-specific cutoff values, high TC level was defined as a TC concentration of ≥5.17 mmol/L, high LDL-C level as an LDL-C concentration of ≥3.36 mmol/L, low HDL-C level as an HDL-C concentration of <1.03 mmol/L, high TG level as a TG concentration of ≥1.12 mmol/L, high non-HDL-C level as a non-HDL-C concentration of ≥3.74 mmol/L, and dyslipidemia as the presence of 1 or more abnormal lipid markers (23). For FPG and UA abnormalities, we used reference levels that are commonly applied to both children and adults. Impaired fasting glucose (IFG) was defined as an FPG level of ≥5.6 mmol/L (24, 25) and high UA level as a UA concentration of >420 mmol/L in boys and >360 mmol/L in girls (26, 27). Non-HDL-C was calculated as TC minus HDL-C (28). RC was calculated as TC minus HDL-C minus LDL-C (12) and divided into 4 groups by quartiles.

Covariates

Potential confounders including maternal age, maternal education, maternal height, prepregnancy weight, weight before delivery, gravidity, parity, delivery mode, gestational age, child sex, birth weight, and breastfeeding duration were extracted from MCH records. Variables of maternal history of gestational diabetes mellitus, paternal education, annual family income, family history of metabolic abnormalities, paternal height and weight, children's snacks consumption, physical activity, sleep duration, and sleep quality over the past 3 months were assessed via a parent-reported questionnaire when the children were in their first year of kindergarten. The family history of metabolic abnormalities was assessed by asking whether the child's grandparents and parents had diabetes, hypertension, hyperlipidemia, or coronary heart disease and was defined as positive if any of these health issues were reported. Prepregnancy BMI was calculated as prepregnancy weight in kilograms divided by square of height in meters. Gestational weight gain was calculated as weight before delivery minus prepregnancy weight (14).

Statistical Analyses

Mean (SD) or median (interquartile range) are used to present continuous variables, and frequencies (%) show categorical variables. Group differences were assessed using ANOVA for means, the Kruskal–Wallis test for medians, and the χ² test for frequencies.

Piecewise linear mixed models were employed to investigate the associations of zBMI growth trajectories with cardiometabolic markers. The zBMI growth trajectories in children from 0 to 3 years of age were modeled with multiple linear splines, which were a series of straight lines that connect to each other at knot points, and the linear associations were evaluated within different periods rather than an entire trajectory (9, 29). The knot points in this study were 6 and 24 months of age, which were determined by evaluating the approximate age in months when the growth's slope or direction changed or approached the child's routine healthcare appointments (13, 14). Following the univariate analyses, the multivariate models were performed with adjusting covariates, including maternal age at delivery (<35, ≥35 years), maternal education (high school or less, college, above college), annual family income (<10,000, 10,000-20,000, ≥20 000 CNY), prepregnancy BMI (<18.5, 18.5-23, 23-27.5, ≥27.5 kg/m²) (30), gestational weight gain (inadequate, appropriate, excessive) (31), gravidity (1, 2 or more), maternal history of gestational diabetes mellitus (without, with), gestational age (preterm, full-term), delivery mode (vaginal delivery, cesarean delivery), paternal education (high school or less, college, above college), paternal BMI (<23, 23-27.5, ≥27.5 kg/m²) (30), family history of metabolic abnormalities (without, with), child sex (male, female), size for gestational age [appropriate (AGA), small (SGA), and large (LGA) for gestational age] (32), breastfeeding duration (<6, ≥6 months), snacks consumption frequency (<1/month, 1-3/month, 1-3/week, 4-6/week, ≥1/day), physical activity duration (<1 hour, 1-2 hours, 2-3 hours, 3-4 hours, >4 hours per day), sleep duration (<10 hours, ≥ 10 hours), and sleep quality (good, medium, poor). Subgroup analyses were performed to examine the coherence of associations by child sex, birthweight, and size for gestational age. Missing values of a covariate as a category with multivariate models. Logistic regression models were used to assess the associations of body composition and obesity with cardiometabolic markers, adjusting for the same covariates.

The 2-tailed P-value was found to be statistically significant at a level of less than .05. The statistical analyses were conducted using the software programs Stata 15.0, SAS 9.4 (SAS Institute, Cary, NC), and R 4.0 (R Foundation for Statistical Computing, Vienna, Austria).

Results

A total of 3822 children were initially enrolled in their first year of kindergarten as a baseline population with completed physical examinations that included body composition test, with 3015 included in the analysis after excluding 74 twins or other multiple births, 178 lacking fetal data, and 555 without blood samples (Fig. 1). The characteristics of participants included and excluded in this study are shown in Supplementary Table S1 (33). In this study, 1564 (51.9%) children were boys, and 1451 (48.1%) were girls. Table 1 shows the characteristics of the samples according to child sex. Compared to girls, boys were more likely to have a shorter gestational age (38.9 vs 39.2 weeks, P < .001), a lower proportion with family history of metabolic abnormalities (57.7% vs 62.2%, P = .013), and a higher proportion of birthweight ≥4000 g (10.6% vs 6.8%, P < .001). In boys, for cardiometabolic markers, the proportions of dyslipidemia (33.7% vs 38.3%, P = .010), high TG level (17.7% vs 21.9%, P = .003), low HDL-C level (5.1% vs 8.3%, P < .001), and high UA level (0.8% vs 2.5%, P < .001) were lower, while the proportion of IFG was higher (7.9% vs 4.8%, P = .001).

Table 2 shows the associations of zBMI growth rates with cardiovascular risk factors. Consistent with the results of the univariate analyses, in the multivariate models, the accelerated zBMI growth rate between 6 and 24 months was positively associated with high TG level [β = .007; 95% confidence interval (CI) 0.002, 0.012 SD units/month] and high LDL-C level (β = .008; 95% CI 0.001, 0.015 SD units/month). Stratified analyses demonstrated significant associations between 6∼24 months’ rapid growth and high TG level (β = .008; 95% CI 0.0002, 0.016 SD units/month) and high LDL-C level (β = .017; 95% CI 0.006, 0.027 SD units/month), and between 2∼3 years of rapid growth and high UA level (β = .064; 95% CI 0.004, 0.124 SD units/month) in boys but not in girls (Fig. 2). Null associations were found for the other outcomes [Supplementary Fig. S1 (33)]. Decelerated growth of zBMI before 6 months was linked to low HDL-C level (β = −.433; 95% CI −0.833, −0.033 SD units/month) and the highest RC quartile (β = −.192; 95% CI −0.361, −0.023 SD units/month) in children with low birthweight, and increased the risks of high TC level (β = −.120; 95% CI −0.225, −0.015 SD units/month), high LDL level (β = −.211; 95% CI −0.326, −0.096 SD units/month), and high non-HDL-C level (β = −.135; 95% CI −0.235, −0.034 SD units/month) in SAG children. The study found that an accelerated zBMI growth rate between 6 and 24 months increased the risks of high TG level (β = .007; 95% CI 0.002, 0.013 SD units/month) in children who were born with a normal birthweight and was linked to high TG level (β = .008; 95% CI 0.002, 0.014 SD units/month) and high LDL level (β = .009; 95% CI 0.001, 0.017 SD units/month) in AGA children. Rapid growth at 2∼3 years of age was associated with IFG in low-birthweight children (β = .086; 95% CI 0.029, 0.144 SD units/month), dyslipidemia (β = .007; 95% CI 0.0004, 0.014 SD units/month) in AGA children, and the highest RC quartile in high birthweight [β = .028; 95% CI 0.009, 0.047 SD units/month, Supplementary Fig. S2 (33)] and LGA children [β = .020; 95% CI 0.006, 0.034 SD units/month, Supplementary Fig. S3 (33)].

After controlling covariates, child obesity at 3 years increased the risks of high TG level [odds ratio (OR) = 3.16; 95% CI 1.78, 5.61] and the highest RC quartile (OR = 1.79; 95% CI 1.03, 3.12) in boys but not in girls (Table 3). Child severe obesity significantly increases the risk of low HDL-C level (OR = 4.57; 95% CI 1.14, 18.28), particularly among boys [OR = 8.84; 95% CI 1.89, 41.44; Supplementary Table S2 (33)]. Table 4 displays the associations of anthropometric and body composition measurements with cardiometabolic markers at 3 years and sex differences. After controlling for variables, FMI was positively correlated with dyslipidemia (OR = 1.13; 95% CI 1.02, 1.25), high TG level (OR = 1.25; 95% CI 1.11, 1.41), high non-HDL-C level (OR = 1.15; 95% CI 1.01, 1.30), and high UA level (OR = 1.89; 95% CI 1.11, 3.22) in boys. FFMI was negatively associated with high TC level (OR = 0.75; 95% CI 0.61, 0.93) and high non-HDL-C level (OR = 0.76; 95% CI 0.61, 0.94) in boys and high LDL-C level in girls (OR = 0.70; 95% CI 0.56, 0.89). FM% was associated with dyslipidemia (OR = 1.03; 95% CI 1.01, 1.05), high TG level (OR = 1.04; 95% CI 1.02, 1.07), high LDL level (OR = 1.04; 95% CI 1.00, 1.07), high non-HDL-C level (OR = 1.04; 95% CI 1.01, 1.07), and high UA level (OR = 1.14; 95% CI 1.02, 1.28) in boys, and zBMI increased the risk of high TG level in boys (OR = 1.28; 95% CI 1.12, 1.46).

Discussion

In this large contemporary cohort study, we examined the association of zBMI growth rates from birth to 3 years and body composition with cardiometabolic health in 3-year-old children. We found that cardiometabolic markers were influenced by growth rates and body composition, particularly in boys. Moreover, the impact of zBMI growth rates on cardiovascular and metabolic health varied among children born with different weights and sizes for gestational age and during specific time spans.

Childhood obesity is a critical public health issue, particularly among very young children, due to its high prevalence and association with increased cardiometabolic risk factors such as dyslipidemia and insulin resistance (2, 3, 34). Early development of obesity predicts future obesity; a 2-year-old child with obesity is more likely to be obese at age 35 than an overweight 19-year-old (35). The high and accelerating growth from birth to age 3.5 years was associated with increased body size at age 9 years (36), and the probability of being overweight or obese in adolescence was almost 90% among children who were obese at 3 years of age (37). Early-onset obesity was also associated with future cardiometabolic abnormalities; cumulative exposure to high BMI from 2 to 3 years of age carried the greatest cardiometabolic risk in adolescence (6, 38).

We observed lower proportions of dyslipidemia in boys, contrary to some previous studies (39). However, obese boys were found to have higher risks of high TG levels and the highest RC quartile compared with those of normal weight. Elevated TG is a typical feature of obesity-related dyslipidemia (40). Moreover, our study specifically focuses on RC, which was reported to have a positive relationship with obesity in children (41). Higher RC levels increase the risk of myocardial infarction, given that remnant cholesterol accumulates in the arterial wall, leading to atherosclerosis, a hallmark of atherosclerotic lesions (12). A previous study found male adolescents with the highest quartile of RC might have a higher future risk of cardiovascular disease (42). Our results demonstrated that girls’ obesity did not increase cardiometabolic risks, which underscored the clear sex disparity in very young children's cardiometabolic health. Given that sex differences in cardiovascular health start at birth, the variations in the relationships between boys and girls may be hereditary (7).

Our study highlights the importance of specific body composition measures such as FMI and FM% in increasing the risks of high UA level and dyslipidemia, particularly high TG and non-HDL-C levels, in boys. The sex differences were also found in other studies; for example, a study reported smaller impacts of change in body fatness on TC and LDL-C in girls than in boys (43). Conversely, FFMI exhibited protective effects against high TC and high non-HDL-C levels in boys and high LDL-C level in girls. These results suggest that less fat mass and more fat-free mass are favorable for cardiometabolic health in children. Additionally, zBMI was solely associated with high TG level in boys, indicating that body composition may be a stronger predictor of negative cardiometabolic profiles in 3-year-old children than weight status alone, which is consistent with previous research (44). In future studies, the mechanisms underlying sex differences in the associations between obesity and measures of cardiovascular health should be studied to provide sex-specific prevention opportunities.

The findings of our study reveal associations between time-specific zBMI growth rates and cardiovascular-metabolic health outcomes. Previous studies assessing early-life growth often involve limited time points; considering the dynamic nature of child growth, there is a pressing need for comprehensive and repeated investigations to discern their association with future cardiometabolic outcomes (45). Children who undergo rapid increases in BMI during early childhood were found to be more likely to experience abnormal changes in their cardiovascular and metabolic health (45). Notably, we identified a critical time window of accelerated zBMI growth rates between 6 and 24 months predisposing to high TG and LDL-C levels, with implications particularly prominent among boys. These results reinforce the notion that the first 1000 days of life are crucial for determining future health outcomes, including cardiovascular health (46, 47). Our study further elucidated the relationship between growth rates and cardiovascular-metabolic health outcomes based on birthweight and size for gestational age. Decelerated zBMI growth before 6 months was associated with low HDL-C level and the highest RC quartile in children with low birthweight and with high TC levels, high LDL levels, and high non-HDL-C levels in SGA children, indicating their vulnerability to metabolic abnormalities with poor growth. Conversely, accelerated growth between 6 and 24 months was linked to high TG level in children born with normal weight and to high TG level and high LDL level in AGA children, while rapid growth between 2 and 3 years was associated with IFG in low-birthweight children, dyslipidemia in AGA children, and the highest RC quartile in children born with high birthweight and LGA. These observations align with the capacity-load model (48), suggesting that the sluggish growth in low-birthweight and SGA children might limit their metabolic capacity, while the rapid weight gain in children with different birthweight or size for gestational age might elevate metabolic load. While it is well documented that children born SGA who experience catch-up growth are at increased risk for metabolic disorders (49), our study also highlights significant risks of accelerated BMI growth for children with different birthweights or born AGA/LGA. Studies have shown that although the exact mechanisms may differ, accelerated postnatal growth, regardless of initial birthweight or size for gestational age, can lead to adverse metabolic outcomes such as obesity, insulin resistance, and cardiovascular diseases (50-56). This suggests that rapid growth modified the metabolic load. Both reduced metabolic capability and increased metabolic load pose challenges to metabolic homeostasis (45). Children with varying birthweight/size for gestational age had distinct appropriate growth patterns in different sensitive periods, which underscored the importance of considering birthweight/size for gestational age when addressing unfavorable growth patterns in specific time periods.

Strengths of our study include the comprehensive exploration of piecewise growth trajectories in the first 3 years and their association with childhood cardiometabolic health, supported by repeated BMI assessments, benefitting the identification of critical windows in which later obesity and cardiometabolic disease risk are being programmed. Additionally, the assessment of multiple anthropometric measures and cardiometabolic markers facilitated a detailed interpretation of the impacts of growth and body composition on health outcomes. Moreover, by adjusting for key covariates such as parental BMI, maternal history of gestational diabetes mellitus, and family history of metabolic abnormalities, we accounted for potential confounding genetic factors, enhancing the robustness of our findings. Given the rising prevalence of overweight and obesity among preschool children (2) and the limited examination of cardiometabolic indicators in children in kindergarten, our focus on 3-year-olds and explorations of potential variations allowed for early detection and targeted interventions to control early-onset abnormalities among very young individuals. Moreover, the presence of sex disparities in cardiovascular health metrics underscored the need for cohort studies to elucidate sex-specific cardiovascular risks.

However, our study also had some limitations that warrant consideration. First, the use of BIA instead of the gold-standard dual-energy X-ray absorptiometry for assessing body composition may have introduced measurement error. While the BIA approach demonstrated a strong correlation with the body composition evaluation achieved using dual-energy X-ray absorptiometry (57), the potential limitations of this method should be acknowledged. Ideally, children with a familial predisposition for hypercholesterolemia should be excluded; however, this variable was not collected in our study. To minimize potential bias, we adjusted for variables such as family history of metabolic abnormalities in the multivariate analyses. Also, reports have linked gestational glucose intolerance, which includes levels below the gestational diabetes threshold, to a higher risk of having overweight or obese children (58). However, we did not accurately define the maternal glycemic state during pregnancy; besides, we relied on the mother's self-reported weight before pregnancy and did not have complete data on relevant factors such as the mother's smoking history or the timing of introducing complementary foods, which may have caused bias in our findings. Severe obesity significantly affects metabolic health later in life (59). In this study, due to the small number of children with severe obesity, some estimates for outcomes could not be accurately calculated because of the insufficient sample size; therefore, future research with larger sample sizes is needed to comprehensively explore the impact of severe obesity on cardiometabolic risks. The validity and reliability of the questionnaire used in this study need to be further evaluated. Furthermore, the lack of a diagnostic cutoff for RC in children necessitated the use of the upper quartile as a proxy. The small sample size in specific strata in our stratified analyses may lead to reduced statistical power and broad CIs, requiring careful interpretation of the point estimates. Some differences existed in characteristics between the included and excluded participants; although we have adjusted for relevant covariates, residual bias cannot be completely ruled out. The study was conducted on an Asian population, which may limit the generalizability of the findings to other ethnic groups. The impact of lifestyle factors in early life and their interaction with parental health characteristics, prenatal and gestational factors, and child-related variables on cardiometabolic risks in children warrants comprehensive exploration in future research.

Conclusions

In conclusion, our cohort study provides valuable insights into the impact of BMI growth and body composition on markers of cardiometabolic disease in very young children. We observed sex- and birthweight-dependent variations in the impacts of growth rates, highlighting the importance of tailored intervention strategies. Addressing childhood obesity and promoting healthy body composition, particularly in boys who are susceptible, is essential for preventing cardiovascular metabolic abnormalities and promoting long-term health outcomes. Future research should aim to elucidate the underlying mechanisms driving these associations.

Acknowledgments

The authors would like to give special thanks to all the staff and participants of the study.

Funding

This research was supported by The National Key Research and Development Program of China (2016YFC1300100, 2016YFC0900600), National Natural Science Foundation of China (72104148), Beijing Hospitals Authority Youth Programme (QML20211301), Beijing Municipal Administration of Hospitals Incubating Program (Px2022052), Research Foundation of Capital Institute of Pediatrics (ERB-2023-02), and Public Service Development and Reform Pilot Project of Beijing Medical Research Institute (BMR2021-3).

Author Contributions

J.W. and G.L. designed the larger study. Z.L. and F.C. designed the present study. J.W., W.L., and G.L. conducted the investigation. Z.L. conducted the analysis and drafted the manuscript. F.C., J.W., Y.C., W.L., T.Z., G.L., and X.X. critically revised and reviewed the manuscript. F.C. and G.L. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

Disclosures

No potential conflicts of interest relevant to this article were reported.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

References

Author notes