Percent body fat, but not body mass index, is associated with cardiometabolic risk factors in children and adolescents

Abstract

Background

The epidemic of overweight and obesity has become a worldwide public health problem. Cardiometabolic diseases may originate in childhood. We investigated the association between percent body fat (PBF) measured by the bioelectrical impedance assay and cardiometabolic risk (CMR) in pediatrics.

Methods

This cross‐sectional study involved 3819 subjects (6–17 years old) in Shanghai. We assessed the association between PBF and body mass index (BMI) with multiple CMR factors. We examined the risk for cardiometabolic abnormalities attributable to overweight and obesity based on age‐ and sex‐specific PBF Z‐scores and BMI Z‐scores, respectively.

Results

PBF, but not BMI, was positively associated with multiple CMR factors in males and females except for total cholesterol in females (all p < 0.05). Compared with the non‐overweight group based on PBF, overweight and obese subjects had increasingly higher odds ratio of dyslipidemia (2.90 (1.99–4.23), 4.59 (2.88–7.32) for males and 1.82 (1.20–2.75), 2.46 (1.47–4.11) for females) and elevated blood pressure (BP) (3.26 (2.35–4.51), 4.55 (2.92–7.09) for males and 1.59 (1.07–2.34), 3.98 (2.27–6.17) for females). Obesity females showed a higher likelihood for hyperglycemia (2.19 (1.24–3.84)) than non‐overweight females. In both sexes, the predictive effect of PBF on dyslipidemia and elevated BP in adolescents was better than that in children. For hyperglycemia, the predictive effect of PBF was better in male adolescents and female children. There was no risk difference for cardiometabolic abnormalities attributable to BMI‐based obesity categories.

Conclusions

PBF but not BMI was associated with CMR. Overweight and obesity categories based on PBF had an increased risk for cardiometabolic abnormalities in children and adolescents.

Highlights

• The incidence of dyslipidemia, hyperglycemia, and elevated blood pressure (BP) was 12.1%, 11.8%, and 18.8%, respectively, in children and adolescents in Shanghai, China.

• Percent body fat (PBF) but not BMI was associated with CMR; meanwhile, overweight and obesity categories based on PBF had an increased risk for cardiometabolic abnormalities in the pediatric population.

• Although PBF could better predict dyslipidemia and elevated BP in adolescents than in children of both sexes, it could better predict hyperglycemia in female children.

1. INTRODUCTION

Cardiometabolic diseases, such as Type 2 diabetes and cardiovascular diseases, are the main causes of mortality, disability, and significant economic burden in modern societies. ¹ Although the clinical features of cardiometabolic diseases usually manifest in adulthood, its well‐recognized risk factors, including dyslipidemia, elevated blood pressure (BP), hyperglycemia, and obesity, may even originate in childhood and track into adulthood. ² , ³ , ⁴ With economic development and lifestyle changes in recent decades, the population with cardiometabolic risk (CMR) has become more common and younger in China. It was estimated that the overall detection rate of obesity, dyslipidemia, hyperglycemia, and elevated BP even reached 12.0%, 15.8%, 24.4%, and 18.2% in Chinese children and adolescents, respectively. ⁵ In view of the trajectory effect and raising pandemic, it is imperative to raise awareness of CMR factors in the pediatric population, which may help to maintain cardiovascular health in youth and prevent cardiovascular risk in adulthood.

The body mass index (BMI) is a convenient and simple index for monitoring obesity. However, due to rapid growth, weight and height are not as proportional in children and adolescents as in adults. ⁶ , ⁷ In addition, it may not accurately distinguish between fat and lean mass and predict other CMRs in the pediatric population. ⁸ , ⁹ , ¹⁰ , ¹¹ High body fat percentage, representing adiposity‐based overweight and obesity, has been shown to be related to other cardiovascular risk factors in children. ¹² The China Child and Adolescent Cardiovascular Health (CCACH) study showed that body fat measured by dual‐energy X‐ray absorptiometry (DEXA) is superior to BMI in identifying abnormal lipid profiles in boys. ¹³ In addition, trunk fat mass, especially abdominal visceral and subcutaneous adipose tissues, was independently associated with higher clustered CMR in youth. ¹⁴ , ¹⁵ Meanwhile, the researchers also found that the bioelectrical impedance assay (BIA), which is a portable, noninvasive technique to measure body fat and suitable for large‐scale screening, agrees well with body fat measured by DXEA. ¹⁶ Up to now, studies on the relationship between BIA‐based percent body fat (PBF) and other CMR of children and adolescents in megacities in China are still quite limited.

In this study, using cross‐sectional data from the Diet and Healthy Survey in Shanghai (the mega city in China), ¹⁷ , ¹⁸ we aimed to examine the correlation between PBF (BIA‐based) and BMI in children and adolescents with CMR and analyze the risk differences for cardiometabolic abnormalities attributable to obesity categories based on PBF and BMI.

2. METHODS

2.1. Study population

From September through October 2015, we conducted a cross‐sectional study, which was designed to examine nutrition intake and health outcomes among students between the ages of 6 and 17 years. Using the multistage stratified sampling method, 60 schools (20 primary schools, 20 junior middle schools, and 20 senior high schools) were selected from 16 districts in Shanghai, China, and representative samples of school children aged 6–17 years were randomly selected according to grade and sex distribution (n = 4320). Subjects with a history of chronic diseases, such as thyroid‐related disease, diabetes, anemia, and malnutrition, had been excluded at the initial inclusion stage. We excluded individuals with incomplete questionnaires (n = 237), missing anthropometric measurements (n = 78), refusing to provide blood (n = 47), and incomplete test data (n = 115). Individuals with hypersensitive C‐reactive protein > 10 mg/L (n = 24) were also excluded to avoid possible acute inflammation. Finally, 3819 participants were included in the present analysis (Figure 1). Our study met the ethical guidelines of the 1975 Declaration of Helsinki and received approval from the Ethics Review Committee of the Shanghai Municipal Center for Disease Control and Prevention (No. 2015‐15). We obtained written informed consent from each participant and the participant's guardians.

Figure 1.

Flowchart of the study cohort. hs‐CRP, hypersensitive C‐reactive protein.

2.2. Demographics and anthropometric measurements

Using an interviewer‐administered questionnaire, information on age, sex, smoking, alcohol, sedentary time, and intentional physical exercise was obtained. Sedentary time was the sum of the time spent using a computer, reading, doing homework, and performing other sedentary activities. Intentional physical exercise was defined as physical exercise for the purpose of health maintenance or fitness. A validated food‐frequency questionnaire was used to collect dietary intake by information on consumption frequency and quantity of typical food items. A detailed description can be viewed in our previous papers. ¹⁷ , ¹⁸ The dietary energy intake was estimated using the Chinese food composition database. ¹⁹

Professional doctors conducted anthropometric measurements following standard protocols in community health centers. Height (SG‐210) and weight (TANITA‐BC601) measurements were performed with the participants wearing light clothing and no shoes. BMI was calculated as weight (kilogram) divided by the square of height (m²). PBF was automatically scanned and calculated by the BIA (TANITA BC601). Waist circumference was measured midway between the costal margin and iliac crest at the end of a normal expiration (Graham‐Field 1340‐2 tape). Waist‐to‐height ratio (WHtR) was calculated as waist circumference (cm) divided by height (cm). Blood pressure was measured three times after a quiet rest using an Omron HEM‐7071 electronic sphygmomanometer (Omron Healthcare) and averaged.

2.3. Laboratory biochemical testing

Five milliliters of venous blood from 7:00 a.m. to 8:00 a.m. was collected following overnight fasting. Using an automatic biochemical analyzer (HITACHI 7080) and Wako reagent, we measured serum concentrations of fasting blood glucose (FBG), triglycerides (TG), total cholesterol (TC), low‐density lipoprotein‐cholesterol (LDL‐C), and high‐density lipoprotein‐cholesterol (HDL‐C). The intra‐assay coefficient of variations (CVs) for TC, TG, LDL‐C, HDL‐C, and FBG were 1.19%, 1.45%, 2.39%, 3.08%, and 2.46%, respectively, and the interassay CVs were 2.91%, 2.83%, 3.12%, 4.03%, and 3.64%, respectively. The laboratory of the Shanghai Center for Disease Control and Prevention, which passed the certification from the Shanghai Clinical Testing Center, performed all the biochemical testing.

2.4. Potential confounders

To control the effects of confounders, several factors that may be associated with CMRs were recorded as covariates, including age, sex, WHtR, dietary energy intake, intentional physical activities, and sedentary time. ¹³ , ¹⁶ Due to there being few smokers and drinkers in the school‐age population investigated, the two items were not included in the covariate analysis.

2.5. Definition of CMR factors

CMR factors included dyslipidemia, hyperglycemia, elevated BP, and obesity. Dyslipidemia in children and adolescents was defined as having any of the following conditions (Table 1): (1) high TC (≥200 mg/dl), (2) high TG (≥110 mg/dl, <10 years old; ≥130 mg/dl, ≥10 years old), (3) high LDL‐C (≥130 mg/dl), and (4) low HDL‐C (<40 mg/dl). ²¹ , ²²

Table 1.

Definitions of CMR

CMR	Definition	References
Dyslipidemia	Any of the following four conditions:	[15, 16]
High TC	TC ≥ 200 mg/dl
High TG	TG ≥ 110 mg/dl (<10 years old) or ≥130 mg/dl (≥10 years old)
High LDL‐C	LDL‐C ≥ 130 mg/dl
Low HDL‐C	HDL‐C < 40 mg/dl
Hyperglycemia	FBG ≥ 100 mg/dl	[17]
Elevated BP	SBP/DBP over 95th percentiles for the age‐ and sex‐specific reference	[18]
Obesity	BMI Z‐scores/PBF Z‐scores: over 95th percentiles	[20]

Abbreviations: BMI, body mass index; BP, blood pressure; CMR, cardiometabolic risk; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL‐C, high‐density lipoprotein; LDL‐C, low‐density lipoprotein; PBF, percent body fat; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride; WHtR, waist‐to‐hip ratio.

Hyperglycemia was defined as having an FBG ≥ 100 mg/dl. ²³ Elevated BP was defined as systolic blood pressure (SBP) and/or diastolic blood pressure (DBP) over age‐specific, sex‐specific, and height‐specific 95th percentiles (P95th) of the Chinese references 2017. ²⁴ Obesity was defined as over P95th according to age‐ and sex‐specific BMI Z‐scores and PBF Z‐scores, respectively. ²⁰

2.6. Statistical analyses

We analyzed demographic, anthropometric characteristics, and laboratory indexes based on sex grouping. Continuous variables with normal or skewed distribution are expressed as mean ± standard deviation (SD) or median (IQR), respectively. The categorical variables are presented as absolute numbers (n) and percentages (%). We analyzed intergroup comparisons using Student's t‐test for normally distributed data, Mann–Whitney U‐test for skewed data, and χ ²‐test for categorical variables. The sex‐ and age‐standardized BMI Z‐scores and PBF Z‐scores were calculated using the standard deviation dispersion method (Z‐scores) to eliminate variability with growth and development; the Z‐scores = (measured value − mean value of the age group)/standard deviation. ²⁵ The differences in Pearson's correlation between PBF Z‐scores and BMI Z‐scores with CMR factors were compared. Using multiple linear regression analysis, associations between BMI Z‐scores and PBF Z‐scores with cardiometabolic parameters were compared. Subsequently, subjects were divided into non‐overweight (<P85th), overweight (P85th to P95th), and obese (>P95th) according to BMI Z‐scores and PBF Z‐scores. ²⁰ Mixed‐effect logistic regression was applied to compare risk differences for dyslipidemia, hyperglycemia, and elevated BP attributable to obesity categories based on PBF and BMI. We did further sensitivity analyses. We divided PBF and BMI into low, middle, and high groups based on −1 and 1 (Z‐scores) as cut points and repeated logistic regression to evaluate the risk stability of BMI and PBF categories for cardiometabolic abnormalities. Furthermore, to better evaluate the predictive significance of PBF in cardiometabolic abnormalities, we used the area under the curve (AUC) and 95% confidence interval (CI) of the receiver‐operating characteristic (ROC) curve to measure the accuracy of PBF in distinguishing CMR in children and adolescents with different sex and age groups. For data analysis, we used the Statistical Package of Social Sciences for Windows (SPSS, version 22.0; IBM Corp) and R statistical software (R Institute, version 3.4.1; SAS Institute Inc.). Two‐sided p < 0.05 was statistically significant.

3. RESULTS

3.1. General characteristics

There were 1897 males (767 children and 1130 adolescents) and 1922 females (742 children and 1180 adolescents) with a mean age of 11.54 ± 3.41 years. PBF was higher in females than in males (23.11 ± 8.20% vs. 18.41 ± 9.66%; p < 0.001). However, there were no significant differences in WHtR and BMI between males and females. Males had lower lipid concentrations (all p < 0.001) but higher FBG and SBP (all p < 0.001). Approximately, 12.1% of the participants had dyslipidemia (5.2% high TC, 6.3% high TG, 3.6% high LDL‐C, and 1.9% low HDL‐C). The overall incidence of hyperglycemia and elevated BP among them was 11.8% and 18.8%, respectively. The prevalence of low HDL‐C, hyperglycemia, and elevated BP (especially elevated SBP) was higher in males than in females (p < 0.001) (Table 2).

Table 2.

Characteristics of the study cohort, stratified by sex

Characteristics	Total (n = 3819)	Males (n = 1897)	Females (n = 1922)	p
Age (years)	11.54 ± 3.41	11.55 ± 3.40	11.54 ± 3.42	0.93
Children, 6–10 years	1509	767 (50.8)	742 (49.2)
Adolescents, 11–17 years	2310	1130 (48.9)	1180 (51.1)
BMI (kg/m2)	19.50 ± 4.50	19.54 ± 4.19	19.45 ± 4.78	0.54
WHtR	0.44 ± 0.06	0.44 ± 0.06	0.44 ± 0.06	0.37
PBF (%)	20.77 ± 9.26	18.41 ± 9.66	23.11 ± 8.20	<0.001
Total cholesterol (mg/dl)	152.94 ± 27.04	150.18 ± 27.55	155.67 ± 26.26	<0.001
Triglycerides (mg/dl)	54.47 (39.86, 77.06)	52.26 (38.09, 74.40)	56.69 (45.52, 78.83)	<0.001
High‐density lipoprotein (mg/dl)	60.65 ± 12.28	59.39 ± 12.36	61.89 ± 12.08	<0.001
Low‐density lipoprotein (mg/dl)	84.54 ± 22.44	83.21 ± 23.05	85.85 ± 21.74	<0.001
FBG (mg/dl)	91.90 ± 8.47	92.98 ± 7.65	90.83 ± 9.09	<0.001
SBP (mmHg)	107.81 ± 13.92	111.25 ± 14.54	104.41 ± 12.37	<0.001
DBP (mmHg)	68.53 ± 8.88	68.46 ± 9.56	68.60 ± 8.15	0.62
High TC (≥200 mg/dl)	198 (5.2)	95 (5.0)	103 (5.4)	0.63
High TG (≥100 mg/dl, 0–9 years; ≥130 mg/dl, 10–19 years)	240 (6.3)	129 (6.8)	111 (5.8)	0.19
Low HDLC (≤40mg/dl)	72 (1.9)	48 (2.5)	24 (1.2)	0.004
High LDLC (≥130 mg/dl)	136 (3.6)	73 (3.9)	63 (3.3)	0.34
Dyslipimia, n (%)	462 (12.1)	240 (12.7)	222 (11.6)	0.29
Hyperglycemia (≥100 mg/dl)	448 (11.8)	273 (14.5)	175 (9.1)	<0.001
Elevated SBP	437 (11.4)	312 (16.4)	125 (6.5)	<0.001
Elevated DBP	496 (13.0)	239 (12.6)	257 (13.4)	0.54
Elevated BP	716 (18.8)	419 (22.1)	297 (15.5)	<0.001
Sedentary time (h/day)	3.5 ± 1.9	3.3 ± 1.5	3.7 ± 1.4	0.01
Dietary energy intake (kcal/day)	2086.07 ± 644.07	2043.73 ± 638.88	2127.73 ± 646.60	<0.001
Intentional physical exercise
No	1801 (47.2)	925 (48.8)	876 (45.6)	0.13
Yes	2014 (52.8）	970 (51.2)	1044 (54.4)

Note: Continuous data are presented as mean ± standard deviation or median (IQR); categorical data are presented as n (%). Bold values indicate statistically significant at p < 0.05.

Abbreviations: BMI, body mass index; BP, blood pressure; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL‐C, high‐density lipoprotein; IQR, interquartile range; LDL‐C, low‐density lipoprotein; PBF, percent body fat; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride; WHtR, waist‐to‐hip ratio.

3.2. Association of PBF and BMI with CMR factors by sex groups

Except for serum TC in females, PBF Z‐scores were extremely significantly correlated with serum lipid concentration and BP in males and females (all p < 0.01), with higher correlation coefficients in males than that in females (for TG: 0.37 vs. 0.20; for LDL‐C: 0.25 vs. 0.13; for SBP: 0.26 vs. 0.20; and for DBP: 0.23 vs. 0.19). However, there was no correlation between PBF Z‐scores and BMI Z‐scores. It is worth noting that BMI Z‐scores were not associated with other CMR factors (Supporting Information: Table 1).

Covariates adjusted regression showed that PBF Z‐scores in males were negatively correlated with HDL‐C (β: −2.97; 95% CI: −3.47 to −2.47; p < 0.001), which were positively associated with other serum lipids (β: 4.42, 95% CI: 3.20–5.63; β: 0.08, 95% CI: 0.07–0.09; β: 6.03, 95% CI: 5.01–7.04; all p < 0.001), FBG (β: 0.35, 95% CI: 0.01–0.70; p = 0.049), and BP (SBP, β: 3.91, 95% CI: 3.41–4.40; DBP, β: 2.30, 95% CI: 1.89–2.72; all p < 0.001). In females, PBF Z‐scores were correlated with CMR factors, except TC (all p < 0.05). Regression coefficients were lower in females than in males. There were either no significant associations between BMI Z‐scores and CMR factors in both sexes (Table 3).

Table 3.

Sex‐stratified linear regression analyses of cardiometabolic risk factors with BMI Z‐scores and PBF Z‐scores in children and adolescents

Items	Males				Females
	PBF Z‐scores		BMI Z‐scores		PBF Z‐scores		BMI Z‐scores
	Β (95% CI)	p	Β (95% CI)	p	Β (95% CI)	p	Β (95%CI)	p
TC	4.42 (3.20 to 5.63)	<0.001	2.25 (−0.75 to 3.86)	0.14	1.06 (−0.14 to 2.26)	0.08	−1.15 (−3.15 to 0.86)	0.26
TG (log‐transformed)	0.08 (0.07 to 0.09)	<0.001	−0.37 (−2.44 to 1.69)	0.72	0.04 (0.03 to 0.05)	<0.001	−0.01 (−0.03 to 0.01)	0.07
HDL‐C	−2.97 (−3.47 to −2.47)	<0.001	−0.02 (−0.03 to 0.01)	0.08	−2.73 (−3.27 to −2.19)	<0.001	0.39 (−0.54 to 1.31)	0.41
LDL‐C	6.03 (5.01 to 7.04)	<0.001	0.44 (−0.34 to 1.40)	0.23	3.31 (2.33 to 4.29)	<0.001	−1.61 (−3.28 to 0.05)	0.06
FBG	0.35 (0.00 to 0.70)	0.049	−0.71 (−2.47 to 1.06)	0.43	0.55 (0.13 to 0.97)	0.01	−0.07 (−0.77 to 0.62)	0.84
SBP	3.91 (3.41 to 4.40)	<0.001	−0.21 (−0.80 to 0.38)	0.48	2.53 (2.03 to 3.03)	<0.001	−0.21 (−1.07 to 0.65)	0.63
DBP	2.30 (1.89 to 2.72)	<0.001	0.33 (−0.55 to 1.21)	0.47	1.61 (1.26 to 1.97)	<0.001	−0.11 (−0.72 to 0.49)	0.72

Note: Adjusted covariates: Age, sedentary time, dietary energy intake, and intentional physical exercise. Bold values indicate statistically significant at p < 0.05.

Abbreviations: BMI, body mass index; BP, blood pressure; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL‐C, high‐density lipoprotein; LDL‐C, low‐density lipoprotein; PBF, percent body fat; SBP, systolic blood pressure; TC, total cholesterol; TG, triglyceride; WHtR, waist‐to‐hip ratio.

3.3. Risk of cardiometabolic abnormalities in different obesity categories based on PBF Z‐scores and BMI Z‐scores

In the multivariate logistic regression model adjusted for WHtR, sedentary time, dietary energy intake, and intentional physical exercises, males in the overweight and obesity groups had significantly increased risks for dyslipidemia (odds ratio [OR]: 2.90, 95% CI: 1.99–2.43, p < 0.001 for overweight; OR: 4.59, 95% CI: 2.88–7.32, p < 0.001 for obesity) and elevated BP (OR: 3.26, 95% CI: 2.35–4.51, p < 0.001 for overweight; OR: 4.55, 95% CI: 2.92–7.09, p < 0.001 for obesity) than those in the non‐overweight group (Figure 2A,C). Males in overweight and obesity groups showed no risk of hyperglycemia compared with non‐overweight males (Figure 2B). Compared with non‐overweight females, those in the overweight group had a slightly increased likelihood of having dyslipidemia (OR: 1.82; 95% CI: 1.20–2.75; p = 0.005) and elevated BP (OR: 1.59, 95% CI: 1.07–2.34, p = 0.02) (Figure 3A,C), and those in the obesity group had higher risks for dyslipidemia (OR: 2.46; 95% CI: 1.47–4.11; p = 0.001), hyperglycemia (OR: 2.19; 95% CI: 1.24–3.84; p = 0.007), and elevated BP (OR: 3.98; 95% CI: 2.27–6.17; p < 0.001) (Figure 3A–C). There were no differences in risk for dyslipidemia, hyperglycemia, and elevated BP by BMI‐based obesity categories (Table 4).

Figure 2.

Risk of cardiometabolism abnormalities attributable to obesity categories based on PBF in males. (A) Dyslipidemia; (B) hyperglycemia; (C) elevated BP; non‐overweight, age‐ and sex‐ specific PBF Z‐scores < P85th; overweight, P85th ≤ PBF Z‐scores ≤ P95th; obesity, PBF Z‐scores > P95th. BP, blood pressure; CI, confidence interval; P, percentile; PBF, percent body fat.

Figure 3.

Risk of cardiometabolic abnormalities attributable to obesity categories based on PBF in females. (A) dyslipidemia; (B) hyperglycemia; (C) elevated BP; non‐overweight, age‐ and sex‐ specific PBF Z‐scores < P85th; overweight, P85th ≤ PBF Z‐scores ≤ P95th; obesity, PBF Z‐scores > P95th. BP, blood pressure; CI, confidence interval; P, percentile; PBF, percent body fat.

Table 4.

Risk for cardiometabolic abnormalities attributed to BMI‐based obesity categories by sex groups

Cardiometabolic abnormalities	BMI‐based obesity category	Males			Females
		Prevalence (%)	Odds ratios	p	Prevalence (%)	Odds ratios	p
Dyslipidemia	Non‐overweight (Ref.)	201/1611 (12.5)	1.00		187/1634 (11.4)	1.00
	Overweight	29/190 (15.3)	1.30 (0.78–2.16)	0.31	22/192 (11.5)	1.14 (0.67‐1.93)	0.64
	Obesity	10/94 (10.6)	0.92 (0.40–2.09)	0.83	13/96 (13.5)	1.45 (0.70‐2.99)	0.32
Hyperglycemia	Non‐overweight (Ref.)	232/1611 (14.4)	1.00		153/1634 (9.4)	1.00
	Overweight	28/190 (14.7)	0.99 (0.60–1.64)	0.97	13/192 (6.8)	0.86 (0.45‐1.65)	0.64
	Obesity	15/94 (16.0)	1.09 (0.53–2.27)	0.81	9/96 (9.4)	1.37 (0.59‐3.18)	0.46
Elevated BP	Non‐overweight (Ref.)	357/1612 (22.1)	1.00		256/1634 (15.7)	1.00
	Overweight	48/190 (25.3)	1.15 (0.75–1.74)	0.53	26/192 (13.5)	0.76 (0.47‐1.23)	0.26
	Obesity	15/94 (16.0)	0.66 (0.33–1.33)	0.25	15/96 (15.6)	0.78 (0.40‐1.54)	0.47

Note: Non‐overweight, age‐ and sex‐specific BMI Z‐scores < P85th; overweight, P85th ≤ BMI Z‐scores ≤ P95th; obesity, BMI Z‐scores > P95th.

Abbreviations: BMI, body mass index; BP, blood pressure; Ref. reference.

In sensitivity analyses, risk trends seen across new PBF categories were similar to previous results in both sexes. We used −1 and 1 (Z‐scores) as the cut‐off points, which are roughly in the 10th and 85th percentiles of our study cohort, respectively. Low, middle, and high PBF groups are roughly equivalent to the low fat, healthy fat, and overweight/obesity groups, respectively. Compared with the low PBF group, the high PBF group had a significantly higher risk of dyslipidemia and elevated BP (Supporting Information: Figures 1A,C and 2A,C). Interestingly, males with high PBF had a slightly higher risk of hyperglycemia (Supporting Information: Figure 1B) than those with low PBF, which did not occur when compared with non‐overweight males (Figure 2B). There were no differences in risk for dyslipidemia, hyperglycemia, and elevated BP according to the new BMI category method (Supporting Information: Table 2).

3.4. Performance of PBF in predicting cardiometabolic abnormalities among different sex and age groups

Using ROC curve analyses, we assessed the performance of PBF Z‐scores in predicting cardiometabolic abnormalities in subjects with different sex and age groups. PBF could well predict dyslipidemia in four sex and age subgroups with AUC from 0.58 to 0.72. Subsequently, PBF also could predict elevated BP in three subgroups, except for female children. For hyperglycemia, there were some disparities in predictive effects in sex and age subgroups (AUC: 0.57 (0.52–0.61) for male adolescents; AUC: 0.60 (0.53–0.67) for female children) (Table 5).

Table 5.

Performance of PBF in predicting dyslipidemia, hyperglycemia, and elevated BP among sex‐ and age‐stratified pediatric population

Items	Group	Males				Females
		n	Prevalence (n, %)	AUC (95% CI)	p	n	Prevalence (n, %)	AUC (95% CI)	p
Dyslipidemia	Total	1896	240 (12.7)	0.66 (0.62 to 0.70)	<0.001	1921	222 (11.6)	0.58 (0.54 to 0.63)	<0.001
	Children	766	109 (14.2)	0.59 (0.53 to 0.65)	0.003	783	105 (13.4)	0.58 (0.51 to 0.64)	0.01
	Adolescents	1130	131(11.6)	0.72 (0.67 to 0.76)	<0.001	1138	117(10.3)	0.59 (0.54 to 0.65)	0.001
Hyperglycemia	Total	1896	275 (14.5)	0.57 (0.53 to 0.60)	0.001	1921	175 (9.1)	0.52 (0.47 to 0.57)	0.34
	Children	766	92 (12.0)	0.56 (0.50 to 0.63)	0.05	783	64 (8.2)	0.60 (0.53 to 0.67)	0.01
	Adolescents	1130	183 (16.2)	0.57 (0.52 to 0.61)	0.006	1138	111 (9.8)	0.48 (0.42 to 0.54)	0.49
Elevated BP	Total	1897	420 (22.1)	0.66 (0.62 to 0.68)	<0.001	1921	297 (15.5)	0.61 (0.57 to 0.65)	<0.001
	Children	766	125 (16.3)	0.58 (0.52 to 0.63)	0.008	783	137 (17.5)	0.53 (0.48 to 0.59)	0.21
	Adolescents	1131	295 (26.1)	0.69 (0.65 to 0.72)	<0.001	1138	160 (14.1)	0.67 (0.62 to 0.72)	<0.001

Note: Children, 6–10 years; adolescents, 11–17 years. Adjusted covariates: Age, WHtR, sedentary time, dietary energy intake, and intentional physical exercises. Bold values indicate statistically significant at p < 0.05.

Abbreviations: AUC, area under the curve; BP, blood pressure; CI, confidence interval; PBF, percent body fat; WHtR, waist‐to‐height ratio.

4. DISCUSSION

This large‐scale health survey provided an opportunity for us to study the association between PBF and CMR in Chinese children and adolescents. The term “CMR” was officially proposed in a joint scientific statement from the American Heart Association and American Diabetes Association in 2006, aiming to call attention to a set of core common factors that can promote the risk of cardiometabolic diseases, such as overweight or obesity, fasting or postprandial hyperglycemia, elevated SBP and DBP, dyslipidemia, and so on. ²⁶ , ²⁷

Our study showed that BIA‐based PBF but not BMI was significantly positively correlated with CMR factors in children and adolescents. Similar results have been reported in previous studies. In 2019, a cross‐sectional study involving 840 children and adolescents from Brazil showed that PBF assessed by multifrequency BIA was superior to BMI and WHtR in identifying unfavorable lipid profiles. ²⁸ A nationwide CCACH study in China, involving 8944 children and adolescents between the ages of 6 and 18 years, showed that body fat measured by DEXA was more closely related to serum TC, LDL‐C, and TG and performed better than BMI in identifying abnormal lipid profiles in males but not in females. ¹³ Recently, a cross‐sectional study conducted in Beijing showed that higher PBF measured by DEXA and BIA was associated with increased BP levels, and the effects were more pronounced in older children. ¹⁶ Li et al. ²⁹ reported that, based on ROC curve analysis, the P70th cut‐off of excess PBF was better at predicting abnormal BP and blood glucose metabolism. These findings suggested that screening and assessment based on body fat rather than weight may be more accurate in monitoring and managing the risk of chronic diseases in children and adolescents. Oxidative stress and inflammation caused by high fat may contribute to metabolic syndrome, diabetes, and cardiovascular diseases. ³⁰ , ³¹ , ³² Adipose tissues are also an important endocrine organ secreting adiponectin and other adipokines. It plays a role in lipid metabolism, insulin signal transduction, inflammation, and oxidative stress. ³³ , ³⁴ Body fat is a novel parameter of childhood obesity, and its application effect needs to be confirmed by more studies. In particular, the application of BIA body fat measurement technology, due to its convenience and popularization, should be evaluated in children's obesity screening.

Interestingly, we observed that the correlation between PBF and CMR, especially dyslipidemia and elevated BP, was stronger in males than in females. This is supported by previous studies. ¹³ , ²⁸ Dallaire and colleagues ³¹ concluded that there were sex differences in cardiovascular function from early fetal life to late adolescence, but reasons for the multifactorial impact of sex have not been thoroughly understood. Even though the total PBF is higher in females than in males, the proportion of peripheral regional fat is greater in females, ³⁵ which had less significance for cardiovascular metabolic health risk than visceral fat and abdominal subcutaneous fat. ¹⁵ , ³⁶ , ³⁷ Our study did not differentiate the distribution of fat, which deserves further investigation. In addition, estrogen has protective effects on CMR in females, which is consistent with the notion that up to menopause, the atherosclerotic risk is lower in women than in men. ³⁸ Estrogen may mediate vasodilation through endothelial‐dependent effects via the estrogen receptor and G protein‐coupled estrogen receptor, ³⁹  and also reduces BP by affecting the renin–angiotensin–aldosterone system and endothelin pathway. ⁴⁰ , ⁴¹ Estrogens may regulate TG and LDL‐C metabolism through multiple mechanisms. ⁴²

Stratified analysis showed that compared with children, PBF in adolescents performed better in the recognition of dyslipidemia and elevated BP in both sexes. There are potential explanations for these results. First, there are significant age differences in body fat composition between children and adolescents due to puberty development. ⁴³ Compared with children, adolescents may have higher abdominal subcutaneous fat, which has a greater impact on cardiovascular metabolism. ³⁶ , ⁴⁴ Previous studies have shown that the relationship between central fat and BP and blood lipids increases as children mature. ¹⁶ , ⁴⁵

In contrast to males, the predictive ability of PBF on hyperglycemia was lower in female adolescents than that in female children. This may be due to the significant increase of sex steroids in puberty and the significant decrease in insulin sensitivity in females. ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ Although sex steroid, insulin secretion increases and blood glucose greatly fluctuates during female puberty development, our study only considered FBG and did not evaluate insulin sensitivity. This is worth further studies in the future.

Our study had several advantages. Specifically, it included a large sample size, a balanced sex and age distribution, and comprehensive CMR factors. Actually, our study had some limitations. First, our study had a cross‐sectional design; therefore, we cannot infer a causal relationship between PBF and CMR factors. Second, AUC in our results was small (<0.70), and the level of evidence‐based evidence was low. The comprehensive prediction ability of related chronic diseases depends on the verification of the long‐term cohort study. Third, we did not evaluate BIA‐based and gold‐standard DEXA‐based PBF in the same study, and we did not deeply distinguish between subcutaneous or visceral fat. Of course, BIA is a noninvasive, convenient, and economical method with confirmed accuracy and reliability. ⁵⁰ , ⁵¹

In conclusion, BIA‐based PBF, but not BMI, was associated with most CMR factors; meanwhile, overweight and obesity categories based on PBF had an increased risk for cardiometabolic abnormalities in the pediatric population. We advocate the development of public health programs that monitor body composition and CMR in school‐aged children and adolescents.

AUTHOR CONTRIBUTIONS

Xin He: Article design; writing. Zhenni Zhu, Jiajie Zang, and Zhengyuan Wang: Cohort establishment; demographic data analysis; Ping Liao: Serum biochemical detection; data management. Yan Shi and Wenjing Wang: Quality control. Chen Fu: Article design; guidance. All authors participated in the writing of articles and agree to publish them.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

ETHICS STATEMENT

This study met the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethical Committee of the Shanghai Center for Disease Control and Prevention (No. 2015‐15). Informed consents were signed by the guardians of all participants.

Supporting information

Supporting information.

He X, Zhu Z, Zang J, et al. Percent body fat, but not body mass index, is associated with cardiometabolic risk factors in children and adolescents. Chronic Dis Transl Med. 2023;9:143‐153. 10.1002/cdt3.54

DATA AVAILABILITY STATEMENT

The data sets generated and/or analyzed during the current study are not publicly available but are available from the corresponding author at a reasonable request.