The China Child and Adolescent Cardiovascular Health (CCACH) project is a comprehensive, cross-sectional population investigation carried out from 2013 to 2015 to evaluate the cardiovascular health of urban children and adolescents in China (14). We selected four northern cities (Beijing, Changchun, Jinan, and Yinchuan) and three southern cities (Shanghai, Chongqing, and Chengdu), all of which are municipalities directly under central government jurisdiction or provincial capitals. In each city, schools were randomly selected to ensure a representative sample in terms of age, sex, and socioeconomic status. In total, 11,435 children underwent dual-energy X-ray absorptiometry (DXA).

The Ethical Review Committee of the Capital Institute of Pediatrics approved this study (approval number: 2012062), and all participants or their guardians provided written informed consent prior to participation.

Before the research began, we developed a unified operation manual for all the measurement procedures. All participants removed their shoes and wore single-layer trousers during the examination. All assessments were conducted by qualified health professionals following a standardized protocol, as described in our previous study (15, 16). Height and weight were measured using a calibrated digital scale (Jianmin II, China Institute of Sport Science, Beijing, China) with a precision of 0.1 kg and 0.1 cm. Waist circumference was measured to the nearest 0.1 cm. Height, weight, and waist circumference measurements were obtained consecutively and averaged. Body mass index (BMI) was calculated as weight (kg) divided by the square of height (m²).

Systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured by trained personnel using a calibrated automatic electronic blood pressure monitor (OMRON HEM-7012, Omron Corporation, Kyoto, Japan). Each child’s resting blood pressure was measured three times, and the mean of the last two measurements was recorded for analysis. If the difference between two consecutive readings exceeded 10 mm Hg, additional measurements were obtained.

After a 10-h overnight fast, blood samples were collected from the antecubital veins using ethylenediaminetetraacetic acid (EDTA)-anticoagulated vacuum tubes.

Samples were centrifuged, aliquoted, and stored at −80 °C until analysis. Total cholesterol (TC), triglycerides (TG), and insulin levels were quantified using enzyme-linked immunosorbent assay (ELISA, Quantikine, R&D Systems, MN, USA). High-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) were measured via a direct method (Sekisui Medical, Tokyo, Japan). Fasting plasma glucose (FPG) was determined by a hexokinase assay (BIOSINO, Beijing, China). Biochemical parameters were analyzed using an automated analyzer (Sekisui Medical, Tokyo, Japan), with an interassay coefficient of variation of <6.0% for all measurements. The Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) was calculated as follows: [Fasting insulin (mU/L) × FPG (mmol/L)]/22.5.

All participants were assessed using Hologic Discovery (A, W, and Wi) fan–beam densitometers. The APEX 4.0 software (Hologic, Bedford, MA, USA) was used to segment the body into anatomical regions, including the head and limbs. The android region, defined as the abdominal area, extends from the midpoint of the lumbar spine at the waist to the superior border of the pelvis. The gynoid region (hip area) is defined as the region between the femoral head and the midpoint of the thigh. Appendicular skeletal muscle mass (ASMM) was assessed and used as the primary indicator of muscle mass, calculated as the sum of lean mass from the arms and legs (ASMM = ASMM_arm + ASMM_leg). Whole-body and regional fat mass (FM, kg) and fat-free mass (FFM, kg) were quantified.

The questionnaire was designed and validated through pilot testing. For children under 12 years of age, the questionnaire was completed with assistance provided by their parents or guardians, whereas children aged 12 years and older self-reported their responses independently. Data were collected on demographic characteristics (sex, date of birth, and ethnicity), lifestyle factors (dietary habits, physical activity, and sleep patterns), early life conditions (preterm birth and birth weight), family medical history (parental obesity, and hypertension), and socioeconomic status (parental education level and annual household income). All participants were asked about their menarche or spermarche experiences and categorized by pubertal stage based on self-reported Tanner criteria.

Table 1 presents dietary intake records and their evaluation based on daily consumption frequency. Fast food was defined as fried items, including churros, fried chicken, hamburgers, and similar products. To facilitate interpretation, food items were categorized into two groups based on median frequency. For analytical clarity, food groups were dichotomized according to median consumption frequency. Physical activity was assessed through self-reported frequency and duration of moderate-to-vigorous physical activity (MVPA) during the preceding 12 months. The WHO recommends ≥60 min of daily MVPA as adequate for children and adolescents; participants not meeting this threshold were classified as insufficient (17). Sedentary behavior was further categorized based on daily sitting duration (<2 h vs. ≥2 h).

Total daily sleep duration includes both nocturnal and daytime sleep. According to the National Sleep Foundation, sleep duration below the following thresholds is classified as insufficient: fewer than 9 h per day for children aged 6–13 years, fewer than 8 h for adolescents aged 14–17 years, and fewer than 7 h for adults aged 18 years or older (18). Experiencing a delay of over 30 min in achieving sleep is categorized as having difficulty initiating sleep. Sleep disorders were assessed based on self-reported symptoms, including snoring, mouth breathing, breath cessation, somnambulism, nightmares, limb jerking, bedwetting, difficulty rising in the morning, or nocturnal headaches.

The obesity threshold was established in accordance with the Chinese National Standard WS/T 586–2018 (19). We employed two standard definitions (SDs) to classify children and adolescents with obesity into MHO and MUO phenotypes. The assessment of the following traditional CR should be conducted initially (12): (1) waist circumference (WC) ≥ 90th percentile for age and sex; (2) SBP and/or DBP ≥ 90th percentile for age and sex; (3) TG levels of ≥1.12 mmol/L for individuals under 10 years and ≥1.46 mmol/L for those over 10 years; (4) HDL-C levels ≤1.03 mmol/L; and (5) FPG levels ≥5.6 mmol/L. MHO-CR was defined as obesity accompanied by one or fewer cardiometabolic risk factors, whereas MUO-CR was defined as obesity with two or more cardiometabolic risk factors. The second definition depends on the existence or non-existence of IR, and there is no universally recognized threshold for IR in children and adolescents. Based on a study conducted in China (20), we determined that the most suitable cut-off value for our cohort was 2.3. Accordingly, MHO-IR was defined as HOMA-IR ≤ 2.3, and MUO-IR was defined as HOMA-IR > 2.3.

Continuous variables were tested for normality using histogram methods and the Shapiro–Wilk test. Those conforming to a normal distribution are presented as ⁻x ± s and compared using an independent sample t-test; non-normally distributed variables are expressed as median (interquartile range, IQR) and analyzed with the Mann–Whitney U test. Categorical variables are reported as proportions (%) and assessed with the chi-square tests. To account for age-related physiological variations (6–18 years), we standardized BMI, WC, and body composition measures using regression-based z-scores (internally studentized residuals). For each outcome variable (e.g., BMI), a linear regression model was fitted with age as the predictor:

where,

e_i = raw residual (observed value—model-predicted value),

σ ̂ = estimated standard deviation of residuals, and.

h_ii = leverage value (a measure of how far an observation deviates from the mean predictor values).

Logistic regression was employed to examine the relationship between each variable and the MHO phenotype, controlling for age, sex, region, and pubertal development. A multi-factor test was subsequently conducted by incorporating all pertinent MHO factors into the final logistic regression model. In the multivariate regression analyses, variance inflation factors (VIFs) were used to evaluate multicollinearity among independent variables. VIF values ≥ 10 indicate severe multicollinearity, suggesting such variables should not be retained simultaneously in the model.

Lifestyle factors and body composition indicators are more likely to influence metabolic health in a multiplicative manner (that is, the effect of one factor changes proportionally with the level of the other factor) rather than a simple additive effect. Therefore, logistic regression analyses were employed to examine the multiplicative interaction between DXA measures and lifestyle factors by including a product term (e.g., FM × Sleep duration) in logistic regression models. Lifestyle factors were entered as bicategorical variables, with adjustments made for age, sex, region, and pubertal development. The analysis used R version 4.3.2 software. A two-tailed p-value of < 0.05 was considered statistically significant.

This study analysed 1,357 participants aged 6–19 years (65.6% male; mean age 10.8 ± 3.6 years). Figure 1 shows the screening flowchart. Table 1 presents demographic and clinical characteristics of participants stratified by MHO and MUO phenotypes according to both definitions. Based on the CR definition, 43.6% of children with obesity were classified as MUO, while 38.9% met MUO criteria by IR thresholds. Among the total cohort, 175 (19.6%) were MUO by both CR and IR definitions, whereas 309 (34.6%) were MHO across both definitions.

The MHO group demonstrated better cardiometabolic profiles. Compared to participants with MUO, the MHO group was younger and had lower BMI, FM, and VAT, and higher FFM. MHO defined by CR criteria was associated with healthier dietary patterns (lower soft drinks and fast food intake) and lower maternal obesity prevalence. MHO, defined by IR criteria, correlated with delayed pubertal onset, reduced sedentary behavior, higher skeletal muscle mass, increased dairy consumption, and higher parental education levels.

Table 2 presents the factors associated with MUO phenotypes after adjusting for age, sex, pubertal development, and residence. Body composition factors, namely FM, android FM, arm FM, VAT, and FFM, significantly predicted MUO under both definitions (all p < 0.03). In contrast, ASMM, arm ASMM, and leg ASMM showed no significant association with MUO status under either definition (all p > 0.05). Among the behavioral factors, soft drinks (p = 0.016), fast food (p = 0.026), and a mother with obesity (p = 0.002) emerged as the most significant determinants of MUO, as defined by CR. Sleep duration (p = 0.052) and parental education level (p < 0.010) were significant factors correlated with MUO, as defined by IR.

To identify the independent factors associated with MUO, variables showing significant associations in the univariate analysis (Table 2) were included in a multivariable logistic regression model. After completely adjusting for age, sex, geographical region, and pubertal development status (Table 3), FM, android FM, VAT, and FFM remained independently associated with MUO status under both CR and IR definitions. For each standard deviation (SD) increase in FM, the odds of MUO-CR and MUO-IR both increased by 21%. Each SD increase in android FM was associated with 41% higher odds of MUO-CR and 20% higher odds of MUO-IR. Similarly, a 1-SD increase in VAT increased the odds of MUO-CR by 27% and MUO-IR by 20%. For FFM, each SD increment elevated the odds of MUO-CR by 16% and MUO-IR by 17%.

Among behavioral factors, participants consuming fast food ≥1 time/week had 36% higher odds of MUO-CR compared to those with less frequent consumption. Longer sleep duration per SD was associated with 27% lower odds of MUO-IR.

To investigate potential interactions between body composition and lifestyle behaviors on MHO and MUO phenotypes, we assessed the most significant lifestyle factors associated with each phenotype under different metabolic criteria. Specifically, fast food consumption was analyzed as a key dietary factor for MUO individuals meeting CR criteria. At the same time, sleep duration was evaluated as a critical behavioral factor for MUO individuals meeting IR criteria. As shown in Table 4, frequent fast food intake (≥1 time/week) was associated with an enhanced effect on MUO-CR in individuals with higher FM (odds ratio (OR) = 1.36, 95% confidence interval (CI): 1.01–1.82; p = 0.042) and arm FM (OR = 1.37, 95% CI: 1.02–1.83; p = 0.036). These findings suggest that FM and arm FM may potentiate the adverse effects of fast food intake on MUO-CR risk. In contrast, FFM (OR = 0.72, 95% CI 0.54–0.97; p = 0.032) appeared to attenuate these effects, indicating a protective role of lean body mass. For MUO-IR, sleep duration interacted multiplicatively with FM, android FM, gynoid FM, trunk FM, arm FM, and FFM (all p < 0.029) (Figure 2).