The role of lifestyle and non-modifiable risk factors in the development of metabolic disturbances from childhood to adolescence

Childhood and adolescent metabolic disturbances (dyslipidemia, hypertension, insulin resistance) are rising alongside the global obesity epidemic, and pediatric metabolic syndrome (MetS) prevalence is increasing. Unfavorable lifestyle factors (unhealthy diet, low physical activity, high media use/sedentary behavior, and low well-being) are linked to overweight, obesity and metabolic disturbances. From the child’s perspective, non-modifiable factors such as family history of MetS components, parental education, maternal BMI, breastfeeding duration, birth weight, and pubertal development also influence risk. Elevated CRP, a marker of inflammation that may mediate links between lifestyle and health, contributes to metabolic disturbances. Metabolic disturbances often persist from childhood to adulthood, increasing morbidity and cardiovascular risk, while remission normalizes risk. Using age- and sex-specific pediatric cut-offs for MetS components and a latent transition modeling framework, this study investigates how lifestyle, non-modifiable factors, and CRP relate to metabolic risk statuses from childhood into adolescence in the IDEFICS/I.Family cohort.

Prior work links unhealthy dietary patterns, low physical activity, and high sedentary/media use with increased pediatric metabolic risk. Psychosocial stress/low well-being has been associated with obesity and metabolic outcomes. Non-modifiable factors (maternal BMI, parental education, birth weight, breastfeeding, pubertal timing, and family history of hypertension, dyslipidemia or type 2 diabetes) have been implicated in metabolic risk. CRP is associated with hypertension, insulin resistance, and central adiposity and may mediate lifestyle–metabolism relationships. Metabolic disturbances track from childhood to later life, reinforcing the need to understand early determinants.

Study design and population: The IDEFICS/I.Family multicenter population-based cohort included children aged 2–9 years at baseline (T0: 2007/2008) from 8 European countries, with follow-ups at T1 (2009/2010) and T3 (2013/2014). Standardized interviews, examinations, and blood sampling were conducted. For this analysis, 3889 children aged 4–15 years who participated at T0 and T3 and provided at least two measurements of all MetS components were included. Exclusions: non-fasting samples (1897 measurements from 1408 children), CRP ≥10 mg/L (none), and medication for diabetes, lipids, hypertension, or obesity (N=54). Outcomes: Age- and sex-specific (and height-specific for blood pressure) reference curves for waist circumference, SBP/DBP, HDL cholesterol, triglycerides, and fasting blood glucose were derived within the cohort; due to assay changes, separate reference curves were used for T0/T1 and T3. Thresholds defined monitoring/action levels (≥90th/95th percentile; HDL ≤10th/5th percentile). Latent statuses (via latent transition analysis, LTA) characterized metabolic profiles: (1) metabolically healthy, (2) abdominal obesity, (3) dyslipidemia, (4) hypertension, (5) several MetS components. Exposures: Lifestyle factors included fruit/vegetable intake frequency (times/day), processed food intake frequency (times/day), psychosocial well-being score (0–48; higher is better), sports club membership (yes/no), and number of media devices in the bedroom (0 vs ≥1). Non-modifiable factors included child age, sex, country, parental education (ISCED low/medium vs high), maternal BMI, family history (hypertension, dyslipidemia, type 2 diabetes), birth weight (g), breastfeeding duration (months), and pubertal status (yes/no). CRP was measured at T0/T1 by nephelometry and at T3 by electrochemiluminescent assay; transformed to age- and sex-specific z-scores using method-specific reference curves. Statistical analysis: LTA (models with 3–7 statuses; 5-status model best by BIC) estimated probabilities of status membership at T0, T1, T3. The vector of status probabilities was treated as compositional data and transformed via additive logratio relative to the metabolically healthy status: ln(P_abdominal_obesity/P_healthy), ln(P_dyslipidemia/P_healthy), ln(P_hypertension/P_healthy), ln(P_MetS/P_healthy). These transformed outcomes were analyzed in multivariate mixed-effects models (SAS Proc MIXED) with a random subject-specific intercept and repeated measures across waves; factor-analytic covariance structure for random effects. Multiple imputation addressed missing covariates (SAS Proc MI/MIANALYZE). Continuous covariates were centered (e.g., fruit/veg to 5 times/day, processed foods to 0 times/day, well-being to 40, maternal BMI to 23 kg/m², birth weight to 3500 g, breastfeeding to 6 months, age to 8 years; CRP z-score to 0). The base model included all lifestyle and non-modifiable factors and their interactions with age; non-significant age interactions (p≥0.10 across all outcomes) were removed for processed foods. CRP z-score (with age interaction) was added subsequently as a potential mediator. Sensitivity analyses for dyslipidemia and hypertension additionally adjusted for BMI z-score.

- Sample and latent statuses: Among 3889 children, the proportions in latent statuses across T0, T1, T3 were: metabolically healthy 67.8%, 61.7%, 64.9%; abdominal obesity 15.2%, 17.4%, 17.5%; dyslipidemia 6.6%, 6.3%, 4.3%; hypertension 4.9%, 5.6%, 2.7%; several MetS components 5.6%, 9.1%, 10.6%. - Non-modifiable factors: • Low/medium parental education increased risk at age 8 for abdominal obesity (OR 1.14) and for several MetS components (OR 1.25) relative to metabolically healthy. • Higher maternal BMI increased risk for all outcomes: abdominal obesity OR 1.29 (95% CI 1.25–1.34), dyslipidemia OR 1.09 (1.07–1.11), hypertension OR 1.10 (1.07–1.12), several MetS components OR 1.47 (1.39–1.55). • Family history of hypertension increased risk across outcomes (directionally consistent; specific ORs not all reported in text). • Early puberty associated with higher risk: abdominal obesity OR 2.43 (1.60–3.69) at age 8 with age interaction 0.75 (0.69–0.81) per year; several MetS components OR 2.46 (1.53–3.96) at age 8 with age interaction 0.71 (0.65–0.77). - CRP: • A 1 SD increase in CRP z-score increased risk for all four metabolic outcomes at age 8 (strong associations; exact ORs not fully detailed in text); effects persisted after mutual adjustment; sensitivity analyses considered BMI z-score for select outcomes. - Lifestyle factors: • Bedroom media (≥1 device) increased risk of several MetS components with strong age-dependence: OR 1.30 (1.18–1.43) at age 8; age interaction 1.18 (1.14–1.23), implying OR ≈1.53 at age 9 and ≈2.97 at age 13. • Not being a sports club member increased risk: dyslipidemia OR 1.16 (1.07–1.26) and several MetS components OR 1.30 (1.18–1.42) at age 8. • Higher well-being associated with lower risk of abdominal obesity: OR 0.90 (0.82–0.98). • No consistent associations were observed for breastfeeding duration, fruit and vegetable intake frequency, or processed food frequency in mutually adjusted models. - Age-dependency: Most associations showed limited age-dependence except for bedroom media and puberty timing, which displayed marked interactions with age.

Findings indicate that both modifiable lifestyle-related behaviors (particularly proxies for physical activity and sedentary behavior: sports club membership and bedroom media presence) and non-modifiable factors (maternal BMI, parental education, pubertal timing, family history of hypertension) contribute to metabolic risk profiles from childhood into adolescence. Elevated CRP, potentially reflecting inflammatory pathways linking lifestyle to metabolic dysregulation, was associated with increased risk across all metabolic outcomes. The strong age-dependent effect of bedroom media suggests that sedentary behavior patterns may become increasingly influential as children age. Early puberty was associated with higher risk for abdominal obesity and multiple MetS components at younger ages, with the relative effect diminishing across ages, consistent with maturation-related changes. Overall, results support the hypothesis that a constellation of behavioral, biological, and familial factors determine transitions among metabolic health statuses, underscoring opportunities for early prevention and the need to consider age effects.

This study demonstrates that multiple lifestyle, familial, developmental, and inflammatory factors shape children’s metabolic risk trajectories. Preventive efforts should be multifactorial, with practical targets including removing media devices from children’s bedrooms and promoting participation in sports clubs, alongside broader strategies addressing family and social determinants. The work highlights the utility of latent transition modeling to track metabolic statuses over time and to inform age-appropriate intervention points.

- Physical activity and sedentary behavior were assessed via proxies (sports club membership; media devices in bedroom) because objective measures were available only in a subset, potentially inducing measurement error. - Dietary exposures were based on reported frequency measures, which may not capture overall diet quality or energy intake comprehensively. - Laboratory assay methods for glucose and lipids (and for CRP) differed between waves (T0/T1 vs T3), necessitating separate reference curves, which may affect comparability across time. - Observational design limits causal inference; residual confounding is possible despite adjustment and multiple imputation for missing covariates. - Generalizability is to European children participating in the IDEFICS/I.Family cohort; attrition between waves may introduce selection effects.