Dietary patterns among European children and their association with adiposity-related outcomes: a multi-country study

The study addresses the high and rising prevalence of overweight/obesity among European school-aged children and the long-term health risks associated with early-life obesity. Dietary quality has worsened over recent decades with shifts toward highly processed diets, potentially increasing obesity risk. While both a priori (e.g., Mediterranean diet indices) and a posteriori (data-driven) methods exist to characterize diet, no prior study had described both types of patterns across multiple European countries in children or prospectively assessed their associations with adiposity-related outcomes. The objective was to derive data-driven and a priori dietary patterns in primary school-age children from six European countries and examine their cross-sectional and prospective associations with zBMI, fat mass percentage, and waist-to-height ratio from childhood to adolescence.

Prior work shows diet-related diseases substantially contribute to premature mortality in Europe. Dietary transitions toward ultra-processed foods are linked to poorer diet quality and increased obesity risk. Studies using data-driven dietary patterns often identify obesogenic “Western” patterns (high in processed foods and sugars) associated with higher obesity risk, though evidence is mixed; healthier patterns rich in fruits, vegetables, and fish typically show weaker or inconsistent associations. Adherence to Mediterranean-like patterns in European children varies, sometimes higher in Northern than Southern Europe, and literature relating Mediterranean diet adherence to adiposity in children/adolescents is inconsistent. Socioeconomic disparities influence diet quality, with lower parental education and family wealth linked to less healthy diets. Methodological considerations include reliance on FFQs and varying definitions/compositions of dietary patterns across studies.

Design and population: Longitudinal analysis within the HELIX project including six European birth cohorts (BiB-UK, EDEN-France, INMA-Spain, KANC-Lithuania, MoBa-Norway, RHEA-Greece). Childhood follow-up at ages 6–11 years (2013–2016): n=1,597 with dietary data; adolescence follow-up at 12–18 years: n=803 with adiposity outcomes. Ethics approvals obtained at each site with informed consent/assent. Dietary assessment: Parent-completed semi-quantitative food frequency questionnaire (FFQ) at childhood visit (previous year intake), harmonized across cohorts; 44 items aggregated to 15 food groups (weekly frequency). Three groups (bread, breakfast cereal, cereals) were excluded due to low KMO (<0.5). A priori index: KIDMED (Mediterranean diet quality index) scored and categorized as high (>4), average (1–4), low (<1). Data-driven patterns: Exploratory factor analysis (EFA) with orthogonal varimax rotation on food groups; adequacy checked with Bartlett’s test (p<0.05) and KMO. Number of factors based on eigenvalues >1, scree plot/parallel analysis, and interpretability. Food groups with absolute loadings ≥0.25 defined each factor. Participant factor scores were computed by summing standardized intakes weighted by retained loadings; adherence categorized into tertiles (low/average/high). Adiposity outcomes: Height and weight measured using standardized equipment; zBMI computed using WHO age- and sex-adjusted reference. Waist circumference measured twice; waist-to-height ratio (WHtR) computed; WHtR ≥0.5 referenced as an obesity indicator. Body composition by bioimpedance (Bodystat 1500), with fat-free mass and fat mass percentage estimated using published equations (Clasey et al. for childhood; Schaefer et al. for adolescence). Covariates: Cohort, child/adolescent age, sex, maternal pre-pregnancy BMI, maternal smoking during pregnancy, maternal education (ISCED low/middle/high), family affluence score (low/middle/high), child sedentary behavior (minutes/day). Statistical analysis: Separate cross-sectional (childhood) and prospective (childhood diet to adolescent outcomes) linear regression models for each outcome (zBMI, fat mass %, WHtR), adjusted for covariates. For KIDMED, high adherence was reference. For data-driven patterns, the healthiest tertile served as reference (e.g., high for Healthy pattern; low for Western pattern). Inverse probability weighting (WeightIt R package, method “energy”) addressed loss to follow-up in adolescent analyses, weighting on cohort, maternal BMI and education, family affluence score, child sex, and sedentary behavior. Two-tailed tests at 5% significance; analyses in R (2022.02.0).

Dietary patterns: Five data-driven patterns identified with cumulative explained variance 45%: Meat (meat, processed meat), Dairy (yogurt, dairy), Western (sweets, beverages, potatoes, bakery products), Healthy (vegetables, fruits, fish), Sweets and fats (sweets, oils and fats). Cross-country differences: Norwegian children (MoBa) had higher intake of fruits (14.1 portions/week) and vegetables (8.5/week), highest Healthy pattern adherence and greater KIDMED adherence; Lithuanian children (KANC) had the highest sweets intake (9.5/week) and the highest Western pattern adherence. KIDMED adherence high overall was 18.2% (range: 31.7% Norway to 5.4% Lithuania). Cross-sectional associations in childhood (Table 3): - Dairy pattern: Low vs high adherence associated with lower zBMI β=-0.18 (95% CI -0.34, -0.02) and lower fat mass % β=-1.22 (-2.27, -0.27). - Healthy pattern: Average vs high adherence associated with higher zBMI β=0.20 (0.04, 0.35); Average and Low vs high associated with higher fat mass % (Average β=1.44 (0.48, 2.39); Low β=1.10 (0.09, 2.12)) and higher WHtR (point increases ~0.01–0.02). - Western, Meat, Sweets and fats patterns: No significant associations with childhood outcomes. - KIDMED index: No significant associations with zBMI, fat mass %, or WHtR. Prospective associations to adolescence (Table 4): - Healthy pattern (childhood): Low vs high adherence associated with higher adolescent zBMI β=0.28 (0.01, 0.55), higher fat mass % β=2.35 (0.14, 4.84), and higher WHtR β=0.02 (0.01, 0.03). - Western pattern (childhood): High vs low adherence associated with slightly lower adolescent WHtR β=-0.02 (-0.03, -0.00); no significant association with adolescent zBMI or fat mass %. - Dairy pattern associations seen cross-sectionally did not persist prospectively. - KIDMED index: No significant prospective associations with adolescent adiposity outcomes.

The study shows substantial cross-country variability in dietary patterns among European children and identifies a consistent relationship between lower adherence to a Healthy dietary pattern (high in fruits, vegetables, fish) and greater adiposity, evidenced by higher fat mass percentage and waist-to-height ratio both cross-sectionally and prospectively, and higher adolescent zBMI. The lack of association between KIDMED (Mediterranean diet adherence) and adiposity may reflect measurement limitations, heterogeneity in how Mediterranean-like diets are expressed across countries, or the index’s insensitivity to specific obesogenic/anti-obesogenic food group combinations in this age group. Null or counterintuitive findings for the Western pattern could reflect underreporting of unhealthy foods, parental modification of diets for overweight children, or heterogeneity in Western pattern composition across populations. Socioeconomic position and national food environments likely contribute to differences in adherence (e.g., higher Healthy pattern adherence in Norway; higher Western/sweets patterns in more deprived cohorts), aligning with evidence that lower parental education and family wealth are associated with less healthy dietary habits. Overall, the results support focusing on improving adherence to healthy dietary patterns in childhood to mitigate adiposity-related risks into adolescence.

This multi-country study derived five data-driven dietary patterns in European school-aged children and found that low adherence to a Healthy pattern (fruits, vegetables, fish) is associated with higher adiposity indicators in childhood and adolescence. KIDMED-based Mediterranean diet adherence was not significantly associated with adiposity. The findings underscore the value of assessing overall dietary patterns to tailor dietary advice and inform public health strategies aimed at preventing excess weight gain from early life. Future research should use validated, multi-method dietary assessments with energy intake estimates, larger samples enabling cohort-level stratification, and consider country-specific food environments and socioeconomic determinants to refine targeted interventions.

- Dietary intake measured by parent-reported FFQ, potentially excluding foods consumed outside parental control (e.g., school meals) and subject to misreporting/underreporting of obesogenic foods. - The FFQ used was not validated; only one FFQ time point was analyzed. - Lack of total energy intake data precluded energy adjustment in models. - Higher socioeconomic status representation may limit generalizability; relatively small sample per country limited stratified analyses. - Potential residual confounding (e.g., genetics, physical activity beyond sedentary time). - Loss to follow-up in adolescence addressed with inverse probability weighting, but selection bias may remain. - Factor analysis decisions (e.g., excluded food groups; loadings threshold) may influence identified patterns and their interpretability.