Obesogenic eating behaviour and dietary intake in German children and adolescents: results from the GINIplus and LISA birth cohort studies

Overweight and obesity among children and adolescents are major public health issues, affecting about 15% in Germany. Diet-related factors, including energy imbalance, contribute to obesity risk, and acquired eating behaviours may influence food intake decisions beyond internal hunger and satiety cues. Three behaviours linked to low responsiveness to internal signals are external eating (susceptibility to external food cues), emotional eating (eating in response to emotional distress), and dietary restraint (cognitive regulation of eating for weight control). These behaviours have been associated with overweight in youth. Puberty is a critical period due to shifts from parent-controlled diets to greater autonomy, peer influence, and body image concerns. Prior studies on behaviour–diet associations in youth report mixed findings and often focus on single behaviours or limited dietary components. This study aims to describe levels of external eating, emotional eating, and dietary restraint at late childhood (10 years) and adolescence (15 years), and to examine their cross-sectional associations with usual dietary intake in participants of the German GINIplus and LISA birth cohorts.

Existing literature indicates that external, emotional, and restrained eating are linked with overweight among children and adolescents. External eating reflects responsiveness to external food cues; emotional eating involves using food to cope with negative emotions; dietary restraint represents cognitive control over eating, often in those struggling with intake and weight control. Prior adolescent studies have yielded mixed results, frequently examining a single behaviour and focusing on select obesity-related foods (e.g., sweets, soft drinks) or nutrients. Some reports associate emotional eating with higher intake of sweets and energy-dense foods, and dietary restraint with lower energy intake and sometimes healthier choices but also potential underreporting. The transition through puberty introduces psychosocial factors (anxiety, depression, appearance anxiety), particularly in females, potentially heightening emotional eating and dietary restraint.

Design and participants: Cross-sectional analyses within two ongoing prospective German birth cohorts: GINIplus (n=5,991 recruits, 1995–1998 in Munich and Wesel; intervention and observation arms) and LISA (n=3,097 recruits, 1997–1999 in Munich, Wesel, Leipzig, Bad Honnef). Analyses used participants from 10- and 15-year follow-ups with complete exposure, outcome, and covariate data (10 years: N=2,257; females 1,082; males 1,175. 15 years: N=1,880; females 1,000; males 880). Ethical approvals were obtained and informed consent given. Eating behaviour assessment: Eating Behaviour and Weight Problems Inventory for Children (EWI-C) administered to participants at 10 and 15 years, focusing on three subscales: External eating (subscale 1; 8 items, score 0–24), Emotional eating (subscale 3; 8 items, score 0–24), and Dietary restraint (subscale 5; 7 items, score 0–21). External and emotional eating measured at both time points; dietary restraint measured at 15 years only (omitted at 10 years to avoid potential adverse influence on younger children). Items rated on a 4-point Likert scale; subscale scores summed and categorized into age- and sex-specific tertiles (T1 low, T2 medium, T3 high) due to skewed distributions. Dietary intake assessment: Self-administered, validated FFQ (80 items) at 10 and 15 years to estimate usual intake. At 10 years, parents assisted; at 15 years, adolescents self-completed. Frequencies and portion sizes converted to grams/day and linked to energy/nutrients using the German Food Code and Nutrient Database (BLS II.3.1). Data quality exclusions: incomplete FFQ blocks or >50% missing items; implausible total energy intakes (females <500 or >3500 kcal; males <800 or >4000 kcal); implausible % energy contributions for specific items (boxplot outliers). Foods aggregated into 17 food groups. Outcomes included all food groups (mostly expressed as % energy; water/tea in mL/day), nutrients (fat, protein, carbohydrate, fibre g/day, total sugar, SFA, MUFA, PUFA, omega-3 and -6 as % energy), and total daily energy (kJ/day). Food groups were categorized into tertiles; nutrients and total energy treated as continuous variables with total, omega-3, and omega-6 PUFA log-transformed. Covariates: Selected from literature and associated variables (p<0.05): exact age; BMI; pubertal status (10 years: onset yes/no; 15 years: stage pre-mid/late/post); siblings (yes/no); moderate-to-vigorous physical activity (low/medium/high); screen time (low/high); total difficulties (Strengths and Difficulties Questionnaire: normal/borderline/abnormal); parental education (low-medium/high); parental BMI; study (GINIplus observation/intervention, LISA); recruitment region (Munich, Leipzig, Bad Honnef, Wesel); total daily energy or total beverage intake. Statistical analysis: Sex-stratified analyses at each time-point. Multinomial logistic regression for categorical outcomes (food group tertiles) and multiple linear regression for continuous outcomes (nutrients, total energy). Models adjusted for all covariates; total energy included in all models except for water and tea (adjusted for total beverage intake). Results presented as relative risk ratios (RRR), beta coefficients (β), or means ratios (MR) for log-transformed variables with 95% CIs. Multiple testing correction using Nyholt method, yielding two-sided α=0.0019 (0.05/26). Sensitivity analyses: (1) Excluding self-reported vegetarians/vegans; (2) Excluding participants with BMI <10th or >90th percentile; within this subpopulation, tested interactions between each eating behaviour and BMI; significant interactions (p<0.0019) visualized as marginal effects by BMI categories (low <25th, medium 25th–<75th, high ≥75th percentile).

- Sample: N10=2,257 (females 1,082; males 1,175); N15=1,880 (females 1,000; males 880). - Descriptives: Median external eating scores ~6–7 at both ages; higher in 10-year-old males; at 15 years, all eating behaviours higher in females. Emotional eating median at 10 years was 1 in both sexes; at 15 years, medians were 3 in females and 1 in males for both emotional eating and dietary restraint. - Associations at 10 years: • Females: Emotional eating (T2 vs T1) associated with medium vegetable intake (RRR=1.84; p=0.0017). • Males: No significant associations for external or emotional eating. - Associations at 15 years: • External eating: Females showed higher total energy intake with higher external eating (β=718 kJ/day; p=0.0002). Males showed a positive association between external eating (T2 vs T1) and high butter intake (RRR=1.96; p=0.0019). • Emotional eating: No significant associations in either sex. • Dietary restraint (assessed only at 15 years): In females, higher dietary restraint associated with lower total energy intake (β=−967 kJ/day; p<0.0001) and lower omega-3 PUFA intake (MR=0.94; p=0.0017 for medium vs low), and with higher odds of high fruit intake (RRR=2.20; p=0.0003) and high whole grain intake (RRR=1.94; p=0.0013) (medium vs low). No significant associations in males. - Sensitivity analyses: Excluding vegetarians/vegans attenuated associations of dietary restraint with fruit and whole grains in 15-year-old females. After excluding BMI extremes (<10th or >90th percentile), associations persisting included: emotional eating with vegetables at 10 years (females), and external eating and dietary restraint with total energy at 15 years (females). Some associations changed (e.g., butter in 15-year-old males no longer significant). - Effect modification by BMI: Significant interactions observed for dietary restraint (with whole grain, oils, sugar-sweetened foods) and emotional eating (with dairy). Direction of effects differed by BMI strata for some outcomes (e.g., positive in low-BMI, inverse in high-BMI groups). Overall, obesogenic eating behaviours showed few robust associations with specific dietary components, with clearer patterns among adolescent females, particularly for total energy intake and selected food groups.

The study found low levels of obesogenic eating behaviours in a predominantly high socioeconomic status cohort, with external eating highest at both ages but shifting from higher in boys at 10 years to higher in girls at 15 years, and both emotional eating and dietary restraint more pronounced in adolescent girls. External eating in 15-year-old girls was linked to higher total energy intake without consistent ties to specific foods, suggesting cue-responsiveness may broadly increase energy intake. Contrary to some prior studies associating emotional eating with sweets and sugar-sweetened beverages, this analysis—after rigorous multiple testing correction—did not find such associations; instead, a moderate positive association of emotional eating with vegetable intake in 10-year-old girls may reflect parental countermeasures promoting healthier foods. Dietary restraint in girls related to lower energy intake and higher fruit and whole grain consumption, aligning with deliberate weight-control behaviours and healthier choices, though a lower omega-3 PUFA intake suggests potential micronutrient shortfalls or selective eating; underreporting among restrained eaters may also contribute. BMI meaningfully modified several associations, indicating behaviour–diet relationships differ across the weight spectrum even within nominally healthy ranges. Comparatively lower behaviour scores than in earlier German samples may reflect cohort composition (higher SES, fewer mental health problems/overweight) and potential parental influence on reporting at younger ages. Overall, the findings emphasize modest and sex- and age-specific links between obesogenic eating behaviours and diet, particularly in adolescent females.

Obesogenic eating behaviours were generally low and showed limited associations with diet in German children and adolescents from the GINIplus and LISA cohorts. The clearest relationships occurred in adolescent females: external eating associated with higher total energy intake, and dietary restraint associated with lower energy intake and higher fruit and whole grain consumption but lower omega-3 PUFA. BMI modified several associations, underscoring the need to consider body weight as an effect modifier in behaviour–diet research. Future studies should include longitudinal analyses across puberty, consider the full spectrum of eating behaviour dimensions, address potential underreporting, and examine BMI-specific pathways to better understand mechanisms and inform targeted prevention.

- Cohort composition likely overrepresents higher socioeconomic status and lower prevalence of overweight/obesity due to non-random loss to follow-up, limiting generalizability. - Cross-sectional design precludes causal inference and allows for potential reverse causation. - Categorization of skewed exposure and outcome variables may reduce information and statistical power, particularly in higher tertiles. - Eating behaviour assessment (EWI-C subscales) is widely used but not formally validated to the authors’ knowledge; only three of ten subscales were included due to respondent burden. - Dietary restraint was not assessed at 10 years, limiting age comparisons for this behaviour. - Potential underreporting of intake among restrained eaters; at 10 years, parental assistance in questionnaires may have led to underreporting or misclassification. - Some associations were sensitive to exclusion of vegetarians/vegans and to exclusion of BMI extremes, indicating potential subgroup-specific effects.