Changes in chronotype and social jetlag during adolescence and their association with concurrent changes in BMI-SDS and body composition, in the DONALD Study

Chronotype reflects individual preference for sleep–wake timing, and misalignment with social schedules leads to social jetlag (SJL). During adolescence (≈9–18 years), chronotype typically shifts later, with eveningness beginning around age 12 and peaking near age 16 in German populations. Late chronotype and higher SJL in adolescence have been linked to higher BMI and adverse body composition. Adolescence is also a critical window for the development of obesity and unfavorable changes in fat mass that can track into adulthood. The study hypothesized that changes toward a later chronotype and increased SJL during adolescence are associated with concurrent detrimental changes in body composition (BMI-SDS, FMI, FFMI), and that these associations may differ across age groups. Using repeated measures over a median of 2.1 years in the DONALD Study, the authors examined change-on-change associations between MSFsc/SJL and body composition.

Prior research largely supports associations between later chronotype or higher SJL and greater BMI/adiposity in adolescents, though most evidence is cross-sectional. Longitudinal findings are mixed and vary by design and exposure measures. For example, college freshmen with late chronotype gained more BMI over 8 weeks, and later bedtimes from adolescence to adulthood predicted BMI increases over years, whereas a small study in older adolescents found no SJL–BMI association—possibly because the critical exposure window had passed as chronotype shifts earlier after mid-adolescence. Mechanistic pathways proposed include irregular sleep timing, shortened sleep, circadian-metabolic misalignment affecting appetite hormones (ghrelin, leptin), metabolic changes, inflammation, cortisol and stress, and pubertal hormonal influences on circadian rhythms. Dietary behavior shifts toward evening intake may further contribute to adiposity from around age 12.

Design and setting: Prospective open cohort (DONALD Study), Dortmund, Germany, with repeated annual assessments from childhood to adulthood. Chronotype/SJL data collection began in 2014 for participants aged ≥9 years. Ethical approval obtained; informed consent provided. Sample: Adolescents aged 9–18 years with ≥2 completed MCTQs and anthropometry within ±1 year of each MCTQ between 2014 and July 2019. Exclusions: data within 2 weeks after clock changes, adult assessments (≥18 years), incomplete MCTQs, missing paired anthropometry, and single-questionnaire participants for overall analyses. Final sample: 213 adolescents (95 female; 45%) from 181 families; 572 questionnaires. Follow-up: median 2.1 years. Exposure assessment: Munich Chronotype Questionnaire (MCTQ) to compute MSFsc (midpoint of sleep on free days corrected for sleep debt) and SJL (absolute difference between mid-sleep on free vs. school days), both continuous (hours:minutes). Outcomes: Annually measured anthropometrics by trained staff. BMI-SDS using German LMS references; body fat percentage from triceps and subscapular skinfolds (Slaughter equations). FMI = fat mass (kg)/height²; FFMI = fat-free mass (kg)/height². FMI was log10-transformed for analyses. Covariates: Age at baseline, sex, time between measurements, age at take-off (ATO), persons per household, maternal BMI; additional data collected included Tanner stage, age at peak height velocity (APHV), total energy intake (3-day weighed records), and physical activity (questionnaire-derived MET-hours converted to kcal/day). Statistical analysis: Linear mixed-effects regression models assessed change-on-change associations: Δ MSFsc or Δ SJL vs. concurrent Δ BMI-SDS, Δ FFMI, or Δ log10-FMI, with individual-level random effects and correlation structures chosen via AIC/BIC. Minimally sufficient adjustment sets were identified via DAG. Effect modification by age and sex was tested; analyses stratified by age groups: <12, ≥12 to ≤15, and >15 years (open cohort allowed individuals to contribute to multiple strata). Multiple imputation (m=5) addressed missing ATO (n=58; 10%) and maternal BMI (n=5; <1%); results combined with MIANALYZE. Sensitivity analyses included adjustment for energy intake, physical activity, maternal education, season, early-life factors (maternal age at birth, gestational weight gain), exclusion of alarm-clock users on weekends, and stratification by Tanner stage (<2 vs. ≥2). Model diagnostics checked multicollinearity, heteroscedasticity, and residual normality. Translational metrics converted log-FMI betas to percent change and absolute FMI change using the baseline median FMI.

- Sample and temporal trends: Median follow-up 2.1 years; participants completed a median of 3 MCTQs. Across follow-up, MSFsc increased by ~39 minutes, SJL by ~17 minutes, and total sleep duration decreased by ~45 minutes per week. BMI-SDS increased (from −0.01 to 0.05), FMI increased by ~11% (median from 4 to 4 kg/m² with distributional shift), and FFMI increased (15 to 16 kg/m²). - Primary associations (adjusted): • Δ MSFsc was associated with higher Δ log10-FMI: β=0.05 (95% CI: 0.01, 0.08; p=0.01), corresponding to ≈5% increase in FMI per 1-hour increase in MSFsc; translating to ≈0.2 kg/m² absolute FMI increase at the sample’s baseline median FMI. • Δ SJL was associated with higher Δ BMI-SDS: β=0.08 SDS units per 1-hour increase (95% CI: 0.01, 0.14; p=0.02). • Δ SJL was associated with higher Δ log10-FMI: β=0.04 (95% CI: 0.003, 0.08; p=0.03), translating to ≈4% FMI increase per 1-hour increase in SJL (≈0.16 kg/m² using median FMI). - No robust associations with FFMI after adjustment: crude associations with FFMI attenuated to null (e.g., Δ MSFsc adjusted β=0.01, 95% CI: −0.16, 0.17; p=0.92). - Age-stratified results: Associations concentrated in the ≥12 to ≤15 years group (p for interaction <0.001). In this group: • Δ MSFsc vs. Δ BMI-SDS: β=0.14 (95% CI: 0.02, 0.26; p=0.03). • Δ SJL vs. Δ BMI-SDS: β=0.17 (95% CI: 0.07, 0.30; p=0.003). • Δ MSFsc vs. Δ log-FMI: β=0.11 (95% CI: 0.05, 0.16; p<0.001). • Δ SJL vs. Δ log-FMI: β=0.12 (p=0.001). Other age groups showed no significant associations. - Sensitivity analyses: Findings were robust to adjustments for energy intake, physical activity, maternal education, season, early-life factors, and to exclusion of weekend alarm-clock users. Associations were present during puberty (Tanner ≥2) and not earlier; no significant sex interactions were detected.

The study extends prior literature by demonstrating change-on-change associations between increasing lateness in chronotype (MSFsc) and SJL with concurrent increases in adiposity (BMI-SDS and FMI) during adolescence. Importantly, the strongest associations were restricted to ages 12–15, aligning with known pubertal and circadian milestones, suggesting a sensitive period wherein circadian delay and social misalignment may promote fat mass accrual. Null adjusted associations with FFMI imply that changes in body composition linked to chronobiological shifts predominantly reflect fat mass rather than lean mass. Potential mechanisms include irregular sleep timing and duration, circadian-metabolic misalignment affecting appetite-regulating hormones (ghrelin, leptin), inflammatory pathways and cortisol, stress, and diet shifts toward evening intake emerging around age 12. The results underscore the dynamic interplay of puberty-related circadian changes and behavioral factors on adiposity, and support targeted interventions (sleep hygiene, reducing evening light/screen exposure, parental monitoring, school start-time policies) to mitigate risk during mid-adolescence.

Changes toward later chronotype and greater social jetlag during adolescence were associated with concurrent increases in BMI-SDS and FMI, with the most pronounced effects between ages 12 and 15 years. These findings highlight mid-adolescence as a vulnerable window for circadian-related increases in body fat. Public health and clinical strategies that align social schedules with adolescent circadian timing and promote consistent, earlier sleep may help prevent unfavorable changes in body composition. Future research should clarify temporal directionality, assess long-term impacts into adulthood, and further consider puberty status in analyses.

- Generalizability may be limited due to a homogeneous German cohort with high socioeconomic status. - FMI and FFMI were derived from skinfold-based estimates of body fat percentage, which may introduce measurement error despite standardized protocols; this could bias associations toward the null. - Potential for residual confounding by unmeasured factors remains, although associations were robust to extensive adjustments and sensitivity analyses. - Missing covariate data (notably ATO and maternal BMI) were handled via multiple imputation; while preferred over complete-case analysis, imputation assumptions may influence estimates. - Chronotype and SJL were assessed by questionnaire (MCTQ), subject to self-report biases. - Open cohort and age-stratified analyses allow individuals to contribute across strata, which may complicate interpretation though addressed by mixed-effects models. - Causality cannot be inferred; directionality between chronobiology and adiposity requires further longitudinal follow-up into adulthood.