Effects of accelerometer-based sedentary time and physical activity on DEXA-measured fat mass in 6059 children

The study addresses the growing epidemic of childhood obesity and the uncertainty about how objectively measured movement behaviors—sedentary time (ST), light physical activity (LPA), and moderate-to-vigorous physical activity (MVPA)—influence objectively measured body composition over time. Existing PA guidelines for youths (e.g., WHO recommending ≥60 min/day of MVPA) are largely based on cross-sectional, short-term, or questionnaire-based evidence. There is a lack of long-term longitudinal data linking accelerometer-measured behaviors to DEXA-derived fat mass, limited understanding of biological mediation (e.g., glucose, insulin, lipids, inflammation), and questions about temporality and potential reverse causation. The purpose is to examine longitudinal associations of cumulative ST, LPA, and MVPA with repeated measures of BMI, waist circumference, total and trunk fat mass, and lean mass from ages 11 to 24 years; assess mediation by metabolic markers; and test temporal causal relationships, including sex-specific effects, using ALSPAC data.

The paper notes that many prior pediatric movement-behavior studies are cross-sectional or short-term and often rely on questionnaires. Over 140 randomized controlled trials have shown little or no effect of school-based PA interventions on BMI. Identified gaps include limited long-term accelerometer-based evidence on ST, LPA, MVPA in relation to progressive changes in DEXA-measured body composition, uncertainty about whether effects are direct or mediated via metabolic pathways (e.g., inflammation, lipid/glucose metabolism, muscle atrophy), and limited evidence on timing and sex differences. Prior experimental and small human studies suggest roles for insulin resistance mechanisms, lipoprotein lipase regulation, oxidative stress, and inflammation. Meta-analyses have provided mixed findings on reallocating ST to LPA versus MVPA for adiposity outcomes, indicating the need for robust longitudinal analyses.

Design and cohort: Longitudinal analysis of the ALSPAC UK birth cohort. Eligible pregnancies were enrolled 1991–1992 with additional recruitment phases; for analyses after age 7, the sample includes 15,447 pregnancies, 15,658 fetuses, and 14,901 children alive at age 1. Clinic visits occurred at ages 11, 15, 24 years. Primary analytical sample: 6059 participants with at least one valid accelerometer measure (ST, LPA, or MVPA) and at least one DEXA body composition measure between ages 11–24. Reduced analytic subsets: 2457 participants with ≥1 valid ST/PA time point and complete DEXA at all three time points; 917 participants with ≥2 valid movement-behavior time points and complete body composition at all three ages. Exposures: Movement behaviors measured via waist-worn ActiGraph accelerometers. At ages 11 and 15: 7 consecutive days; at age 24: GT3X+ for 4 consecutive days. Valid day: ≥10 h wear (excluding ≥10 min consecutive zero counts); inclusion required ≥3 valid days. Counts aggregated in 60-second epochs (raw 30 Hz). Processing via Kinesoft 3.3.75. Cutpoints: ages 11 and 15—Evenson thresholds (ST 0–<100 cpm; LPA 100–2296 cpm; MVPA >2296 cpm). At 24 years—Troiano MVPA cutoff 2020 cpm. MVPA categories: <40, 40–<60, ≥60 min/day. Outcomes: Anthropometry (height, weight, waist), BMI computed as kg/m². Body composition (total fat mass, trunk fat mass, lean mass) assessed by DEXA (GE Lunar) at ages 11, 15, 24; lean mass-to-fat mass ratio computed. Repeatability coefficient for body fat mass ~0.5 kg. Covariates: Time-varying and baseline factors including age, sex, heart rate, systolic/diastolic blood pressure, smoking status, family history of hypertension/diabetes/high cholesterol/vascular disease, socioeconomic status, fasting metabolic markers (glucose, insulin, HDL-C, LDL-C, triglycerides), high-sensitivity C-reactive protein. Pubertal timing noted (all had attained puberty by 17 years). Missing data: Variables not MCAR (Little’s test p<0.0001). Multiple imputations by chained equations in SPSS v27: 20 imputations, 10 iterations, constraints set to observed min/max; pooled results reported. Skewed variables log-transformed as needed. Statistical analyses: Generalized linear mixed-effects models (GLMM) with identity link for repeated measures assessed longitudinal associations of cumulative ST, LPA, and MVPA (continuous and MVPA categorical) with BMI, waist, total/trunk fat, lean mass, and lean:fat ratio. Random intercepts for participants and family clusters; model selection by BIC. Adjustment sequence: Model 1 unadjusted; Model 2 adjusted for cardiometabolic, socioeconomic, lifestyle covariates; Model 3 additionally adjusted for another movement behavior (e.g., adjust ST model for LPA); Model 4 mutually adjusted for remaining behavior (e.g., adjust ST model also for MVPA). Sex-specific GLMMs conducted. Multiple testing correction with Sidak. Mediation analysis: Structural equation modeling estimated natural direct and indirect effects of cumulative glucose and insulin (and in sensitivity analyses, lipids and hsCRP, and either fat or lean mass depending on outcome) on associations of cumulative ST, LPA, MVPA with cumulative total/trunk fat and lean mass. Bootstrapping with 1000 samples. Proportions of mediation/suppression quantified; ≥1% considered partial. Temporal causal path analyses: Autoregressive cross-lagged models examined temporal relations among ST, LPA, MVPA and total fat mass, lean mass across ages 11 (T1), 15 (T2), and 24 (T3). Models adjusted for covariates; significance p<0.05 indicated temporal precedence. Compositional analysis: Isometric log-ratio transformation applied to examine relative time in ST, LPA, MVPA vs body composition outcomes (reported in Supplementary). Analyses performed in SPSS v27 and AMOS v27.

- Trajectories: From ages 11 to 24 years, ST rose from ~6 to ~9 h/day; LPA fell from ~6 to ~3 h/day; MVPA showed a U/J-shaped pattern with a mid-adolescent dip. - Per-minute associations over 13 years: Each additional min/day of ST associated with +1.3 g total fat mass; each min/day of LPA associated with −3.6 g; each min/day of MVPA associated with −1.3 g. - MVPA intensity/dose: Persistently achieving ≥60 min/day MVPA was associated with a −2.8 g decrease in total fat mass per each minute/day of MVPA relative to persistently <40 min/day. Categorical MVPA (vs <40 min/day) associated with lower fat and trunk fat mass; ≥60 min/day also linked to higher lean mass. - Magnitude of contribution during growth (approximate): ST potentially contributed ~0.7–1.0 kg (≈7–10%) to the ~10 kg total fat mass gained from childhood to young adulthood; LPA potentially reduced fat mass by ~0.95–1.5 kg (≈9.5–15%); time in MVPA associated with 70–170 g (≈0.7–1.7%) reduction per 10 kg gain, noting benefits increase at ≥60 min/day intensity. - Sex-specific findings: In the full cohort, cumulative ST was associated with increased BMI, fat, trunk fat, and lean mass; effects on fat/trunk fat were stronger in females. Cumulative LPA associated with decreased BMI, waist, total and trunk fat, and lean mass in both sexes. Cumulative MVPA and persistent ≥60 min/day MVPA associated with decreased total and trunk fat in both males and females; lean mass increased with higher MVPA. - Mediation/suppression: Glucose generally did not mediate associations. Insulin and LDL-C partially mediated/suppressed some relationships. For MVPA, insulin mediated ~8–9% of the association with decreased total/trunk fat and mediated ~19.5% of the positive association with lean mass; LDL-C also contributed (e.g., 12.8% mediation for total fat from 15 to 24 years in sensitivity analysis). For ST, insulin and LDL-C showed small suppression effects on fat outcomes; lean mass partially mediated the positive association of ST with total fat in some analyses. For LPA, hsCRP partially mediated associations with decreased total fat (~3–8%); insulin mediated ~5–13% in some analyses. - Temporal (cross-lagged) results (n=917): Higher total fat mass at 11 years predicted lower MVPA at 15 years; higher MVPA at 15 predicted lower total fat mass at 24. Higher lean mass at 11 predicted higher ST at 15; higher ST at 15 predicted lower lean mass at 24. No strong bidirectional relationships for ST/LPA with adiposity; MVPA showed meaningful temporal paths with fat mass and lean mass (MVPA15→lean mass24 positive). - Compositional analyses: Time allocated to ST relative to LPA/MVPA associated with higher total and trunk fat mass; reallocating time to LPA or MVPA relative to ST associated with lower total and trunk fat mass.

Findings indicate that movement behaviors are independent and potentially causal determinants of body composition changes from childhood through young adulthood. Cumulative sedentary time is deleterious for adiposity outcomes even after extensive adjustment and mutual control for LPA and MVPA, while LPA and MVPA confer protection against gains in total and trunk fat mass. Mediation analyses suggest metabolic pathways—particularly insulin and LDL cholesterol—partly explain how MVPA lowers adiposity and increases lean mass, whereas glucose plays little mediating role. The temporal analyses support that childhood adiposity may reduce subsequent MVPA, and that increased MVPA in adolescence leads to lower adiposity in young adulthood, underscoring the importance of early-life MVPA. LPA demonstrated effects comparable to or greater than MVPA in reducing fat mass, implying that increasing LPA at scale (e.g., replacing sedentary time with LPA) could be a pragmatic target, especially for youths unable or unwilling to perform MVPA. The compositional perspective reinforces that reducing sedentary time in favor of LPA/MVPA is beneficial. Collectively, results have implications for updating pediatric activity guidelines to emphasize both reducing sedentary time and substantially increasing LPA and sustaining ≥60 min/day of MVPA, particularly emphasizing mid-adolescence timing and earlier intervention.

From ages 11 to 24 years, sedentary time increased and light physical activity decreased markedly, while MVPA showed a mid-adolescent dip. Sedentary time was associated with greater total and truncal fat mass, whereas cumulative LPA and MVPA were associated with lower fat mass. Each minute/day of ST corresponded to a 1.3 g increase in fat mass; each minute/day of LPA and MVPA corresponded to 3.6 g and 1.3 g decreases, respectively. Persistently achieving ≥60 min/day of MVPA produced the strongest reductions in fat mass and increased lean mass, with partial mediation through insulin and LDL-C. Temporal analyses suggest that higher childhood adiposity reduces adolescent MVPA, and higher adolescent MVPA reduces young-adult adiposity. Public health guidance could emphasize reducing ST by about 3 h/day and increasing LPA by a similar amount, while promoting sustained ≥60 min/day MVPA from early childhood. Future research should include experimental and mechanistic studies across diverse populations to clarify pathways and optimize timing and intensity of interventions.

- Generalizability: Cohort predominantly White; findings may not extend to other racial/ethnic groups. - Observational design: Potential residual confounding from unmeasured factors (e.g., quantitative sleep measures unavailable). - Attrition and missing data: Loss to follow-up and missing covariates required multiple imputations; although characteristics were similar between included and excluded participants, bias cannot be fully excluded. - Biomarker timing: Lack of fasting blood samples at baseline (age 11) and misalignment of biomarker collection (e.g., blood at 17 years vs accelerometer/adiposity at 15 or 24 years) may affect mediation analyses; addressed with sensitivity analyses at 15 and 24 years. - Device/cutpoint differences: Different accelerometer models and MVPA cutpoints (Evenson vs Troiano) across ages could influence MVPA trajectories (U/J-shape) and estimates, though absolute agreement between devices is high and sensitivity analyses support robustness.