Total energy expenditure is repeatable in adults but not associated with short-term changes in body composition

The study addresses whether total energy expenditure (TEE) is a reliable, repeatable trait over time and whether lower TEE predisposes individuals to weight or fat gain. Obesity is prevalent and driven by energy imbalance, making accurate TEE measurements essential. Prior evidence on TEE as a risk factor for obesity is mixed, with some studies suggesting associations between low energy expenditure and subsequent weight gain, particularly in controlled settings, while others find no predictive relationship in free-living adults or children. Methodological issues such as small sample sizes, slow natural changes in weight/adiposity, and uncertainty about TEE measurement repeatability may explain conflicting findings. The authors aim to (1) determine whether age affects repeatability of TEE, and (2) test whether TEE is associated with short-term changes in weight or body composition, using repeated DLW-based TEE measures from an international database.

Earlier DLW studies reported that women with obesity have similar body size- and composition-adjusted TEE as normal-weight women, while pediatric studies have reported both that low TEE predicts greater fat gain and, conversely, that high TEE predicts faster fat gain in preadolescent girls. Two studies using whole-room calorimetry found adults with low 24-h energy expenditure were more likely to gain mass over 2.0–6.7 years. In contrast, several longitudinal studies in infants, children, and adult women found TEE did not predict changes in body fat percentage. Energy expenditure estimated via heart-rate monitoring showed age-dependent associations with fat mass change (inverse under 54 years; positive over 54 years). Regarding repeatability, basal metabolic rate and 24-h expenditure show repeatability in humans and other animals, though repeatability can decline with longer intervals and field conditions. TEE repeatability evidence is limited: small studies showed good short-term reproducibility and one study (N=50, ages 20–50) showed agreement over up to 2.5 years. It remained unclear whether TEE repeatability depends on measurement interval and whether children or older adults differ in repeatability.

Data source and inclusion: Repeated TEE measurements were extracted from the IAEA DLW database (version 3.1.2), which standardizes TEE calculations across studies. Inclusion criteria were age ≥1 year, at least two TEE measures, healthy status, and not engaged in athletic training. The final sample comprised 696 TEE measures from 348 adults (21–89 years) and 114 TEE measures from 47 children (ages 2, 4, and 6; no repeated measures between 7–20 years). Mean interval between measures within individuals was 1.9 ± 2.88 years (range 0.04–8.2 years). Body weight, TEE, fat-free mass (FFM), and fat mass (FM) were taken from the database; FFM was estimated by isotope dilution with corrections for age- and weight-related variation in dilution space and FFM hydration; FM and body fat percentage were calculated as body weight minus FFM. Repeatability analysis: Repeatability (intra-class correlation, R = VG/(VG+VR)) of ln-transformed TEE was estimated using mixed-effects models (rptR in R 3.6.2), with individual ID as a random effect and FFM, FM (both ln-transformed), sex, and age as fixed effects to obtain adjusted repeatability. Analyses were performed for all individuals combined, and separately for adults and children. Parametric bootstrapping provided 95% CIs; significance was tested via likelihood ratio tests (LRTs) and permutation tests. Body mass repeatability (adjusted for sex and age) was also estimated. Association analyses between TEE and changes in body composition (adults 20–60 y): Two complementary approaches were used on (a) all eligible adults 20–60 y (N=267; interval 7.4 ± 12.2 weeks) and (b) a subset with intervals >4 weeks (N=53; 29.1 ± 12.8 weeks). - Approach 1 (covariance decomposition): Multivariate Bayesian mixed models (MCMCglmm) estimated phenotypic (rP), among-individual (rIND), and within-individual (rE) correlations between unadjusted TEE and FM (Model 1) or body fat percentage (Model 2); repeated for the >4-week subset (Models 3 and 4). Models included individual ID as a random effect, and sex, age, and FFM as fixed effects (FFM fitted only for TEE). Continuous variables were standardized (mean 0, variance 1). Unstructured variance–covariance matrices were used. Priors were inverse-gamma; MCMC: 900,000 iterations, 30,000 burn-in, thinning 250; three chains with convergence assessed by trace/density inspection and Gelman–Rubin <1.1. - Approach 2 (adjusted TEE metric): A multiple regression predicted TEE from FFM (ln), FM (ln), age, and sex; adjusted TEE for each measurement was computed as (Observed TEE / Predicted TEE) × 100. Associations were tested between (i) adjusted TEE at time 1, (ii) the mean of adjusted TEE1 and adjusted TEE2, and (iii) the difference (adjusted TEE2 − adjusted TEE1), versus subsequent changes in body weight and body fat percentage. Additional models evaluated change in adjusted TEE per week versus future changes in body weight and body fat percentage. Analyses were performed in both datasets (all adults and >4-week subset).

- Repeatability of adjusted TEE (all individuals combined): R = 0.54, SE = 0.035; 95% CI 0.472–0.608; PLRT < 0.0001; PPermutation < 0.001. - Adults: adjusted TEE repeatable with R = 0.64, SE = 0.033; 95% CI 0.578–0.703; PLRT < 0.0001; PPermutation < 0.001. - Children: adjusted TEE not repeatable, R = 0.00, SE = 0.077; 95% CI 0.000–0.262; PLRT = 1.0; PPermutation = 1.0. - Body mass repeatability higher than TEE: • All individuals: R = 0.96, SE = 0.004; 95% CI 0.952–0.967; PLRT < 0.0001; PPermutation < 0.001. • Adults: R = 0.94, SE = 0.006; 95% CI 0.929–0.952; PLRT < 0.0001; PPermutation < 0.001. • Children: R = 0.38, SE = 0.107; 95% CI 0.166–0.583; PLRT < 0.0001; PPermutation = 0.012. - Adjusted TEE repeatability did not decline with increasing time between measurements (Supplementary Fig. 1a–c). - Covariance decomposition (Bayesian mixed models): No significant phenotypic (rP), among-individual (rIND), or within-individual (rE) correlations between unadjusted TEE and FM or body fat percentage in either dataset. Examples: • Model 1 (TEE × FM, N=267): rP = −0.07 (95% CI −0.16 to 0.04); rIND = −0.09 (−0.21 to 0.05); rE = 0.04 (−0.10 to 0.16). • Model 2 (TEE × %FM, N=267): rP = −0.04 (−0.14 to 0.07); rIND = −0.01 (−0.15 to 0.12); rE = −0.05 (−0.19 to 0.06). • Model 3 (TEE × FM, >4 weeks subset N=53): rP = 0.09 (−0.16 to 0.31); rIND = 0.07 (−0.29 to 0.40); rE = 0.29 (−0.02 to 0.47). • Model 4 (TEE × %FM, >4 weeks subset N=53): rP = 0.19 (−0.06 to 0.38); rIND = −0.23 (−0.14 to 0.50); rE = 0.18 (−0.09 to 0.40). All 95% CIs include zero. - Adjusted TEE associations with change outcomes (N=267 unless noted): • Adjusted TEE at time 1 vs. change in body weight: estimate −0.001 ± 0.002; t = −0.612; df = 265; P = 0.541; adjusted R2 = −0.002. • Adjusted TEE at time 1 vs. change in body fat percentage: estimate 0.020 ± 0.017; t = 1.206; df = 265; P = 0.229; adjusted R2 = 0.001. • Difference in adjusted TEE (TEE2 − TEE1) vs. change in body weight: positive association, estimate 0.009 ± 0.003; df = 265; P = 0.01; adjusted R2 = 0.020 (subjects with greater adjusted TEE2 tended to weigh more). • Average adjusted TEE vs. change in body weight: estimate 0.001 ± 0.003; t = 0.396; df = 265; P = 0.692; adjusted R2 = −0.003. • Neither average adjusted TEE nor difference in adjusted TEE associated with change in body fat percentage (both datasets). • Subset (>4 weeks, N=53): trend towards a negative relationship between adjusted TEE1 and change in body weight (P = 0.094), but no other significant associations. - The observed positive relationship between changes in adjusted TEE and weight change contradicts a simple model predicting ~−0.1 kg per 10% TEE increase per week (based on ~7 MJ per kg), challenging the expectation that decreased TEE drives weight gain in the short term.

Findings indicate that adjusted TEE is a stable, repeatable individual trait in adults over periods up to years, but not in children aged 2–6 years, likely reflecting developmental changes in organ size, activity, and behavior. Despite substantial inter-individual variation in TEE (even after accounting for body size/composition), neither unadjusted nor adjusted TEE predicts short-term changes in body fat mass or body fat percentage in adults. The only significant association observed was that increases in adjusted TEE between measurements were positively associated with weight gain, contrary to a simple energy balance expectation of weight loss with higher TEE. These results suggest that over the weeks-to-months timescales examined, variations in TEE are not a primary driver of changes in adiposity; other factors such as compensatory changes in energy intake, constrained energy expenditure, and behavioral adaptations may play larger roles. The durability of adult metabolic rate underscores consistent individual differences, but these do not translate into predictable short-term changes in body composition.

The study demonstrates that total energy expenditure measured by DLW is highly repeatable in adults but not in young children, and that lower TEE is not a risk factor for, nor higher TEE protective against, short-term weight or fat gain in free-living adults. Multi-model analyses found no meaningful within- or among-individual correlations between TEE and body fat. These findings reconcile mixed prior literature by highlighting the importance of large samples and reliable measurements, and suggest that interventions or predictions based solely on TEE are unlikely to affect short-term adiposity outcomes. Future work should examine longer timeframes, incorporate objective physical activity metrics, resting energy expenditure, and organ size measures to parse mechanisms underlying metabolic variation and its health consequences, particularly across developmental stages.

- Lack of additional physiological and behavioral measures (e.g., objective physical activity, organ size, resting expenditure) limits mechanistic interpretation of TEE variability and repeatability, especially in children. - Short observational intervals (weeks to months for adult analyses) may not capture small cumulative effects of TEE on weight over longer timeframes; a trend in the >4-week subset suggests possible associations over longer periods that could not be confirmed. - Potential measurement error in FFM/FM estimation (e.g., hydration variability) can inflate calculated rates of change over short intervals; authors mitigated this by analyzing a >4-week subset but residual error may remain. - No repeated measurements for ages 7–20 years limit inferences about developmental changes in TEE repeatability across childhood/adolescence. - Sample restricted to healthy, non-athlete participants from multiple studies; generalizability to clinical populations or athletes may be limited.