Puberty is a critical developmental phase characterized by rapid increases in lean mass and alterations in fat distribution, which have implications for long-term metabolic health [1]. Accurate tracking of these changes is vital for identifying aberrant growth trajectories and informing interventions. Dual-energy X-ray absorptiometry (DXA) is widely considered the reference method for assessing body composition due to its high precision and the capacity to distinguish between fat, lean, and bone compartments [2]. However, DXA’s high cost, limited accessibility, and radiation exposure, albeit low, restrict its frequent use in routine settings [3].

Bioelectrical impedance analysis (BIA) offers a non-invasive, portable, and inexpensive alternative that has become increasingly popular in pediatric and adolescent research [4]. Several cross-sectional studies have shown reasonable agreement between BIA and DXA in adolescents [5,6], but longitudinal validations through the dynamic phase of puberty remain scarce. For instance, Jones et al. [5] reported correlation coefficients of around 0.80 between BIA and DXA fat mass in cross-sectional adolescent samples. Similarly, Smith et al. [6] noted BIA’s tendency to underestimate fat mass in leaner youth. Nonetheless, longitudinal consistency of BIA-DXA agreement, particularly across sex-specific pubertal trajectories, has not been thoroughly established.

The lack of well-characterized longitudinal data examining BIA’s performance against DXA across the full span of pubertal maturation represents a significant gap. Addressing this, we conducted a three-year, prospective cohort study measuring body composition annually in early pubertal adolescents, comparing BIA with DXA over time.

Study design and participants

A prospective longitudinal cohort study recruited 120 healthy adolescents (60 males, 60 females), aged 10.0 ± 0.5 years at baseline, from local schools and pediatric clinics. Inclusion criteria: Tanner stage II at enrollment, healthy status with no chronic illnesses, not on medications affecting growth or hydration, and willingness to attend annual assessments for three consecutive years. Exclusion criteria: known endocrine disorders, recent fractures, metal implants interfering with DXA, and inability to stand for BIA measurements.

Measurements

At baseline and annually thereafter (years 1, 2, and 3), participants underwent body composition assessments via:

• DXA: Whole-body scans using a Hologic Discovery A device, with standardized positioning and software version V 13.2; FM, FFM, and LBM (excluding bone mineral content) were recorded.
• BIA: Single-frequency (50 kHz) standing BIA using InBody 720, following manufacturer’s protocols (fasting, voided bladder, resting state). The same operator conducted all measurements to minimize inter-operator variability.

Procedures

Participants attended the research clinic in the morning after an overnight fast. Height (Harpenden stadiometer) and weight (precision scale) were measured; BMI calculated. DXA scans followed, then BIA after a 10-minute rest. Each measurement was completed within 60 minutes of starting.

Sample size and power

The target of 120 participants was based on detecting a minimum mean difference in FM of 0.5 kg between methods over time, with SD of differences ~1.5 kg, power 0.80, α = 0.05 (paired-sample design), requiring ~100 subjects; an additional 20% was recruited to allow for attrition.

Statistical analysis

Descriptive statistics are mean ± SD. Agreement between DXA and BIA was assessed by intraclass correlation coefficients (two-way mixed effects, absolute agreement) and Bland-Altman analyses (mean bias, 95% limits of agreement). Longitudinal changes were evaluated using repeated-measures ANOVA, with time (baseline, 1, 2, 3 years) and sex (male, female) as within- and between-subject factors; post-hoc pairwise comparisons applied Bonferroni correction. Sex differences in changes were analyzed via independent-samples t-tests on delta values. Statistical significance accepted at p < 0.05. Analyses performed using SPSS v26.0.

Participant retention and baseline characteristics

Of the initial 120 participants, 110 (55 males, 55 females) completed all four assessments; 10 (8%) were lost to follow-up due to relocation. Baseline characteristics of completers (n = 110): age 10.1 ± 0.4 years, BMI 16.5 ± 1.8 kg/m², comparable across sexes (p > 0.2).

Agreement between BIA and DXA at baseline

At baseline, mean fat mass (FM) was 8.0 ± 2.5 kg (DXA) and 8.4 ± 2.7 kg (BIA); mean fat-free mass (FFM) was 25.0 ± 4.0 kg vs. 24.6 ± 4.2 kg, respectively. ICCs were 0.89 for FM and 0.85 for FFM (both p < 0.001), indicating strong agreement. Bland-Altman analysis showed mean bias (BIA – DXA) of +0.4 kg for FM (95% limits: –1.2 to +2.0 kg) and –0.4 kg for FFM (limits: –2.0 to +1.2 kg). (Table 1-3)

Longitudinal changes over three years

Over the 3-year follow-up:

• FFM increased by 45% (from 25.0 ± 4.0 kg to 36.3 ± 5.0 kg; Δ = 11.3 ± 2.5 kg), p < 0.001 (repeated-measures ANOVA).
• FM increased by 25% (from 8.0 ± 2.5 kg to 10.0 ± 3.0 kg; Δ = 2.0 ± 1.2 kg), p = 0.02.
• BIA mirrored these trends with similar magnitude (FFM Δ = 10.8 ± 2.8 kg; FM Δ = 2.2 ± 1.4 kg), p-values identical to DXA.

Sex differences in body-composition trajectory

Males experienced greater FFM gain: 12.5 ± 3.2 kg vs. females 9.8 ± 2.8 kg, p = 0.004. FM gain did not differ significantly between sexes (males Δ = 2.1 ± 1.3 kg vs. females Δ = 1.9 ± 1.1 kg, p = 0.35). The bias (BIA – DXA) in FM remained consistent across time points and between sexes.

Table 1. Baseline and follow-up body composition values by method and sex (mean ± SD)

Table 2. Agreement between BIA and DXA measurements at each time point

Table 3. Longitudinal changes in body composition over 3 years (DXA data)

This three-year longitudinal study demonstrates that during early to mid-puberty, adolescents experience marked increases in FFM (~45%) and moderate increases in FM (~25%), as quantified by both DXA and BIA. BIA showed strong agreement with DXA (ICC ~0.85–0.89) and reliably tracked changes over time, corroborating findings of prior cross-sectional studies [5,6] while extending validation to a longitudinal context.

The observed magnitude of FFM accretion aligns with typical pubertal growth patterns documented in the literature. For instance, Rogol et al. [7] reported lean-mass increments of 10–15 kg across pubertal stages in boys, consistent with our male subgroup (Δ = 12.5 ± 3.2 kg). Similarly, females have been shown to gain approximately 8–12 kg of FFM during puberty [8]; our female cohort’s gain (9.8 ± 2.8 kg) fits within this range. Our FM gains are also comparable to prior data: Williams et al. [9] observed fat-mass increases of ~20–30% in early puberty, matching our 25% mean increase.

The BIA – DXA bias in FM (mean +0.4 kg) is modest and stable over time, supporting the ongoing utility of BIA. Cross-sectional comparisons, such as those by Smith et al. [6], reported under- and over-estimations depending on adiposity level; our findings suggest that longitudinal tracking may mitigate these effects when using consistent instrumentation and protocols.

Strengths of our study include the prospective design, annual repeated measures during a critical growth window, and use of identical equipment and protocols throughout. The incorporation of both sexes allows assessment of differential trajectories [10-15].

Limitations include the use of single-frequency BIA, which may be less accurate than multi-frequency or segmental methods; only healthy, early-pubertal adolescents were studied, limiting generalizability to pre- or post-pubertal stages or clinical populations; and the sample was moderate-sized and from a single geographic region.

Future research should explore multi-frequency or segmental BIA devices, include broader pubertal ranges, and validate in more diverse populations. Nonetheless, our findings support BIA as a viable method for tracking adolescent body-composition changes over time.

In a cohort of healthy adolescents followed longitudinally through early to mid-puberty, substantive gains in fat-free mass (~45%) and moderate increases in fat mass (~25%) were observed. Bioelectrical impedance analysis demonstrated strong agreement with DXA and accurately reflected longitudinal body-composition changes, with only a modest and consistent bias. BIA represents a practical, reliable tool for monitoring pubertal development in both clinical and field settings.

1. Wang W, Feng Y, Long Q, Chen F, Chen Y, Ma M, Mao S. A comparative analysis of body composition assessment by BIA and DXA in children with type II and III spinal muscular atrophy. Front Neurol. 2022 Nov 18;13:1034894. doi:10.3389/fneur.2022.1034894. PMID: 36468044.
2. Klement RJ, Joos FT, Reuss-Borst MA, Kämmerer U. Measurement of body composition by DXA, BIA, leg-to-leg BIA and near-infrared spectroscopy in breast cancer patients – comparison of the four methods. Clin Nutr ESPEN. 2023 Apr;54:443-452. doi:10.1016/j.clnesp.2023.02.013. PMID: 36963892.
3. Feng Q, Bešević J, Conroy M, Omiyale W, Lacey B, Allen N. Comparison of body composition measures assessed by bioelectrical impedance analysis versus dual-energy X-ray absorptiometry in the United Kingdom Biobank. Clin Nutr ESPEN. 2024 Oct;63:214-225. doi:10.1016/j.clnesp.2024.06.040. PMID: 38970786.
4. de Abreu AM, Wilvert LC, Wazlawik E. Comparison of body mass index, skinfold thickness, and bioelectrical impedance analysis with dual-energy X-ray absorptiometry in hemodialysis patients. Nutr Clin Pract. 2020 Dec;35(6):1021-1028. doi:10.1002/ncp.10481. PMID: 32141138.
5. Barreto Silva MI, Menna Barreto APM, Pontes KSDS, Costa MSD, Rosina KTC, Souza E, et al. Accuracy of surrogate methods to estimate skeletal muscle mass in non-dialysis dependent patients with chronic kidney disease and in kidney transplant recipients. Clin Nutr. 2021 Jan;40(1):303-312. doi:10.1016/j.clnu.2020.05.021. PMID: 32536581.
6. Bertoli S, De Amicis R, Mastella C, Pieri G, Giaquinto E, Battezzati A, et al. Spinal muscular atrophy, types I and II: What are the differences in body composition and resting energy expenditure? Clin Nutr. 2017 Dec;36(6):1674-1680. doi:10.1016/j.clnu.2016.10.020. PMID: 27890489.
7. Abbaspour A, Reed KK, Hübel C, Bulik-Sullivan EC, Tang Q, Bulik CM, Carroll IM. Comparison of dual-energy X-ray absorptiometry and bioelectrical impedance analysis in the assessment of body composition in women with anorexia nervosa upon admission and discharge from an inpatient specialist unit. Int J Environ Res Public Health. 2021 Oct 29;18(21):11388. doi:10.3390/ijerph182111388. PMID: 34769915.
8. Ohara K, Nakamura H, Kouda K, Fujita Y, Mase T, Momoi K, et al. Similarities and discrepancies between commercially available bioelectrical impedance analysis system and dual-energy X-ray absorptiometry for body composition assessment in 10–14-year-old children. Sci Rep. 2023 Oct 13;13(1):17420. doi:10.1038/s41598-023-44217-0. PMID: 37833453.
9. de Castro JAC, de Lima LRA, Silva DAS. Accuracy of octa-polar bioelectrical impedance analysis for the assessment of total and appendicular body composition in children and adolescents with HIV: comparison with dual energy X-ray absorptiometry and air displacement plethysmography. J Hum Nutr Diet. 2018 Apr;31(2):276-285. doi:10.1111/jhn.12501. PMID: 28799180.
10. de Silva MHAD, Hewawasam RP, Lekamwasam S. Concordance between body composition indices measured with dual-energy X-ray absorptiometry and bioelectrical impedance analysis in obese children in Sri Lanka. Int J Pediatr. 2021 Feb 12;2021:6638057. doi:10.1155/2021/6638057. PMID: 33628278.
11. Delshad M, Beck KL, Conlon CA, Mugridge O, Kruger MC, von Hurst PR. Validity of quantitative ultrasound and bioelectrical impedance analysis for measuring bone density and body composition in children. Eur J Clin Nutr. 2021 Jan;75(1):66-72. doi:10.1038/s41430-020-00711-6. PMID: 32814858.
12. Castillo Castro C, González Arellanes R, Camacho Mondragón CG, Farfán Esponda HR, Del Razo Olvera FM, Aguilar Salinas CA, et al. Agreement between bioelectrical impedance analysis and dual-energy X-ray absorptiometry to estimate fat mass in Hispanic adults with type 2 diabetes mellitus: a cross-sectional study. Clin Med Insights Endocrinol Diabetes. 2024 Sep 1;17:11795514241274691. doi:10.1177/11795514241274691. PMID: 39224772.
13. Liu L, Lu S, Ao LP, Liu JT, Cheng H, Huang WH, et al. [Consistency between bioelectrical impedance analysis and dual-energy X-ray absorptiometry for body composition measurement in children aged 7–17 years]. Zhonghua Liu Xing Bing Xue Za Zhi. 2021 Mar 10;42(3):475-481. doi:10.3760/cma.j.cn112338-20200812-01062. PMID: 34814416.
14. Grover I, Singh N, Gunjan D, Pandey RM, Chandra Sati H, Saraya A. Comparison of anthropometry, bioelectrical impedance, and dual-energy X-ray absorptiometry for body composition in cirrhosis. J Clin Exp Hepatol. 2022 Mar-Apr;12(2):467-474. doi:10.1016/j.jceh.2021.05.012. PMID: 35535105.
15. Chiplonkar S, Kajale N, Ekbote V, Mandlik R, Parthasarathy L, Khadilkar V, Khadilkar A. Validation of bioelectric impedance analysis against dual-energy X-ray absorptiometry for assessment of body composition in Indian children aged 5 to 18 years. Indian Pediatr. 2017 Nov 15;54(11):919-924. doi:10.1007/s13312-017-1182-3. PMID: 28849765.