Sample
The cohort study entitled "Analysis of Behaviors of Children During Growth" (ABCD – Growth Study) is an initiative designed to identify the impact of physical activity/sports participation on different health aspects of adolescents. The ABCD – Growth Study has been carried out in Presidente Prudente since 2017 (~ 200,000 inhabitants and human development index 0.806; western state of São Paulo, Brazil) by researchers of the Laboratory of Investigation in Exercise (LIVE) from Sao Paulo State University – UNESP, the campus of Presidente Prudente. The ethics committee of UNESP approved the study (process number 1.677.938/2016), and parents/guardians and adolescents signed the written consent form.
More details about the sampling process can be found elsewhere. [18–21] In brief, at baseline, researchers contacted school units and sports clubs spread across the metropolitan region of the city to request authorization to contact adolescents; 11 of the units contacted granted access to the adolescents. In the first contact with the adolescents, researchers explained all the aims and inclusion criteria of the cohort study and written consent forms were delivered to those ones who reported fulfilling all inclusion criteria. The inclusion criteria adopted were: 1) 11–18 years old, 2) parents' consent form signed, 3) if contacted in a sports club, at least one year of training experience in order to characterize consistent engagement; if contacted in a school unit, at least one year without the regular practice of sport or exercise; and 4) absence of orthopedic disease, limiting physical activity.
In the second contact, the researchers collected the signed consent forms. The researchers contacted only adolescents who returned the signed consent form, by phone, to schedule interviews at the university facilities. Baseline data collection was performed at LIVE in 2017, and follow-up data collection in 2018. In 2017, 285 adolescents started the cohort study and the researchers followed them during the entire year. Due to exclusions (n = 95 [did not wish to take part, did not participate in all measurements, and moved to another city]), in 2018 the measurements involved 190 adolescents (Fig. 1).
Body Composition Aspects
Bone mineral density (BMD, g/cm²), lean soft tissue (LST, kg), and body fatness (percentage values [%]) were measured at the university laboratory in a temperature-controlled room using dual-energy x-ray absorptiometry (Lunar DPX-NT; General Electric Healthcare, Little Chalfont, Buckinghamshire, UK) with GE Medical System Lunar software (version 4.7). A trained researcher performed all scans and tested the scanner quality before the first exam of each day. The coefficient of variation for this device was 0.66% (in whole-body BMD analysis, n = 30 participants not involved in this study), defining the regions of interest (ROIs) as the extremities (upper limbs, lower limbs, and spine) as suggested by the General Electric Healthcare company and reported in previous studies. [22–24] The scans were performed using a standardized protocol with the participants remaining in the supine position and wearing only light clothing. Regional analysis for BMD of upper limbs, lower limbs, spine, and whole body (less the head) occurred off-line after the scans had been performed.
Absorptiometry scans were performed twice in this study (baseline and follow-up), and the outcomes considered were the absolute change (subtraction of baseline values from follow-up values) in LST (whole-body and upper limbs), body fatness (whole-body), and BMD (upper limbs, lower limbs, spine, and whole-body less head).
Resistance Training
The RT (strength training, weight-bearing, circuit-based exercise, etc.) engagement was assessed by face-to-face interview at both baseline and follow-up. Researchers interviewed the adolescents about their current engagement in RT (yes or no), previous time of engagement (months), and weekly frequency (days). Considering the responses, three groups were created for this manuscript. The first, denominated “non-engagement” (n = 121 [boys n = 90 and girls n = 31]) included adolescents who did not report any engagement in RT at either collection moment. The second group was denominated “Irregular engagement” (n = 44 [boys n = 26 and girls n = 18]), composed of adolescents who reported RT practice at only one of the moments, either baseline or the 12-month follow-up (n = 30 quit RT between baseline and follow-up and n = 14 started RT between baseline and follow-up). The third group, denominated “Frequent engagement” (n = 25 [boys n = 14 and girls n = 11]), was composed of adolescents who reported RT practice at both collection moments.
Covariates
Sex (boys/girls), chronological age (years), supplementation use, skipping breakfast, previous time of engagement in RT, and somatic maturation were the main covariates considered in this manuscript. Sex, chronological age, previous time of engagement in RT (in months), skipping breakfast (number of days per week skipping the meal [categorized as no days and at least one day]), and use of any nutritional supplement (categorized as none, weight loss, and muscle mass gain) were assessed at baseline during the face-to-face interview. Somatic maturation was estimated by the peak of height velocity (PHV), through mathematical models based on anthropometric measurements proposed by Mirdwald et al.[25] The equations indicate the time in years to reach maximum height.
Baseline values of the outcome were used as covariates due to their impact on the changes over time. In the multivariate models considering BMD outcomes (whole-body, spine, upper and lower limbs), the LST at baseline of the respective body segment was considered an additional covariate.
Statistical analysis
Descriptive statistics are composed of mean values, standard-deviation values, and 95% confidence intervals (95%CI). The Pearson correlation was used to assess the relationship between RT parameters (previous time of practice and days training per week) and changes in body composition aspects. Dependent variables were treated as absolute changes (Δ), while comparisons according to the different patterns of engagement in RT were performed through analysis of variance (ANOVA). Mean comparisons that were statistically significant in the ANOVA were rerun on multivariate models based on analysis of covariance (ANCOVA), controlled by covariates (post-hoc tests were Tukey and Bonferroni, respectively). The multivariate models considered the overall sample, and stratified by sex. Levene’s test identified goodness-of-fit parameters in the multivariate ANCOVA models. All data analysis was conducted using the statistical software BioEstat (version 5.0) and the significance level (p-value) was set at < 0.05.