Article | Published:

Epidemiology and Population Health

# Associations of changes in BMI and body fat percentage with demographic and socioeconomic factors: the ELANA middle school cohort

## Abstract

### Background and objectives

Adolescent weight and fat gain is determined by multiple factors. This study examined the association between changes in body fat indicators, excessive weight and fat gain, and sociodemographic factors among Brazilian adolescents over a 4-year period.

### Methods

Body mass index (BMI) and body fat percentage (BFP) of 809 middle school students (mean age: 11.8 ± 1.15 years) were evaluated annually, from 2010 to 2013. Linear mixed effects models were used to assess the trajectories of BMI and BFP in both boys and girls according to the type of school attended (public or private) and skin colour. General estimating equations logistic regression analyses were performed to investigate associations between sociodemographic variables and the development of overweight or high BFP.

### Results

Girls from private schools (p = 0.003) and white boys (p = 0.041) experienced bigger increases in BMI, as compared to girls from public schools and black/brown boys, respectively. White boys also had an increased chance of presenting excessive weight (OR = 3.28; CI 95%: 1.13–9.52) and BFP (OR = 3.32; CI 95%: 1.38–8.01) gain than black/brown boys. Conversely, white girls were less likely to present excessive body fat gain when compared to black/brown girls (OR = 0.42; CI 95%: 0.18–0.96).

### Conclusions

Adolescents who experienced better socioeconomic conditions, especially boys, were more likely to have greater increases in body fat indicators. Our findings contribute to the better understanding of BMI trajectories and body composition changes during puberty, as well as demonstrates the relationship between socioeconomic variables and adiposity indicators among adolescents in middle-income countries.

## Introduction

The prevalence of overweight and obesity among adolescents worldwide is high. According to the 2013/2014 Health Behaviour in School-aged Children International Report survey, approximately 20% of European adolescents were overweight, with boys having a significantly higher risk [1]. From 1980 to 2013, the worldwide prevalence of childhood overweight and obesity has risen by 47% [2] however, recent studies have indicated that this trend has since stabilised [3] and even decreased [4] in some developed countries. In Brazil, approximately 21.7% of boys and 19.4% of girls between the ages of 10 and 19 years presented overweight or obesity in 2008 [5], and 23.7% of students aged 13 to 17 years presented overweight or obesity in 2015 [6]. Adolescent obesity has raised concern worldwide because studies have shown that excessive weight and body fat are related to cardiovascular risk factors both at this age [7] and later in life [8, 9].

Socioeconomic status (SES) is associated with overweight among children and adults [10], with high-income countries tending to have a strong inverse association between SES and overweight [11] and middle- and low-income countries tending to have a direct association [12]. In Brazil, overweight is more prevalent among adolescents with a higher SES [5], although the prevalence of overweight increases at a faster rate among lower-income groups [13]. To investigate the relationship between risk factors and obesity and body composition changes among adolescents, the growth processes of this age group should be monitored [14, 15]. However, few longitudinal studies have been conducted in low- and middle-income countries because of the complexity and cost of those investigations.

In Brazil, some cohort studies have been conducted that evaluated anthropometric or body composition changes during adolescence [16,17,18,19]. Yet, none of these studies performed repeated evaluations after short time intervals. The Adolescent Nutritional Assessment Longitudinal Study (ELANA) was conducted in the metropolitan area of Rio de Janeiro and followed a cohort of high school adolescents and a cohort of middle school adolescents for three and 4 consecutive years, respectively. Results from the high school ELANA cohort showed that boys of higher SES had greater BMI increases than those of lower SES and all girls [20]. However, the most evident body changes as a result of physiological development occur during early adolescence, which is represented by the middle school cohort, and many factors can affect weight and fat gain during this phase. Therefore, the aim of this study was to examine changes in body mass index (BMI) and body fat percentage (BFP) for 4 consecutive years during early adolescence and to elucidate the association between excessive weight and fat gain and demographic and socioeconomic variables among the middle school ELANA cohort.

## Subjects and methods

### Study design and population

The ELANA was a longitudinal study that followed adolescents from two cohorts, a middle school cohort (attended the sixth grade in 2010 and was evaluated until 2013; mean age: 11.8-years-old—early adolescence) and a high school cohort (attended the first year of high school in 2010 and was evaluated until 2012; mean age: 15.7-years-old—intermediate or late adolescence) [20]. Both cohorts were followed at four private and two public schools located in the metropolitan region of Rio de Janeiro, Brazil. The present study considered data from the middle school cohort. All students who were included in the ELANA middle school cohort were enroled in this study, having met the eligibility criteria at baseline (not having a physical or mental condition that prevented completing the questionnaire and assessment in anthropometric measures, and not being pregnant or lactating at baseline).

Sample size was calculated based on 80% power, a significance level of 5%, assuming an unbalance of 50% between groups and a correlation of BMI among repeated measurements of 0.90, in order to detect a difference of one unit in BMI at the end of the study [21]. For each of the four proposed strata (sex and type of school attended), 147 students would need to be included. Therefore, the whole sample size for the cohort was estimated to be 588 adolescents.

The ELANA was approved by the Ethics Committee in Research of the Institute of Social Medicine of the State University of Rio de Janeiro (certificate number 0020.0.259.000-09). Participation in the study was voluntary and written informed consent was obtained from the adolescents’ parents.

### Data collection

All data collection were repeated annually and followed the same order of schools taken at baseline with the objective of having a 12-month interval between evaluations. Anthropometric measurements were performed by trained research assistants and field supervisors performed the application of questionnaires. They explained to the students the procedure of completing the questionnaires, solved their doubts, and checked each questionnaire, in order to detect errors of incomplete information.

### Anthropometric and body composition data

All students were evaluated during the time of physical education classes, and wearing the proper uniform, which consisted of light clothing. Anthropometric data were collected according to Lohman protocols [22] and standardised by Habicht techniques [23] at baseline and Norton and Olds techniques [24] in further phases of the study. Body weight was obtained using a portable electronic scale (Kratos®) with a maximum capacity of 180 kg and variation of 50 g. Considering the high precision of the scale, weight measurements were performed only once. Height was measured using a portable stadiometer (Alturexata®), with a range of 0–213 cm and a variation of 0.1 cm. Two height measurements were taken with a maximum variation of 0.5 cm between both. The mean of the measurements was used for the analyses.

Usually 1 week prior to the assessment of body composition, performed by electrical bioimpedance (BIA), students were advised to follow a protocol, which included having taken in enough fluids, not drinking alcohol or coffee and not taking diuretic and laxative medicines the day before, and also not practicing physical activity eight hours before the test. BIA was performed using the body fat analyser RJL System, by means of four electrodes placed on two hands and two feet. From the resistance value obtained from the BIA, a specific age- and sex-equation was used to estimate body fat free mass [25]. Then, fat mass was obtained by finding the difference between body mass and body fat free mass. BFP was estimated by the following equation: $$\frac{{{\mathrm{Body}}\,{\mathrm{weight}}\,\left( {{\mathrm{kg}}} \right) - {\mathrm{fat}}\,{\mathrm{free}}\,{\mathrm{mass}}\,({\mathrm{kg}})}}{{{\mathrm{Body}}\,{\mathrm{weight}}\,({\mathrm{kg}})}} \times 100$$.

### Sexual maturation, socioeconomic, and demographic data

Age, sex, type of school attended, and skin colour were obtained by a self-report questionnaire. Respondents were asked to choose an option for their skin colour among white, black, yellow, brown or indigenous, according to the Brazilian Institute of Geography and Statistics. The type of school attended was used as a proxy of SES, since there is a strong relation between family income and studying in private schools in Brazil [26]. Sexual maturation stage was investigated using the self-evaluation technique validated by Saito [27], which focused on the development of genitalia for boys and breasts for girls, according to Tanner’s criteria [28]. For both sexes, the prepubertal period was classified when adolescents evaluated themselves on stage 1, the beginning of growth spurts in stages 2 and 3 for boys and stage 2 for girls, peak of growth velocity in stage 4 for boys and 3 for girls, and growth deceleration in stage 5 for boys and stages 4 and 5 for girls [29].

### Data processing and analysis

BMI (kg/m²) was classified into four categories according to the World Health Organisation’s proposed age- and sex-specific cut offs: underweight, normal weight, overweight and obesity [30]. BFP was classified as elevated when values equalled or exceeded 25% for boys and 30% for girls [31]. Continuous variables were tested for normal distribution by the Shapiro–Wilk test. Since variables did not have normal distribution, the Mann–Whitney test was used to compare groups. For categorical variables, frequencies were obtained using the chi-square test to compare independent groups. These descriptive analyses were performed using the Statistical Programme for the Social Sciences software, version 19.0 (SPSS, Chicago, IL). WINPEPI software was used for the partition of the chi-square analysis. Statistical significance was assumed when the p-value was < 0.05.

In the descriptive analysis of socio-demographic variables, pubertal status, anthropometric measures, and body fat percentage, the students who participated in the present study at baseline were compared with those that had been assessed at two or more stages of follow-up to identify possible bias as a result of loss of follow-up. BMI and BFP means, overweight and high BFP frequencies were compared between types of school and skin colour, at baseline, stratified by sex.

In order to assess BMI and body fat trajectories, linear mixed effect models were conducted using the Proc Mixed procedure of the Statistical Analysis System (SAS), version 9.3 (SAS, Institute Inc., Cary, NC). This type of analysis allowed incomplete data tracking and enabled the estimation of fixed effects common to individuals belonging to the same group and the random effects specific to each individual, through structures of variance and covariance, which provided improved quality of settings and more accurate estimates [32]. To assess differences between changes in BMI and BFP over time among adolescents according to the type of school (public or private) and skin colour (white or black/brown), an interaction term that was composed of age (used as a time effect) and the categorical variable of interest (e.g., age⁎type of school) was used. The null hypothesis was that trajectory differences between the two types of school or skin colour categories would be constant over time. For this analysis, we assumed an unstructured variance-covariance pattern [15]. Analyses comparing types of school were adjusted by skin colour, and vice versa. Also, all trajectory analyses were stratified by sex and adjusted by sexual maturation at baseline. In order to achieve better visualisation of growth on the trajectory graphs, time points (2010, 2011, 2012 and 2013) were plotted on the x-axis.

In addition, changes over time of the prevalence of two categorical outcomes, high BFP and overweight or obesity, were evaluated according to skin colour and type of school attended using general estimating equations logistic regression models (PROC GENMOD procedure in SAS 9.3). Similar to the linear mixed model analysis, an interaction term between age and the specific sociodemographic characteristic was applied, and the relative risk for presenting high BFP, overweight, or obesity was obtained. These analyses were also stratified by sex and estimates were obtained from models adjusted by sexual maturation on baseline and skin colour (when comparing types of school) or the type of school (for the skin colour analyses).

Also, analyses of excessive gain in BMI and BFP were performed in the following three steps: (1) we assessed the conditional relative gain in each of these variables [33] by regressing current z-score measures on all previous z-score measures and sexual maturation from 2010 to 2013; (2) residual values obtained in the regression were then distributed in z-scores and values > 1 z-score were classified as excessive gain; and (3) logistic regression analysis, stratified by sex, was performed to compare the likelihood of adolescents of different types of school and skin colours to experience excessive weight and body fat gain. However, this analysis was restricted only to those students who underwent measurements in 2010 and 2013 (n = 399 [49.3%] for BMI and n = 398 [49.2%] for BFP).

## Results

At baseline (2010, T0), 809 of the 945 eligible students participated in the data collection. Anthropometric measurements were completed for 792 of these students, and bioimpedance data were collected for 787 of these students. In 2011 (T1), 642 of the students had BMI data and 641 of the students had BFP data. In 2012 (T2), BMI data was available for 537 of the students and BFP data was available for 535 of the students. In the last year (2013, T3), BMI data was available for 481 of the students and BFP data was available for 480 of the students. For all follow-up, most of the participants were males (≈53%) and studied in private schools (ranging from 63 to 70%). Longitudinal analysis included all 803 adolescents who had at least one BMI measurement and the 800 adolescents who had at least one BFP measurement. The flowchart of participants during the study is shown in Fig. 1.

In the comparison of sociodemographic, anthropometric, and body composition data at baseline of adolescents who attended only the first phase of the ELANA and who participated in two or more data collection processes (Table 1), a higher proportion of public school students was observed among participants only at baseline than those with two or more repeated measures (48.0% and 35.1%, respectively, p = 0.006). Similarly, the age median was higher among baseline-only participants, as compared to those in two or more data collection points (11.9 years and 11.4 years, respectively, p < 0.001).

In regards to girls, there were no differences between the mean baseline BMI and BFP according to skin colour or the type of school attended. Similarly, frequencies of overweight or obesity and high BFP did not differ according to skin colour or type of school. In regards to boys, both mean baseline BMI (20.6 and 19.7, respectively, p = 0.026) and BFP (23.8% and 19.1%, respectively, p < 0.001) were higher among those attending private schools than those attending public schools, and baseline BFP was also higher among white than black/brown boys (23.3% and 21.0%, respectively, p = 0.002). Frequency of overweight or obesity was also higher among white than among black/brown boys (50.3% and 40.4%, respectively, p = 0.043) and among private school students than public school students (54.1% and 28.9%, respectively, p < 0.001). The frequency of high BFP was higher in white boys (43.7% and 30.4%, respectively, p = 0.005) and those attending private schools (46.2% and 19.6%, respectively, p < 0.001) – Supplementary table 1.

Results on BMI and BFP changes after 4 years are shown in Fig. 2 and Supplementary Table 2. Significantly higher increases in BMI units were observed in girls attending private schools (p = 0.003) and in white boys (p = 0.041). BFP increase was observed in girls and a decrease in boys, but the trajectories did not differ according to skin colour or type of school attended in both sexes.

Temporal changes in the prevalence of overweight or obesity and high BFP according to skin colour and type of school attended are shown in Fig. 3 and Supplementary Table 3. Crude and adjusted relative risks indicated that white boys had a lower risk for developing overweight or obesity at the end of the follow-up than that of the black/brown boys (p = 0.008).

When analysing the associations between sociodemographic factors and excessive gains of BMI and BFP at the end of the 4 years, white boys were observed to have an increased chance of presenting excessive weight (OR = 3.28; CI 95%: 1.13–9.52) and BFP (OR = 3.32; CI 95%: 1.38–8.01) gain than black/brown boys. After adjusting by type of school attended, white girls were less likely to present excessive body fat gain when compared to black/brown girls (OR = 0.42; CI 95%: 0.18–0.96) (Fig. 4 and Supplementary Table 4).

## Discussion

This study found that over the 4 years, white boys participants of middle school cohort had a higher increase in BMI and higher risk of having an excessive gain of BMI and BFP, as compared to black/brown boys. In addition, white boys had a 15% less risk of developing overweight or obesity throughout the study. This seemingly conflicting result (greater chance of gaining excessive weight but lesser chance of presenting overweight or obesity) for white boys probably occurred because they already presented higher prevalence of overweight or obesity than black/brown boys at baseline. The girls attending private schools had greater increases in BMI from 2010 to 2013, as compared to the girls from public schools; however, white girls had a lower risk of presenting excessive gain in body fat than black/brown girls.

The associations of socioeconomic and demographic factors with BMI trajectory and excessive weight gain in the ELANA high school cohort were evaluated in another study [20]. A higher increase in BMI was also observed in white boys (p = 0.04), as compared to black/brown boys and all girls, and in boys attending private schools (p = 0.01), as compared to those in public schools and all girls. Boys in private schools also had a higher risk (OR = 2.27, CI 95%: 1.06–4.85) of presenting excessive weight gain when compared to boys from public schools. The findings of both ELANA cohorts suggest that, in the context of the metropolitan area of Rio de Janeiro, adolescents that experience better socioeconomic conditions, especially boys, are more likely to have greater increases in adiposity indicators.

Other longitudinal studies assessing BMI that were conducted in Brazil have shown similar results. A cohort study conducted in the Midwestern region of Brazil followed 2 405 children from 0- to 5-years-old at baseline for 11 years and found that children from higher socioeconomic positions tended to have both higher BMI-for-age z-scores and obesity prevalence during adolescence. In the same way, a greater increase in BMI was observed for boys in higher socioeconomic positions than those from middle and low socioeconomic positions, while girls in the low socioeconomic position group showed lower increases in BMI, as compared to others [19]. Similarly, a population-based longitudinal study of 255 children living in the Brazilian Amazon who were evaluated three times between 2003 and 2009 found that higher household wealth was related to greater BMI-for-age z-scores at and after the age of 7 years for both sexes [34].

The relationship between BMI and socioeconomic conditions observed in Brazil differs from what has been observed in high-income countries, where adolescents from lower socioeconomic levels are more likely to have greater increases in adiposity indicators [35, 36]. Racial and ethnic disparities in body weight and adiposity have been studied for some years, mainly in the United States of America, and seem to occur as a result of differences in diet, physical activity, access to medical care, social environment, and various biological factors, such as genetics and developmental perspectives [37].

In a 5-year cohort study conducted in the boroughs of south London, 5863 adolescents aged 11–12 years at baseline, black and mixed brown girls were significantly more likely to present overweight or obesity than white or Asian girls, even after adjustment for socioeconomic status. Although mean waist circumference was also larger in black and mixed brown girls in all 5 years, the mean annual rate of increase was lower in these ethnic groups [38]. In the present study, black/brown girls were also at an increased risk for having excessive fat gain over 4 years. However, in the context of Brazil, race and ethnicity are a social construct and have also been used as indicators of socioeconomic conditions. Therefore, it was expected that results of the skin colour analysis followed the same direction of what was observed by type of school (used as a proxy of SES). Although the associations between type of school and skin colour trended in the same direction for boys, the association was not as clear for girls, with private school girls having a higher BMI increase than those in public schools and white girls being less likely to experience excessive body fat gain. Nevertheless, this association happened only after including type of school attended in the analysis, suggesting that this variable acts as a negative confounder in the relation between skin colour and SES among girls in the present sample [39].

One possible explanation for a higher SES leading to higher risk of excessive weight gain could be related to the quality and quantity of food served in private schools. Data of the most recent National School Health Survey (2015) indicated that Brazilian ninth graders from public schools were more likely to have lunch or dinner with parents (74.8% and 69.6%, respectively) and eat breakfast at least 5 days per week (64.9% and 61.6%, respectively) than those from private schools. In addition, students from private schools had more access to snack bars at schools (92.0% and 54.0%, respectively) and spent over 3 h more per week on sedentary activities, such as watching TV or using videogames (65.2% and 54.5%, respectively), than adolescents in public schools [6].

A limitation of this study was that, in 2013 (T3), only 60% of the original sample that participated in 2010 was able to be evaluated, which corresponds to a loss to follow-up of 40%. Losses to follow-up were higher among public school students, as has been observed that school dropout or migration to “youth and adult education” rates were two to nine times higher among public school students in Rio de Janeiro compared to students in private schools from 2010 to 2013 [40]. Thus, logistic regression analyses considering excessive gains in body fat indicators, which were applied only for those participating both at baseline and at the end of follow-up, may have been affected. However, in the trajectory analysis, mixed effect models were used to take into account missing data and differences between the amounts of measurements in each subject, making it possible to analyse data of all 809 participants. Therefore, it was possible to identify that girls attending private schools and white boys had greater increases in BMI over time, which was similar to what has been observed for excessive gain analyses, especially for boys. This indicates that losses to follow-up did not affect the results.

Strength of this study was that the ELANA was the first study in Brazil to propose annual evaluations of two cohorts within a short follow-up time, including the whole range of adolescence. We conclude that, in general, adolescents from higher socioeconomic conditions, especially boys, are more vulnerable to higher gains of BMI and body fat. Our findings contribute to the better understanding of BMI trajectories and body composition changes during puberty, as well as demonstrates the relationship between socioeconomic variables and adiposity indicators among adolescents in middle-income countries.

## References

1. 1.

World Health Organization. Growing up unequal: gender and socioeconomic differences in young people’s health and well-being. In: Inchley J, Currie D, Young T, Samdal O, Torsheim T, Augustson L, et al., editors. Health Behaviour in School-aged Children (HBSC) study: International report from the 2013/2014 survey. Copenhagen: WHO Regional Office for Europe; 2016. p. 292.

2. 2.

Ng M, Fleming T, Robinson M, Thomson B, Graetz N, Margono, et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980–2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet. 2014;384:766–81.

3. 3.

Miqueleiz E, Lostao L, Regidor E. Stabilisation of the trend in prevalence of childhood overweight and obesity in Spain: 2001–11. Eur J Public Health. 2016;26:960–3.

4. 4.

Keß A, Spielau U, Beger C, Gausche R, Vogel M, Lipek T, et al. Further stabilization and even decrease in the prevalence rates of overweight and obesity in German children and adolescents from 2005 to 2015: a cross-sectional and trend analysis. Public Health Nutr. 2017; 20: 3075–83.

5. 5.

Instituto Brasileiro de Geografia e Estatística - IBGE. Pesquisa de Orçamentos Familiares 2008-2009: Antropometria e Estado Nutricional de Crianças, Adolescentes e Adultos no Brasil. Rio de Janeiro: Ministério do Planejamento, Orçamento e Gestão; 2010, 130 pages.

6. 6.

Instituto Brasileiro de Geografia e Estatística - IBGE. Pesquisa Nacional de Saúde do Escolar (PeNSE): 2015.  Rio de Janeiro: Ministério do Planejamento, Orçamento e Gestão; 2016, 132 pages.

7. 7.

Zhao M, Bovet P, Ma C, Xi B. Performance of different adiposity measures for predicting cardiovascular risk in adolescents. Sci Rep. 2017;7:43686.

8. 8.

Bjerregaard LG, Jensen BW, Ängquist L, Osler M, Sørensen TIA, Baker JL. Change in overweight from childhood to early adulthood and risk of type 2 diabetes. N Engl J Med. 2018;378:1302–12.

9. 9.

Llewellyn A, Simmonds M, Owen CG, Woolacott N. Childhood obesity as a predictor of morbidity in adulthood: a systematic review and meta-analysis. Obes Rev. 2016;17:56–67.

10. 10.

Bambra CL, Hillier FC, Cairns J-M, Kasim A, Moore HJ, Summerbell CD. How effective are interventions at reducing socioeconomic inequalities in obesity among children and adults? Two systematic reviews. Public Heal Res. 2015;3:1–446.

11. 11.

Chung A, Backholer K, Wong E, Palermo C, Keating C, Peeters A. Trends in child and adolescent obesity prevalence in economically advanced countries according to socioeconomic position: a systematic review. Obes Rev. 2015;17:276–95.

12. 12.

Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. Obesity and socioeconomic status in developing countries: a systematic review. Obes Rev. 2012;13:1067–79.

13. 13.

Lima NP, Horta BL, Motta JVDS, Valença MS, Oliveira V, Santos TV Dos, et al. [Evolution of overweight and obesity into adulthood, Pelotas, Rio Grande do Sul State, Brazil, 1982-2012]. Cad saúde pública. 2015; 31: 2017–25.

14. 14.

Loomba-Albrecht LA, Styne DM. Effect of puberty on body composition. Curr Opin Endocrinol Diabetes Obes. 2009;16:10–15.

15. 15.

Fitzmaurice GM, Laird NM, Ware JH. Longitudinal and clustered data. In: Fitzmaurice GM, Laird NM, Ware JH, editors. Applied Longitudinal Analysis. Hoboken, New Jersey: John Wiley and Sons; 2011, p. 1–18.

16. 16.

González DA, Nazmi A, Victora CG. Growth from birth to adulthood and abdominal obesity in a Brazilian birth cohort. Int J Obes (Lond). 2010;34:195–202.

17. 17.

Victora CG, Sibbritt D, Horta BL, Lima RC, ScD TC, Wells J. Weight gain in childhood and body composition at 18 years of age in Brazilian males. Acta Paediatr. 2007;96:296–300.

18. 18.

Fonseca FL, Brandão AA, Pozzan R, Campana ÉMG, Luiz Pizzi O, Magalhães MEC, et al. Excesso de peso e o risco cardiovascular em jovens seguidos por 17 anos: estudo do Rio de Janeiro. Arq Bras Cardiol. 2010;94:207–15.

19. 19.

Muraro AP, Gonçalves-Silva RMV, Ferreira MG, Sichieri R. Effects of social mobility from childhood to adolescence on BMI. Public Health Nutr. 2016;19:814–21.

20. 20.

Moreira NF, Sichieri R, Reichenheim ME, Oliveira ASDDe, Da Veiga GV. The associations of BMI trajectory and excessive weight gain with demographic and socio-economic factors: the Adolescent Nutritional Assessment Longitudinal Study cohort. Br J Nutr. 2015;114:2032–8.

21. 21.

Twisk JWR. Applied Longitudinal Data Analysis for Epidemiology. Cambridge: Cambridge University Press; 2003. https://doi.org/10.1017/CBO9781139342834.

22. 22.

Lohman TG, Martorell R. Anthropometric standardization reference manual. Human Kinetics Books: Champaign, IL; 1988.

23. 23.

Habicht J-P. Estandarizacion de metodos epidemiológicos cuantitativos sobre el terreno. Bol Sanit PAnam. 1974;75:375–84.

24. 24.

Pederson D, Gore C. Error en la medición antropométrica. In: Norton K, Olds T, editors. Antropometrica. Argentina: Biosystem Servicio Educativo; 2000, p. 61–70.

25. 25.

Houtkooper LB, Going SB, Lohman TG, Roche AF, Van Loan M. Bioelectrical impedance estimation of fat-free body mass in children and youth: a cross-validation study. J Appl Physiol. 1992;72:366–73.

26. 26.

​Instituto Brasileiro de Geografia e Estatística - IBGE. Síntese de indicadores sociais: uma análise das condições de vida da população brasileira - 2010. Rio de Janeiro: Ministério do Planejamento, Orçamento e Gestão; 2010, 317 pages.

27. 27.

Saito MI. Maturação sexual: auto avaliação do adolescente. Pediatr (São Paulo). 1984;6:111–5.

28. 28.

Tanner JM. Growth at adolescence. 2nd ed. Oxford: Blackwell Scientific publication; 1962.

29. 29.

Bagni UV, Luiz RR, Veiga GV da. Overweight is associated with low hemoglobin levels in adolescent girls. Obes Res Clin Pract. 2013;7:e218–9.

30. 30.

de Onis M. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007;85:660–7.

31. 31.

Williams DP, Going SB, Lohman TG, Harsha DW, Srinivasan SR, Webber LS, et al. Body fatness and risk for elevated blood pressure, total cholesterol, and serum lipoprotein ratios in children and adolescents. Am J Public Heal. 1992;82:358–63.

32. 32.

Spyrides MHC, Struchiner CJ, Barbosa MTS, Kac G. Análise de dados com medidas repetidas. In: Kac G, Sichieri R, Gigante DP, editors. Epidemiologia Nutricional. Rio de Janeiro: Fiocruz e Atheneu; 2007, p. 245–60.

33. 33.

Adair LS, Fall CH, Osmond C, Stein AD, Martorell R, Ramirez-Zea M, et al. Associations of linear growth and relative weight gain during early life with adult health and human capital in countries of low and middle income: findings from five birth cohort studies. Lancet. 2013;382:525–34.

34. 34.

Lourenço BH, Villamor E, Augusto RA, Cardoso MA. Influence of early life factors on body mass index trajectory during childhood: a population-based longitudinal analysis in the Western Brazilian Amazon. Matern Child Nutr. 2015;11:240–52.

35. 35.

Lane SP, Bluestone C, Burke CT. Trajectories of BMI from early childhood through early adolescence: SES and psychosocial predictors. Br J Health Psychol. 2013;18:66–82.

36. 36.

Shin SH, Miller DP. A longitudinal examination of childhood maltreatment and adolescent obesity: results from the National Longitudinal Study of Adolescent Health (AddHealth) Study. Child Abus Negl. 2012;36:84–94.

37. 37.

Krueger PM, Reither EN. Mind the gap: Race/ethnic and socioeconomic disparities in obesity. Curr Diab Rep. 2015;15:95.

38. 38.

Wardle J, Brodersen NH, Cole TJ, Jarvis MJ, Boniface DR. Development of adiposity in adolescence: five year longitudinal study of an ethnically and socioeconomically diverse sample of young people in Britain. BMJ. 2006;332:1130–5.

39. 39.

Mehio-Sibai A, Feinleib M, Sibai TA, Armenian HK. A positive or a negative confounding variable? a simple teaching aid for clinicians and students. Ann Epidemiol. 2005;15:421–3.

40. 40.

Ministério da Educação-Brasil, Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira. Indicadores Educacionais. http://portal.inep.gov.br/web/guest/indicadores-educacionais (accessed 18 Jan 2018).

## Acknowledgements

This study was funded by the National Council for Scientific and Technological Development (grant 47667/2011-9), the Research Support Foundation of the State of Rio de Janeiro (grants E26/ 110•847/2009, E26/110•626/2011 and E-26/110.774/2013) and Coordination for the Improvement of Higher Education Personnel (grant 23038.007702/2011-5).

### Funding

The Adolescent Nutritional Assessment Longitudinal Study (ELANA) was funded by the National Council for Scientific and Technological Development (grant 47667/2011-9), the Research Support Foundation of the State of Rio de Janeiro (grants E26/ 110.847/2009, E26/110.626/2011 and E-26/110.774/2013) and Coordination for the Improvement of Higher Education Personnel (grant 23038.007702/2011-5).

## Author information

### Affiliations

1. #### Department of Social and Applied Nutrition, Institute of Nutrition Josué de Castro, Federal University of Rio de Janeiro, Rio de Janeiro, 21941-901, Brazil

• Milena Miranda de Moraes
•  & Gloria Valeria da Veiga
2. #### Faculty of Health Sciences, Federal University of Grande Dourados, Dourados, 79804-970, Brazil

• Naiara Ferraz Moreira
3. #### Department of Social Nutrition, Nutrition Institute, State University of Rio de Janeiro, Rio de Janeiro, 20550-900, Brazil

• Alessandra Silva Dias de Oliveira
4. #### Department of Epidemiology, Institute of Social Medicine, State University of Rio de Janeiro, Rio de Janeiro, 20550-900, Brazil

• Diana Barbosa Cunha
•  & Rosely Sichieri

### Conflict of interest

The authors declare that they have no conflict of interest.

### Corresponding author

Correspondence to Milena Miranda de Moraes.