Association between risk behaviors and adiposity indicators in adolescents from Southern Brazil

Abstract

The aim of this study was to analyze the association between risk behaviors and adiposity indicators in adolescents and to discuss some methodological aspects related to this relationship. We evaluated 1,321 adolescents (55.2% female) aged 10–16 years. Relative body fat (%fat) by measurement of triceps and subscapular skinfold thickness and waist circumference (WC) were used as total and central adiposity indicators, respectively. Physical inactivity, time spent in front of the TV, the consumption of soda and/or chocolate, alcohol, and tobacco smoking were analyzed as risk behaviors. Information about the socioeconomic status (categorized into three levels) and nutritional status of the mother (overweight or normal weight) were used as adjustment factors in the analyses of prevalence ratio (PR) of the outcomes and their associated 95% confidence intervals (95% CI). The chi-square test and Poisson regression were used for statistical analyses. Low associations were found between risk behaviors and adiposity indicators. Tobacco smoking was the most positively correlated behavior with adiposity in girls (%fat: PR = 1.61; 95% CI = 1.04–2.47; WC: PR = 1.90; 95% CI = 1.17–3.08) and in adolescents whose mothers were normal weight (%fat: PR = 2.31; 95% CI = 1.33–4.03; WC: PR: 2.31; CI: 1.19–4.46). Additionally, as an important methodological issue, we highlighted the assessment of risk behaviors in adolescents as crucial to producing more robust evidence on the subject. Of the investigated behaviors, we concluded that tobacco smoking is the behavior most associated with adiposity indicators.

Keywords

Adolescent behavior obesity lifestyle risk factors epidemiology

Introduction

Among the major public health problems in childhood and adolescence, excess body weight has received great attention in recent decades. This is due to the high prevalence of excess body weight observed in several countries and its negative health consequences (Ebbeling et al., 2002; Ludwig, 2007; Sikorski et al., 2012). To develop actions to improve this situation, studies have tried to identify the main determinants of excess weight in specific pediatric populations (Assunção et al., 2012).

Issues related to lifestyle and the adoption of risk behaviors at early ages, such as low levels of physical activity, a sedentary lifestyle, the consumption of foods with a high energy density and the use of licit and illicit drugs, have been proposed as causes for the current obesity epidemic (Currie et al., 2008). In Brazil, according to data from the National Survey of School Health (BIGS, 2012), which evaluated students in ninth grade, only 30% of them met their recommendations for physical activity, 79% spent more than two hours/day watching TV, 41% ate sweets regularly, 26% consumed alcohol, and 5% were tobacco smokers.

Although studies conducted in developed countries have indicated a causal effect between specific behaviors and adiposity (Boulos et al., 2012; Hancox et al., 2004; Neumark-Sztainer et al., 2012; Ortega et al., 2007), little evidence is available for middle- and low-income countries (Bozza et al., 2014). In order to bring about a discussion on the topic and highlight the problems of decision making based on weak evidence, we analyzed the association between risk behaviors and adiposity indicators in adolescents from Southern Brazil and discussed some methodological issues related to this relationship.

Methods

Sample

A school-based epidemiological study was conducted with 1,321 adolescents aged 10–16 years from public schools in the city of Londrina, in 2011. Located in the north of the state of Paraná, Brazil, Londrina has 515,707 inhabitants (28,822 school children in the age range of interest), a Human Development Index and gross domestic product of .824 and US$ 4,442,229.50, respectively (BIGS, 2013).

The process of sampling was carried out in two stages. All public schools in the city were listed by geographical region (north, south, east, west, and center), and two schools from each region were then randomly selected. Classes from the selected schools were then randomly ordered and five were chosen. All students who studied in the morning or afternoon periods were invited to participate in this study. Schoolchildren were excluded if they frequently used medication, which made their participation impossible, were under treatment for any illness, or who did not return the terms of consent duly signed by their parents or guardians. The sample size was calculated for expected prevalence of metabolic syndrome = 4%, α = .05, error = 2% points; design effect = 2.0, 20% + for losses/refusals. Initially, the sample was composed of 1,395 adolescents. Of these, 74 did not have complete data for the variables of interest. Thus, the final sample consisted of 1,321 adolescents (Table 1). All procedures in this investigation were approved by the Ethics Committee in Research of the Londrina State University.

Table 1.

General characteristics of the sample.

	Male (n = 593) (95% CI)	Female (n = 728) (95% CI)	Total (N = 1,321) (95% CI)
Age (years)
10–12	29.0 (25.6–32.5)	33.6 (30.3–37.0)	31.5 (29.2–34.0)
12–14	43.4 (39.6–47.3)	43.2 (39.8–46.8)	43.3 (40.7–45.9)
14–16	27.6 (24.2–31.2)	23.2 (20.3–26.3)	25.2 (22.9–27.5)
Socioeconomic status
Highest	35.0 (30.7–39.5)	31.7 (28.1–35.5)	33.1 (30.3–36.0)
Middle	33.9 (29.6–38.4)	32.5 (28.9–36.3)	33.1 (30.3–35.9)
Lowest	31.2 (27.1–35.6)	35.8 (32.1–39.6)	33.8 (31.1–36.7)
Physical inactivity	70.9 (67.2–74.4)	88.2 (85.7–90.3)	80.4 (78.2–82.4)
High TV watching	44.4 (40.5–48.4)	45.5 (41.9–49.1)	45.0 (42.3–47.7)
Inadequate dietary pattern	65.8 (61.2–69.5)	70.1 (66.7–73.3)	68.2 (65.6–70.6)
Alcohol consumption	18.9 (16.0–22.2)	18.3 (15.7–21.3)	18.6 (16.6–20.7)
Tobacco smoking	8.3 (6.4–10.8)	8.3 (6.5–10.6)	8.3 (6.9–9.9)
Maternal overweight	55.8 (50.9–60.6)	55.7 (51.6–59.8)	55.8 (52.6–58.9)
High WC	19.2 (16.3–22.6)	24.0 (21.0–27.2)	21.9 (19.7–24.2)
High % body fat	20.2 (17.2–23.7)	26.1 (23.0–29.4)	23.5 (21.3–25.8)

Note: CI: confidence interval; socioeconomic status: highest = A and B classes; middle = C2 class; lowest = C1, D, and E classes; WC: waist circumference.

Considering a statistical power of 80%, confidence interval (CI) of 95%, and the distribution of subjects in the categories of the independent variables, this sample size allowed us to identify prevalence ratios (PRs) of above 1.42 (risk)/.61 (protection), regardless of risk behavior or gender.

Dependent variables

The relative body fat (%fat) and waist circumference (WC) were adopted as indicators of total and central adiposity, respectively. For this, the thickness of the triceps and subscapular skinfolds was measured by an experienced evaluator using a compass of the Lange® brand, suitably calibrated (accuracy = .5 mm), following the recommendations of Harrison et al. (1988). For quality control, the technical error of measurement (TEM) was calculated from measurements duplicated in 90 adolescents (pilot project), who were not included in the sample of this study, but had similar characteristics with regard to age, race, somatic maturation and level of adiposity. The TEMs were 4.8% and 3.5% for subscapular and triceps skinfold thicknesses, respectively. From this information, the %fat was estimated using the equation proposed by Boileau et al. (1985) and the adolescents were categorized dichotomously according to the cutoff proposed by Williams et al. (1992). Additionally, the WC was measured by a single examiner using a metal tape measure of the Sanny® brand with a precision of one millimeter. The recommendations of Katzmarzyk (2004) were adopted both for the measurement itself and for the categorization of the values of WC (normal/high).

Independent variables

Physical inactivity, amount of time spent watching TV, consumption of sodas and/or chocolate, alcohol, and tobacco use were the risk behaviors for the adiposity indicators in this study. This information was obtained through questionnaires administered by an experienced evaluator and assisted by two assessors trained for this purpose. For quality control of the data, the instruments were reapplied (seven-days interval) to 10% of the sample for calculating the intraclass correlation coefficient (ICC, to continuous and ordinal variables) and κ coefficient (to dichotomous variables).

As an indicator of the level of physical activity, playing sports during leisure time was assessed using the questionnaire developed by Baecke et al. (1982) (ICC = .88). Adolescents who reported having engaged in sports activities of moderate to high intensity for at least 240 minutes/week during the previous four months were considered physically active (Fernandes et al., 2012).

The amount of time spent watching TV was calculated by summing the minutes spent during the weekdays and weekends (ICC = .77). Based on the daily weighted average, sedentary behavior was defined as more than four hours/day in front of the TV.

The consumption of sodas and/or chocolate was evaluated by means of one question (ICC = .76) concerning the frequency of consumption (no consumption, one to three days, four to six days, and seven days) during the previous week. The cutoff used for inadequate dietary pattern was the consumption of these foods on four or more days per week. Information about alcohol consumption (κ = .58) and tobacco smoking (κ = .64) was obtained in a dichotomous manner, by indicating frequent use (at least once a week) in the past 30 days.

Confounding variables

For statistical control, in addition to age and sex, socioeconomic status (SES) and the mother’s body mass index (BMI_mother) were evaluated, by means of a questionnaire filled by the parents or guardians.

While SES was evaluated using the instrument of the Brazilian Association of Research Companies (BARC, 2008), which allows stratification by eight categories (A1, A2, B1, B2, C1, C2, D, and E [lowest]), the BMI_mother was obtained through the request for self-reported body weight and stature.

Statistical analyses

The statistical sample power was estimated by Epi Info program, version 7. Distribution of relative frequencies and their respective 95% CI, according to sex, were used for the characterization of the sample. Chi-square test of independence was used to check the association between risk behaviors and adiposity indicators (bivariate analysis). Poisson regression was used with robust estimate of variance to obtain the adjusted PRs and their 95% CIs. This method has been used in order to not overestimate the effect measure mainly in high-prevalent outcomes (Barros and Hirakata, 2003). PRs were adjusted by chronological age and SES (stratified by gender analysis) and by chronological age, SES, and sex (stratified by BMI_mother). After bivariate analyses, independent variables were included in the adjusted multivariate model. All analyses used SPSS 17.0 with a p value below .05 for statistically significant associations.

Results

Crude associations between risk behaviors and adiposity indicators as well as the adjusted PRs (95% CI) are presented in Table 2. In boys, an association was only observed between physical inactivity and total adiposity, and those with insufficient physical activity presented a higher probability of total adiposity (% fat) compared to the more physically active (PR = 1.66, 95% CI: 1.10–2.50; p = .016). Among the girls, dietary pattern and tobacco use were associated with total adiposity. Compared to girls with an adequate diet, those with an inadequate diet were less likely to have higher total adiposity (PR = .76, 95% CI: .59–.98; p = .037). Smoking increased the probability of excess in both central adiposity (PR = 1.90, 95% CI: 1.17–3.08; p = .009) and total adiposity (PR = 1.61, 95% CI: 1.04–2.47; p = .032).

Table 2.

Association between risk behaviors and indicators of adiposity in adolescents.

	High WC				High % body fat
	M	PR_adj (95% CI)	F	PR_adj (95% CI)	M	PR_adj (95% CI)	F	PR_adj (95% CI)
Physical inactivity
No	16.3%	1	24.4%	1	14.0%	1	29.3%	1
Yes	20.8%	1.27 (0.86–1.86)	24.1%	0.96 (0.64–1.44)	23.2%	1.66 (1.10–2.50)	25.6%	0.90 (0.62–1.29)
p	.206	.227	.952	.841	.011	.016	.471	.563
TV watching
0–4 hours	17.1%	1	23.3%	1	18.1%	1	26.7%	1
>4 hours	22.9%	1.20 (0.82–1.74)	25.4%	1.06 (0.79–1.42)	24.1%	1.23 (0.87–1.75)	26.4%	0.99 (0.77–1.27)
p	.088	.344	.514	.715	.081	.244	.924	.931
Dietary pattern
Adequate	16.0%	1	28.2%	1	19.1%	1	31.1%	1
Inadequate	21.5%	1.29 (0.87–1.93)	22.2%	0.81 (0.62–1.06)	21.7%	1.16 (0.82–1.64)	24.3%	0.76 (0.59–0.98)
p	.117	.210	.091	.121	.197	.406	.059	.037
Alcohol consumption
Not frequent use	20.0%	1	24.8%	1	21.4%	1	25.4%	1
Frequent use	17.1%	0.92 (0.58–1.45)	20.2%	0.92 (0.60–1.41)	17.1%	0.90 (0.55–1.47)	29.8%	1.05 (0.73–1.52)
p	.509	.712	.269	.709	.326	.661	.306	.779
Tobacco smoking
Not frequent use	19.0%	1	23.0%	1	20.9%	1	24.5%	1
Frequent use	24.4%	1.12 (0.53–2.36)	36.4%	1.90 (1.17–3.08)	17.8%	0.90 (0.46–1.76)	45.5%	1.61 (1.04–2.47)
p	.378	.767	.026	.009	.620	.760	.001	.032

Note: N = 1321; M: male; F: female; PR_adj: prevalence ratio adjusted by chronological age and socioeconomic status; CI: confidence interval; WC: waist circumference.

With stratification by BMI_mother (Table 3), only tobacco use was associated with adiposity indicators. In adolescents whose mothers were of normal weight (BMI < 25 kg/m²), higher central adiposity (PR = 2.31, 95% CI: 1.19–4.46; p = .013) and total adiposity (PR = 2.31, 95% CI: 1.33–4.03; p = .003) were observed among those who reported smoking regularly. However, regardless of sex, chronological age, and SES, adolescents of overweight mothers had a 1.67 (95% CI: 1.27–2.15) and 1.48 (95% CI: 1.15 – 1.88) higher probability of high central and total adiposity (p < .001), respectively, when compared with adolescents of normal weight mothers.

Table 3.

Association between risk behaviors and indicators of adiposity in adolescents according to the BMI_mother.

	High WC				High % body fat
	BMI_mother < 25	PR_adj (95% CI)	BMI_mother > 25	PR_adj (95% CI)	BMI_mother < 25	PR_adj (95% CI)	BMI_mother > 25	PR_adj (95% CI)
Physical inactivity
No	10.8%	1	23.8%	1	10.8%	1	22.8%	1
Yes	17.5%	1.53 (0.75–3.11)	26.7%	1.09 (0.73–1.62)	20.2%	1.79 (0.88–3.66)	28.7%	1.33 (0.89–1.99)
p	.158	.245	.541	.674	.060	.108	.234	.166
TV watching
0–4 hours	14.5%	1	25.7%	1	19.1%	1	25.3%	1
>4 hours	18.2%	1.21 (0.77–1.89)	28.0%	1.09 (0.82–1.45)	18.2%	0.97 (0.65–1.47)	31.0%	1.23 (0.93–1.62)
p	.319	.404	.545	.559	.826	.899	.154	.154
Dietary pattern
Adequate	16.0%	1	29.3%	1	19.8%	1	29.9%	1
Inadequate	16.4%	1.03 (0.64–1.64)	24.7%	0.87 (0.65–1.17)	18.2%	0.91 (0.59–1.38)	26.8%	0.92 (0.69–1.23)
p	.920	.915	.262	.358	.700	.645	.455	.572
Alcohol consumption
Not frequent use	16.3%	1	27.8%	1	17.1%	1	29.0%	1
Frequent use	15.0%	1.04 (0.54–1.99)	19.8%	0.78 (0.50–1.21)	26.7%	1.53 (0.95–2.47)	21.9%	0.77 (0.50–1.17)
p	.802	.911	.108	.264	.079	.082	.160	.217
Tobacco smoking
Not frequent use	15.2%	1	26.4%	1	17.3%	1	27.3%	1
Frequent use	30.4%	2.31 (1.19–4.46)	24.2%	1.24 (0.65–2.38)	39.1%	2.31 (1.33–4.03)	33.3%	1.40 (0.83–2.38)
p	.045	.013	.781	.513	.009	.003	.451	.211

Note: N = 1321; BMI_mother: mother’s body mass index; PR_adj: prevalence ratio adjusted by sex, chronological age and socioeconomic status; CI: confidence interval; WC: waist circumference.

Discussion

The main results of this study indicate weak associations between behavioral risk factors and adiposity indicators in adolescents from Southern Brazil when analyzed by sex and BMI_mother. It is important to highlight that these associations seem to be different to some extent between the sexes and mother’s BMI category. The only behavioral risk factor positively associated with both the adiposity outcomes was tobacco use in girls and adolescents whose BMI_mother < 25 kg/m².

Our findings do not corroborate many other studies, especially those that endorse recommendations for the prevention and treatment of obesity (Barlow, 2007; Davis et al., 2007; Spear et al., 2007). To consensually establish causality between behaviors and biological outcomes in young people (short term), some factors should be considered, such as, the experimental design, the method for obtaining the variables, the characteristics of the population and the duration, intensity and frequency of the behaviors.

In this sense, the cross-sectional design adopted in this study is a limitation. Although it is useful to detect disease frequencies and risk factors, as well as to identify population groups that are more affected, this approach does not allow the establishment of temporality between variables. However, short term longitudinal designs should not be encouraged since a large part of the morbidities attributed to lifestyle have a long latency period. Therefore, in order to discuss reverse causality, a characteristic of cross-sectional designs, the stratified analysis of BMI_mother was performed, hypothesizing that adolescents whose mothers are overweight are more prone to the effects of reverse causality.

The method for obtaining measurements is the second limitation of this study. Information about lifestyle, mainly in middle and low-income countries has been obtained predominantly using questionnaires (Farias Júnior et al., 2010; Hallal et al., 2007). However, particularly in adolescents, data collected from questionnaires should be viewed with caution since many behaviors are susceptible to social approval. In this case, questionnaires were used in the school environment where the presence of peers or uncertainty about the confidentiality of the information to their parents may affect the reporting of the data (Brener et al., 2003). Although we took care of the reproducibility of the questionnaires’ results, the absence of a gold standard method to measure physical activity does not allow us to control the validity. However, it is worth highlighting the methodological care adopted, such as sample selection, quality control of the measures, and statistical power of the analyses, as strengths of this study.

Based on the analyses, we believe that the associations found between physical inactivity, sedentary behavior, and alcohol consumption with adiposity (central and total) did not suffer the effect of reverse causality since no result was significant in the stratified analysis with BMI_mother. Regarding eating habits, the results seem to be more influenced by sex (girls may have reported their habits distortedly) than mother’s nutritional status (no significant association). The association between adiposity and tobacco use, less reported in the literature and even an inverse determination product, positively observed in girls, remained even in the adolescents whose mothers had BMI < 25 kg/m². This indicates that, among children whose mothers are overweight, tobacco use was found to have no relationship with adiposity indicators. In this instance, we concluded that the measurement collection may have more influence than the reverse causality.

Others studies conducted in Brazil have also not observed associations between risk behaviors and adiposity indicators in adolescents (Moraes and Falcao, 2013; Silva et al., 2011), corroborating our findings. Further investigations, unpublished because they do not show positive associations, should also be considered. Therefore, we highlight the importance of discussing the methodological problems inherent in cross-sectional designs, in the sense of looking carefully at the observed results and trying to overcome the problems.

A relevant point relates to the detailed characteristics of the selected behaviors, such as the intensity, duration, and frequency of the habit or combinations thereof. Studies have suggested that the relationship between specific behaviors and negative outcomes depends on how this behavior is conducted (Ambrosini et al., 2012; Byun et al., 2012; Ortega et al., 2007) and their combination may soften/potentiate the health risk outcomes (Chaput et al., 2012; Rey-Lopez et al., 2012). Thus, the consideration of habits in isolation may not reflect the true position, creating biased conclusions and negatively influencing the practical applications of the results for public policies.

Finally, we conclude that, among the risk behaviors investigated in this study (physical inactivity, sedentary behavior, inadequate dietary pattern, alcohol, and tobacco use), the latter is the behavior most associated with adiposity indicators in adolescents from Southern Brazil. Based on the foregoing discussion, in studies of this nature, the authors reinforce the need to advance methodological aspects, mainly with regard to research design, methods of measuring behaviors detailed, and grouped analyses of adolescents’ behavioral risks.

Footnotes

Acknowledgments

The authors thank Décio Barbosa, Danielle Venturini, Alessandra Okino, and Jair Oliveira for research support, Mariana Carnelossi and Sandra Kawaguti for acquisition of data, Coordination for the Improvement of Higher Education Personnel (CAPES, Brazil) for scholarships (DRPS, CMT, MBB), and National Council of Scientific and Technological Development (CNPq/Brazil) for funding the project and for scholarship productive research (ERVR and ESC).

Conflict of interest

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received financial support by National Council of Scientific and Technological Development (CNPq/Brazil) for the research. No financial support was received for authorship and/or publication of this article.

References

Ambrosini

Emmett

Northstone

. (2012) Identification of a dietary pattern prospectively associated with increased adiposity during childhood and adolescence. International Journal of Obesity 36(10): 1299–1305.

Assuncao

Muniz

Dumith

. (2012) Predictors of body mass index change from 11 to 15 years of age: the 1993 Pelotas (Brazil) birth cohort study. Journal of Adolescent Health 51(6 Suppl): S65–S69.

Baecke

Burema

Frijters

(1982) A short questionnaire for the measurement of habitual physical activity in epidemiological studies. The American Journal of Clinical Nutrition 36(5): 936–942.

Barlow

(2007) Expert committee recommendations regarding the prevention, assessment, and treatment of child and adolescent overweight and obesity: summary report. Pediatrics 120(Suppl 4): S164–S192.

Barros

Hirakata

(2003) Alternatives for logistic regression in cross-sectional studies: an empirical comparison of models that directly estimate the prevalence ratio. BMC Medical Research Methodology 3: 21–34.

Boileau

Lohman

Slaughter

(1985) Exercise and body composition in children and youth. Scandinavian Journal of Medicine & Science in Sports 17: 17–27.

Boulos

Vikre

Oppenheimer

. (2012) ObesiTV: how television is influencing the obesity epidemic. Physiology & Behavior 107(1): 146–153.

Bozza

Campos

Bacil

EDA

. (2014). Sociodemographic and behavioral factors associated with body adiposity in adolescents. Revista Paulista de Pediatria 32(3): 241–246.

Brazilian Association of Research Companies (BARC) (2008) Brazil economic classification criterion 2008. Available at: http://www.abep.org/codigosguias/Criterio_Brasil_2008.pdf (accessed 21 February 2013).

10.

Brazilian Institute of Geography and Statistics (BIGS) (2012) National Survey of School Health. Rio de Janeiro: Ministry of Health, Brazil. Available at: www.ibge.gov.br/home/estatistica/populacao/pense/…/pense_2012.pdf (accessed 07 March 2014).

11.

Brazilian Institute of Geography and Statistics (BIGS) (2013) Gross Domestic Product of Municipalities. Available at: http://www.ibge.gov.br/cidadesat/xtras/perfil.php?codmun=411370&search=parana|londrina (accessed 5 July 2013).

12.

Brener

Billy

Grady

(2003) Assessment of factors affecting the validity of self-reported health-risk behavior among adolescents: evidence from the scientific literature. Journal of Adolescent Health 33(6): 436–457.

13.

Byun

Dowda

Pate

(2012) Associations between screen-based sedentary behavior and cardiovascular disease risk factors in Korean youth. Journal of Korean Medical Science 27(4): 388–394.

14.

Chaput

Lambert

Mathieu

. (2012) Physical activity vs. sedentary time: independent associations with adiposity in children. Pediatric Obesity 7(3): 251–258.

15.

Currie

Gabhainn

Godeau

. (2008) Inequalities in Young People’s Health: HBSC International Report from the 2005/2006 Survey. Copenhagen: World Health Organization.

16.

Davis

Gance-Cleveland

Hassink

. (2007) Recommendations for prevention of childhood obesity. Pediatrics 120(Suppl 4): S229–S253.

17.

Ebbeling

Pawlak

Ludwig

(2002) Childhood obesity: public-health crisis, common sense cure. Lancet 360(9331): 473–482.

18.

Farias Júnior

Lopes

Florindo

. (2010) Validity and reliability of self-report instruments for measuring physical activity in adolescents: a systematic review. Cadernos de Saúde Pública 26(9): 1669–1691.

19.

Fernandes

Reichert

Monteiro

. (2012) Characteristics of family nucleus as correlates of regular participation in sports among adolescents. International Journal of Public Health 57(2): 431–435.

20.

Hallal

Dumith

Bastos

. (2007) Evolution of the epidemiological research on physical activity in Brazil: a systematic review. Revista de Saúde Pública 41(3): 453–460.

21.

Hancox

Milne

Poulton

(2004) Association between child and adolescent television viewing and adult health: a longitudinal birth cohort study. Lancet 364(9430): 257–262.

22.

Harrison

Buskirk

Carter

. (1988) Skinfold thicknesses and measurement technique. In: Lohman

Roche

Martorell

(eds) Anthropometric standardization reference manual. Champaign: Human Kinetics Books, p.55–80.

23.

Katzmarzyk

(2004) Waist circumference percentiles for Canadian youth 11-18y of age. European Journal of Clinical Nutrition 58(7): 1011–1015.

24.

Ludwig

(2007) Childhood obesity—the shape of things to come. The New England Journal of Medicine 357(23): 2325–2327.

25.

Moraes

Falcao

(2013) Lifestyle factors and socioeconomic variables associated with abdominal obesity in Brazilian adolescents. Annals of Human Biology 40(1): 1–8.

26.

Neumark-Sztainer

Wall

Story

. (2012) Dieting and unhealthy weight control behaviors during adolescence: associations with 10-year changes in body mass index. Journal of Adolescent Health 50(1): 80–86.

27.

Ortega

Ruiz

Sjostrom

(2007) Physical activity, overweight and central adiposity in Swedish children and adolescents: the European Youth Heart Study. International Journal of Behavioral Nutrition and Physical Activity 4: 61.

28.

Rey-Lopez

Ruiz

Vicente-Rodriguez

. (2012) Physical activity does not attenuate the obesity risk of TV viewing in youth. Pediatric Obesity 7(3): 240–250.

29.

Sikorski

Luppa

Brahler

. (2012) Obese children, adults and senior citizens in the eyes of the general public: results of a representative study on stigma and causation of obesity. PLoS One 7(10): e46924.

30.

Silva

Pelegrini

de Lima

ESJM

. (2011) Epidemiology of whole body, peripheral, and central adiposity in adolescents from a Brazilian state capital. European Journal of Pediatrics 170(12): 1541–1550.

31.

Spear

Barlow

Ervin

. (2007) Recommendations for treatment of child and adolescent overweight and obesity. Pediatrics 120(Suppl 4): S254–S288.

32.

Williams

Going

Lohman

. (1992) Body fatness and risk for elevated blood pressure, total cholesterol, and serum lipoprotein ratios in children and adolescents. American Journal of Public Health 82(3): 358–363.