Introduction

Disordered eating behaviors involve inappropriate methods for body weight control, including restrictive diets, skipping meals, exclusion of high energy density foods of the diet, prolonged fasting, episodes of binge eating, and compensatory behaviors, such as the use of laxatives, diuretics, and self-induced vomiting [1]. These behaviors aim to avoid weight gain or promote weight loss, usually due to excessive appreciation of lean body esthetics and dissatisfaction with body image [1]. Consequently, there is an increased risk of developing of eating disorders, such as anorexia nervosa, bulimia nervosa, and binge eating disorder [2], which are severe conditions that may be fatal [3].

Different methods to investigate disordered eating behaviors have been used in epidemiological studies. Irrespective of the method used, a high prevalence of disordered eating behaviors has been observed among adolescents [4, 5].

Because they affect a phase of intense growth and development, disordered eating behaviors during adolescence are worrying, as they can affect the diet consumed, both quantitatively and qualitatively, compromising the achievement of high energy and nutrient demands.

Studies investigating the association between nutritional inadequacies and the presence of disordered eating behaviors among adolescents, observed a lower intake of energy [6, 7] and micronutrients, such as zinc [6], and vitamins B6 and B12 [7] among individuals exhibiting disordered eating behaviors. Of note, populations of low socioeconomic status may be more vulnerable to poorer food consumption [8] and the presence of disordered eating behaviors in this population is particularly worrying.

Conversely, higher values of body mass index and body fat have also been observed in adolescents with disordered eating behaviors [4, 9]. However, these findings are controversial as some studies did not find such associations [10, 11].

The aim of this study was to investigate the association between disordered eating behaviors and energy and nutrient intake, anthropometric profile, and body composition among adolescents from public schools in the metropolitan region of Rio de Janeiro, which are most frequently attended by individuals of lower socioeconomic level.

This study is the first in Brazil to evaluate the association between abnormal food patterns or unusual food patterns and energy and nutrient intake in a probabilistic sample of predominantly low socioeconomic level adolescents.

Methods

In this study were analyzed data from students aged 15–19 years that were part of a larger cross-sectional study developed with a probabilistic sample of students from the 5th grade of middle school to the 3rd year of high school in 13 of the 33 public schools located in the city of Niterói, Southeast region of Brazil. The main objective of the larger study was to evaluate changes in nutritional status over a 5-year period, based on two studies conducted in 2003 and 2008. This article used the data from the 2008 study. The socio-demographic characteristics of Niterói are described in Bagni et al. [12]. The type of school attended by students has been considered a proxy of socioeconomic status that has been used in Brazilian studies [13, 14] so that those who attend public schools are considered, in the majority, of lower economic status.

The initial sample of the larger study consisted of 894 students aged 12–19 years distributed across 30 randomly selected classes (with an expected average of 30 students per class) who met the following eligibility criteria: not being pregnant, and not have a physical disability which would hinder the anthropometric evaluation and complete the questionnaire. Of these, 53 students did not present the consent form signed by theirs parents, 101 refused to participate in the study, and 40 did not attend the data collection. Because adolescents under 15 years of age did not perform well in completing the questionnaires chosen as research instruments, i.e., the Eating Attitudes Test (EAT) and the Bulimic Investigatory Test Edinburgh (BITE), due to their complexity, only subjects aged 15 years or above were selected to participate in this study, consisting of 498 adolescents in total. Of these, 11 did not respond to the questionnaires used to investigate disordered eating behaviors. Finally, this study included 487 individuals in the analysis examining the association between disordered eating behaviors and anthropometric profile and body composition.

This sample size allowed an estimation of the prevalence of disordered eating behaviors of approximately 25% [15], with a 95% confidence interval and an absolute precision of 5 percentage points, considering the effect of sample design by cluster in one stage (classes) [16]. Only subjects who completed the 3-day food record (n = 371) were included in the assessment of the association between disordered eating behaviors and energy and nutrient intake.

Data were collected between May 2008 and April 2009 by trained staff. Measurements were taken with adolescents wearing light clothing and barefoot. Weight and height were measured according to the protocols described by Gordon et al. [17]. Weight was measured using a portable electronic scale (Kratos®), with a maximum capacity of 150 kg, resolution of 50 g. The height was measured using a portable anthropometer (Alturexata®), with amplitude of 213 cm, and resolution of 0.1 cm. Duplicate measurements were taken, and the average of both measurements was used as the valid measure. A maximum variation of 0.5 cm between the two measurements was allowed. The waist circumference (WC) was measured in duplicate with an inelastic tape measure, as the smallest horizontal circumference of the trunk, with an accuracy of 0.1 cm, according to the protocol proposed by Callaway et al. [18]. A maximum variation of 1.0 cm between the two measurements was allowed, obtaining the mean.

The nutritional status was classified according to sex- and age-specific BMI cutoff points based on the reference curve proposed by the World Health Organization [19].

The body composition was evaluated by Bioelectric impedance (BIA), using a portable analyzer BIA Analyzer (101Q, RJL Systems®, Detroit, USA). Participants were instructed to avoid coffee and alcohol consumption and not to practice physical activity on the day of the evaluation, to minimize possible measurement errors. The use of the BIA for body composition evaluation in children and adolescents was previously validated [20]. Body fat (kg) was estimated by calculating the difference between body mass (kg) and fat free mass (kg), obtaining relative (%) body fat (BF).

Food consumption to estimate energy and nutrient intake was evaluated by a 3-day food record (2 non-consecutive weekdays and 1 weekend day). The adolescents were instructed by the team nutritionist to keep as complete and detailed a record as possible. Photographs of standard kitchen utensils that are normally used to serve food were shown to students, aiming to facilitate the recording of portion sizes. At the time of delivery, all records were rigorously checked by the researchers in the presence of the adolescent, to clarify any doubts regarding the process of completing the questionnaire.

The composition of the diet in energy (kcal), macronutrients (carbohydrates, proteins and lipids) and micronutrients was as follows: calcium (mg), folate (µg), zinc (mg), vitamin C (mg), vitamin A (µg/day of retinol), vitamin B12 (µg) and iron (mg) were evaluated using NutWin® software version 1.5.2.51 [21]. The database of the software uses information from the chemical composition of food of the United States Department of Agriculture (USDA). For foods that were not available using the software, information on nutritional composition was obtained from food composition tables [22, 23, 24], or from the labels of industrialized foods.

To estimate individual energy and nutrient intake, the mean of a 3-day food record was deattenuated by intra-individual variability using the method proposed by Iowa State University with PC-SIDE® version 1.0 software (Software for Intake Distribution Estimation for the Windows OS, Department of Statistics, Iowa State University, 2003) [25]. The deattenuated values of nutrient intake was also adjusted for total energy consumption, according to the residual method [26].

Inadequate intake of micronutrients was evaluated according to the reference values of the estimated average requirement (EAR) [27,28,29]. This was presented as the prevalence of individuals with consumption below the EAR value [30]. For iron, the prevalence of inadequate intake was estimated by a manually determined probabilistic approach, due to its non-normal distribution [28].

The disordered eating behaviors were identified by two self-filling questionnaires: (1) the Eating Attitudes Test (EAT-26), a 26-question model proposed by Garner et al. [31], translated and adapted into Portuguese by Nunes et al. [32]. Individuals who scored ≥ 20 (EAT+) were considered as susceptible to the development of anorexia nervosa with an “abnormal food pattern”; (2) the Bulimic Investigatory Test Edinburgh (BITE), developed by Henderson and Freeman [33], translated and adapted into Portuguese by Cordás and Hochgraf [34]. This questionnaire is composed of 33 questions with two subscales: symptom and severity. The first evaluates the presence of bulimic symptoms and the second the intensity of these symptoms. A score ≥ 10 (BITE+) on the symptom scale was suggestive of an “unusual food pattern” [34]. The severity scale was evaluated only for those cases where the score on the symptom scale was greater than or equal to 10. A score below 5 was categorized as low severity, a score between 5 and 9 was categorized as moderate severity, and a score equal or above 10 was categorized as high severity [35].

Statistical analyzes were performed considering the effect of sample design by cluster (classes) and the expansion of the sample by the sampling weight of each individual (1/Pr, where Pr = probability of each adolescent has participated of the study), using the commands for complex samples of the Statistical Package for the Social Sciences—SPSS version 19.0 (IBM Corp, Armonk, NY, 2010).

The variables investigated were: the presence of abnormal eating patterns, evaluation using EAT (yes or no), and the presence of unusual dietary patterns evaluated by BITE (yes or no); nutritional status (overweight, grouping overweight and obesity or non-overweight, grouping adequate weight and underweight); percentage of body fat (no excess < 25% for boys and < 30% for girls, with excess ≥ 25% for boys and ≥ 30% for girls); caloric, macronutrients (carbohydrates, proteins, and lipids) and micronutrient (calcium, folate, zinc, vitamin C, vitamin A, vitamin B12 and iron) intake as continuous and dichotomous variables (adequate or inadequate intake).

A descriptive analysis for anthropometrics and disordered eating behaviors variables was performed by the means of frequencies and their respective 95% confidence intervals (95% CI). The statistical differences between the frequencies were tested by the Pearson Chi-square test. The magnitude of the association between the variables was investigated by mean of the prevalence ratio (PR) and the 95% CI and linear regression, respectively, applying the SPSS General Linear Model procedure. The p value < 0.05 was used for statistical significance. Recognizing the heterogeneity between boys and girls regarding the variables investigated, all analyzes were stratified by sex.

The research was approved by the Committee for Ethics in Research of the Instituto de Pediatria e Puericultura Martagão Gesteira of the Federal University of Rio de Janeiro (protocol number 27/08). Participation was voluntary, and only students who had consent forms signed by their parents were allowed to participate.

Results

Of the 487 participants, 61% were female and the mean age was 16.9 years (95% CI 16.6–17.2). The prevalence of abnormal food patterns evaluated by the EAT questionnaire was 7.4% and the prevalence of unusual food patterns evaluated by the BITE questionnaire was 18.9%. The presence of at least one disordered eating behavior was 22.1%, while 4.2% of adolescents had a positive result for both instruments. These behaviors were significantly (p < 0.05) more frequent in girls than in boys (Table 1).

Table 1 Prevalence (%) and 95% confidence intervals of BMI and % body fat classification, and disordered eating behaviors in total adolescents and according to sex
Full size table

In the assessment of the severity scale for the BITE questionnaire, 72.5% of adolescents were classified as low severity, 21.1% as moderate severity and 6.4% as high severity, with no significant differences between the sexes (data not shown).

Overweight was observed in 16.5% of adolescents, obesity in 3.7%, and low weight in only 1.0%, with no differences between the sexes. Excess body fat was observed in 25% of the adolescents, being six times higher in girls (37%) than in boys (6.7%, p < 0.001) (Table 1).

Overweight girls presented a 2.4 (95% CI 1.6–3.5) higher prevalence of unusual food patterns when compared to those without overweight (Table 2). The association between abnormal food patterns and the variables of interest were not investigated in boys due to the low prevalence of this behavior.

Table 2 Prevalence (%) of abnormal food patterns (EAT+) and unusual food patterns (BITE+), and association with anthropometric and body composition characteristics (prevalence ratio and 95% confidence intervals), according to sex, in adolescents from public schools
Full size table

The nutrients with the highest frequency of inadequacy were calcium (100% of adolescents) and folate (95.6%). Over half of the adolescents presented inadequate intake of zinc (59.4%), vitamin A (60.7%) and vitamin C (56.8%), and one-fourth of the subjects had inadequate iron intake. The frequency of vitamin C inadequacy was lower among girls with abnormal food patterns (Table 3). It was also observed that vitamin C mean intake was higher among girls with an abnormal food pattern and calcium mean intake was lower among boys with an unusual food pattern (Table 4).

Table 3 Prevalence (%) of inadequate intake of micronutrients according to the presence of abnormal eating patterns (EAT+) and unusual eating patterns (BITE+), according to sex, in adolescents of public schools
Full size table
Table 4 Means (Standard Error) of energy and nutrient intake for presence or absence of abnormal food patterns (EAT + or EAT−) and unusual food patterns (BITE + or BITE−), according to sex, in adolescents of public schools
Full size table

Discussion

One of the main results of this study is that overweight girls had a higher risk of developing unusual food patterns than non-overweight girls. Greater intake of vitamin C was observed among girls with abnormal food patterns, resulting in a lower frequency of inadequacy of this vitamin in this group. Calcium intake was lower among boys with unusual food patterns compared to those without unusual food patterns.

The association found in the present study between unusual food patterns and overweight has been previously observed by Miotto et al. [36] in a study evaluating Italian students aged 15–19 years which found a positive association between high BMI values and higher BITE scores. However, BMI was not significantly associated with abnormal food patterns, assessed by the EAT score, as observed in our study. In Brazil, Ferreira et al. [37], using a simplified questionnaire, found an association between binge eating and restrictive dieting and overweight in adolescents living in a region of low socioeconomic level, regardless of sex. These findings were supported by a later study targeting the same population [4].

Further studies have confirmed positive associations between disordered eating behaviors and BMI [4, 9, 37] and percentage of body fat [9]. This finding is understandable considering that adolescents with overweight and excess of body fat are more likely to develop concerns about body weight and body image, increasing the risk of inadequate dietary practices for weight control.

Our study demonstrated a higher intake of vitamin C among girls with abnormal food patterns, resulting in a lower frequency of vitamin C inadequacy in this group. This could be explained by an increased consumption of fruits and vegetables, rich sources of vitamin C, reflecting a greater concern with body weight by the preference for foods with a lower caloric content. These results are consistent with the study conducted by Dunker and Phillipi [38], in which girls with abnormal eating patterns showed more frequent fruit and vegetable consumption than those without such behavior. Tsai et al. [39] followed-up girls aged between 11 and 15 years in Taiwan reporting a higher intake of fiber, derived from an increased consumption of fruit and vegetables, among girls with abnormal eating patterns. However, the authors also observed a lower energy intake among these adolescents and this was not found in the present study.

It is important to highlight that a number of approaches to control body weight, such as increasing fruit and vegetable intake and reducing foods rich in fats and sugars, which are characteristic of adolescents with disordered eating behaviors, are acceptable, and even desirable, in moderation. However, the benefits of choosing healthier foods do not outweigh the losses caused by excessively restrictive diets that may compromise the health of young people and fail to meet daily energy and nutrient requirements.

Aparicio-Llopis et al. [7], examined the association between disordered eating behaviors (evaluated by the EAT) and energy and nutrient intake in 495 Spanish adolescents. The authors found that girls with disordered eating behaviors had a lower intake of vitamin C, energy, and macro and other micronutrients (calcium, iron, magnesium, potassium, phosphorus, sodium, thiamine, vitamin E, B6 and B12, pantothenic acid and folic acid) than those without these behaviors. Estecha et al. [6] also found lower intakes of energy and other nutrients, such as sodium, potassium, phosphorus, iron, zinc, vitamin B2 and niacin and no difference in vitamin C intake in Spanish girls.

In the present study, the mean calcium intake was lower among boys with an unusual food pattern than among those without disordered eating behaviors. However, there were no differences between the groups in the inadequacy of calcium intake, reaching 100% of the adolescents, regardless of the presence or absence of disordered eating behaviors. The inadequate intake of calcium by Brazilian adolescents was also observed in the National Food Survey, with calcium being the nutrient with the highest percentage of inadequacy [40]. This result is worrying considering the essential physiological functions of this nutrient, particularly in adolescents. Approximately 40% of the total bone mass of the skeleton is acquired during puberty, determining adult bone mass, which in turn is a significant indicator of fracture risk later [29]. Furthermore, inadequate intake of calcium and dairy products has been associated with overweight and increased body fat among adolescents [41], which could lead to a vicious cycle, as excess weight is related to disordered eating behaviors.

Considering that populations of low socioeconomic status are more vulnerable to poor dietary intake [8] the presence of disordered eating behaviors in this population could be potentially more serious because these can further compromise the diet. However, regardless of these behaviors, nutrient intake was inadequate in a high proportion of adolescents, reflecting the low quality of the Brazilian adolescent diet predisposing them to nutritional deficiencies.

A limitation of this study is the cross-sectional design which does not allow the determination of a cause-and-effect relationship between the variables. It is also likely that adolescents with disordered eating behaviors have difficulties reporting their actual food intake because of their complex relationship with diet and weight control. This may have led to overestimation of food consumption [42]. Future research, with a longitudinal design and using a combination of methodologies, is required to assess food intake among adolescents and clarify the associations between disordered eating behaviors and food consumption. Another limitation is the lack of socioeconomic level assessment, but in Brazil is well known that public schools are most frequented per adolescents from low socioeconomic level [13].

This study found that the magnitude of disordered eating behaviors was high among the adolescents investigated, particularly the unusual food pattern evaluated by the BITE, and that they occur more frequently in girls, particularly among girls with excess weight. Regarding dietary intake, such behaviors may lead to lower calcium intake in boys and higher intake of vitamin C in girls. However, the low nutritional quality of the diet in general, regardless of the presence or not of disordered eating behaviors, is highlighted. It is important to note the fundamental role of nutrition in this age group and the need for public policies to encourage healthier eating practices that may contribute to a decrease in the prevalence of disordered eating behaviors and nutritional inadequacies.