INTRODUCTION
Adolescence is a period of life characterized by important physical, behavioral and brain changes 1,2. Nutritional requirements and dietary habits are modified during this age-span, which could lead to different nutrition-related disorders with adverse consequences for overall health status 3.
Eating disorders (ED) are complex diseases characterized by disruptive eating behaviors linked to body shape dissatisfaction, disinterest in food and an intense fear of gaining weight 4. According to scientific literature, anorexia nervosa, bulimia nervosa and binge eating disorder are the most investigated severe clinically defined ED. The origin of ED is multifaceted, being involved a combination of psychological (i.e., positive self-esteem, emotional well-being), biological (i.e., sex) and social factors (i.e., family connectedness) 5. A peak of incidence in the number of ED has been reported in adolescents between 15-19 years old 6, with symptoms starting during pre-adolescence and early adolescence 7.
Since ED are associated to adverse consequences for physical health and psychosocial and cognitive functioning 5, the early detection of people at risk of developing ED is essential to prevent its progression. Evidence from previous studies in adolescents suggest that cognitive deficits pre-exist and underlie the etiology of ED 8, while high academic performance could act as a protective factor 9. However, limited and non-conclusive research has been undertaken regarding the link between the risk of developing ED and academic performance in adolescents. To this extent, Veses et al. 10 found higher risk of ED among non-overweight adolescent girls with low academic performance, and among overweight adolescent boys with low school grades in physical education (PE). Conversely, Gil et al. 11 did not find differences in academic performance between adolescents at risk and not at risk of ED.
Although ED is commonly related to normal or low body weight, obesity and ED may coexist 12. Since obesity is a weight-related disorder with a high prevalence among adolescents, it might affect body satisfaction and perception leading to disruptive eating behaviors. Overweight and obesity have been positively associated with an increased risk of developing ED in adolescents 10,13,14. These disorders share multiple similarities and may occur simultaneously 15 or as a consequence of each other 16. In addition, recent research has also pointed out the negative association of body fatness with cognition in adolescents 17. Body image dissatisfaction, weight-related teasing, unhealthy behaviors, and health-related problems closely linked to overweight and obese adolescents have been postulated as potential mechanisms by which excess of body fat negatively influences academic outcomes and cognitive functions 17.
Given the high prevalence of obesity and the key role of body weight on the risk of developing ED and cognition in adolescents, the main aim of the present study was to analyze the association between the risk of ED and academic performance in healthy adolescents, considering the effect of weight status. We additionally examined: a) the differences in academic performance according to ED categories; and b) the risk of ED according to weight status.
METHODS
PARTICIPANTS
The DADOS (Deporte, Adolescencia y Salud) Study is a 3-year longitudinal research project (from 2015 to 2017) aimed to assess the influence of physical activity on health, cognition and psychological wellness through adolescence. All the participants were recruited from secondary schools and sport clubs of Castellon (Spain), and met the general DADOS inclusion criteria: born in 2001, enrolled in second grade of secondary school and free of any chronic disease. The results presented in this study belong to baseline data obtained between February and May of 2015. From the total DADOS study sample, 261 adolescents (138 boys) with valid data for at least risk of eating disorders, weight status, and academic performance were included in the analyses.
Adolescents and their parents or guardians were informed of the nature and characteristics of the study, and all provided a written informed consent. The DADOS study protocol was designed in accordance with the ethical guidelines of the Declaration of Helsinki of 1961 (last revision of Fortaleza, Brazil, 2013) and approved by the Research Ethics Committee of the Universitat Jaume I of Castellon.
WEIGHT STATUS
Briefly, body weight was measured to the nearest 0.1 kg using an electronic scale (Seca(r) 861, Hamburg, Germany) with the participant lightly dressed and without shoes. Height was measured to the nearest 0.1 cm using a wall-mounted stadiometer (Seca(r) 213, Hamburg, Germany). Measures were assessed in duplicate by trained members of the DADOS research group following standardized procedures 18 and average measures were used for data analysis. Body mass index (BMI) was calculated as weight/height square (kg/m2).
THE RISK OF EATING DISORDERS
The risk of ED was assessed using the Sick Control on Fast Food (SCOFF) questionnaire 19. The questionnaire consists of five questions regarding deliberate vomiting, loss of control over eating, weight loss, body image distortion and impact of food on life. A value of 1 is assigned to each affirmative answer, ranging the SCOFF overall score from 0 to 5. A score ≥ 2 in the SCOFF score indicates a likely case of suffering from ED. The Spanish version of the SCOFF validated for adolescents was used 20, which has shown accurate internal consistency and criteria validity for screening ED in adolescents (81.9% sensitivity and 78.7% specificity) 21.
ACADEMIC PERFORMANCE
Academic performance was assessed by two components:
The final academic grades from the 1st course of secondary school provided by each school's secretary office. The following indicators were included in the analyses: individual grades for Math, Language and Physical Education (PE) and grade point average (GPA) score. Language is the grade of Catalan, the official teaching language at school. GPA score was defined as the single average for Geography and History, Natural Science, Math, Spanish, Catalan, English and Physical Education grades. All the subjects are measured on a ten-point scale, where 1 is the worst and 10 is the best.
The Spanish version of the validated Science Research Associates Test of Educational Abilities (TEA) 22. This test measures three basic abilities: verbal ability (command of language), numeric ability (speed and precision in performing operations with numbers and quantitative concepts), and reasoning ability (the aptitude to find logical ordination criteria in sets of numbers, figures or letters). Scores for the three abilities were obtained by adding positive answers. Overall academic ability was calculated by adding the three abilities' scores (verbal + numeric + reasoning). This battery test provides three complexity levels based on the age range of the sample. The present work used the level 3 designed for adolescents aged 14 to 18 years (reliability: verbal α = 0.74, numeric α = 0.87, reasoning α = 0.77 and overall academic ability α = 0.89) 22.
COVARIATES
Pubertal stage was self-reported according to the five stages defined by Tanner and Whitehouse 23. It is based on external primary and secondary sex characteristics, which are described by the participants using standard pictures according to Tanner instructions.
The Family Affluence Scale "FAS" developed by Currie et al. 24 was used as a proxy of socioeconomic status (SES), which is based on material conditions in the family such as car ownership, bedroom occupancy, computer ownership and home internet access.
Parental educational attainment was self-reported by both parents as primary school, secondary school, bachelor studies or university degree. Study levels were dichotomized in two groups: no university degree vs university degree for each parent. Maximum parental educational attainment was used in the analyses.
STATISTICAL ANALYSIS
Descriptive characteristics are presented as mean and standard deviation (SD) or percentages. Differences between sexes were examined using the t-test and Chi-square test for continuous and nominal variables, respectively. As preliminary analyses showed no significant interactions between sex and ED risk scores (p > 0.10), all analyses were performed with the total sample.
Partial correlation analyses controlling for sex, pubertal stage, SES and parental educational attainment were performed to examine the associations of the risk of ED with weight status and academic performance.
Linear regression analyses were performed to investigate the associations of the risk of ED (SCOFF score ranging from 0 to 5) with academic performance. We created three regression models: a) unadjusted model; b) model adjusted for sex, pubertal stage, SES and parental educational attainment; and c) model adjusted for confounders in model 2 plus BMI. Analyses of covariance (ANCOVA) were conducted to examine whether academic performance differed between eating disorders risk categories (eating disorder risk vs non-eating disorder risk). These analyses were adjusted for sex, pubertal stage, SES, parental educational attainment and BMI.
Finally, binary logistic regression analysis controlling for sex, pubertal stage, SES and parental educational attainment was performed to estimate the probability of the risk of eating disorders according to weight status (non-overweight vs overweight including obesity). All the analyses were performed using the IBM SPSS Statistics for Windows version 22.0 (Armonk, NY: IBM Corp), and the level of significance was set to p < 0.05.
RESULTS
Descriptive characteristics of the study sample by sex are displayed in Table 1. Weight status was similar for boys and girls, with a 13% of overweight or obese adolescents. The risk of developing ED was of 12%, with no differences among sexes. Girls reported higher loss of control over eating and body image distortion than boys (26.0% vs 13.8% and 17.9% vs 8.0%, respectively; all p < 0.05). Academic performance did not show differences among sexes except for numeric ability that was higher in boys (p < 0.001).
Data are presented as mean ± SD or percentages. Differences between sexes were examined by t-test or Chi-square test. Statistically significant values are in italics.
*Includes overweight + obese participants.BMI: body mass index; SCOFF: Sick Control on Fast Food; PE: Physical Education; GPA: grade point average. A score ≥ 2 in the SCOFF questionnaire indicates risk of eating disorders. Overall indicates the sum of the three abilities scores.
Partial correlations among all the study variables controlling for sex, pubertal stage, SES and parental educational attainment are shown in Table 2. The risk of ED (SCOFF score) was positively associated with weight status (r = 0.262, p < 0.001) and negatively associated with all the academic grades (r ranging from -0.261 to -0.186; all p < 0.01), and with verbal and numeric abilities (r = -0.146 and -0.125, respectively; all p < 0.05).
BMI: body mass index; SCOFF: Sick Control on Fast Food; PE: Physical Education; GPA: grade point average. Overall indicates the sum of the three abilities scores. p-value = and *p < 0.05, †p < 0.01, ‡p ≤ 0.001.
*p < 0.05
†< 0.01
‡p ≤ 0.001.
Table 3 presents the associations between the risk of ED and academic performance variables. In model 1 (unadjusted model), the risk of ED was negatively associated with all the academic grades (Math, Language, PE and GPA; β ranging from -0.152 to -0.202), while no significant associations were found with academic abilities. After further controlling for sex, pubertal stage, SES and parental educational attainment (model 2; β ranging from -0.181 to -0.253), and BMI (model 3; β ranging from -0.214 to -0.141), the results for academic grades remained the same. Regarding academic abilities, the risk of ED was negatively associated with verbal and numeric abilities in model 2, and with verbal ability in model 3 (all p < 0.05).
PE: Physical Education; GPA: grade point average; β: standardized coefficient. Model 1: unadjusted. Model 2: adjusted for sex, pubertal stage, socioeconomic status and parental educational attainment. Model 3: adjusted for model 2 + body mass index. Overall indicates the sum of the three abilities scores. Statistically significant values are highlighted in italics.
Figure 1 displays the differences in academic grades between categories of risk of developing ED, adjusting for potential confounders. Non-risk of ED adolescents (SCOFF score < 2) showed higher academic grades (all p < 0.01) compared with those adolescents with risk of ED (SCOFF score ≥ 2). No differences were found for academic abilities between categories of ED risk (data not shown).
Binary logistic regression analysis (Table 4) showed that the risk of ED was three times higher in overweight or obese adolescents than in non-overweight participants (OR = 3.016; 95% CI = [1.212; 7.504]; p = 0.018).
DISCUSSION
The main finding of our study suggests an inverse association between the risk of ED and academic performance in adolescents. Adolescents without risk of ED had higher scores in all the academic grades. Moreover, overweight or obese adolescents showed higher risk of ED. These findings extend the existing scientific knowledge by highlighting the impact of ED on academic performance in adolescents and the importance of considering weight status.
The estimated prevalence of the risk of developing ED in our sample (12.3%) did not show differences between sexes, concurring with Veses et al. 13. However, previous studies have reported higher prevalence rates of the risk of ED in Spanish (i.e., 17% or 21%) 10,25 and German (22%) 26 adolescents showing differences between boys and girls.
The present study reveals an inverse association between the risk of ED and academic grades, verbal ability and numeric ability in adolescents. Additional analyses according ED categories showed higher scores in all the academic grades (Math, Language, PE and GPA) among those adolescents without risk of ED. Our results concur with Veses et al. 10, who analyzed a sample of 1,877 Spanish adolescents showing an inverse association between the risk of ED and academic grades. However, Gil García et al. 11 did not find differences in academic performance between Spanish adolescents with different levels of risk of developing ED (at risk vs not at risk). Such inconsistent findings may be due to methodological aspects, such as the different tools used to assess academic performance and ED or the multifactorial character of academic performance.
Previous studies in patients with ED have shown poor academic and cognitive performance 27. For instance, Yanover et al. 28 analyzed a sample of 1,584 college students concluding that eating disturbance may have a negative impact on academic outcomes. In 2011, Sarrar et al. 29 analyzed a sample of 30 patients with anorexia nervosa showing deficits in cognitive flexibility. More recently, Weider et al. 30 found that patients with anorexia nervosa showed lower performance when compared with the control group, on eight out of 13 cognitive measurements. A possible explanation might be that increased levels of metabolic factors (i.e., cortisol serum) in patients with ED alter synaptic function and brain areas closely linked to learning and memory 31, which may impair cognitive functioning 32. In addition, psychological aspects related to ED could affect attitude, interest, motivation and other behaviors linked to academic success which may impair academic performance in adolescents 33,34.
Our analyses indicate that overweight and obese adolescents might present higher risk of ED than non-overweight and obese adolescents in accordance with prior research 13,35,36. This finding highlights the importance of dealing with obesity and ED jointly 12 and could be explained by different mechanisms. Body dissatisfaction and weight-related teasing in overweight and obese adolescents may encourage behaviors associated with weight control, increasing the risk of ED 37,38. In addition, the lack of self-esteem derived from body image dislike might have relevance for the development of ED in overweight and obese adolescents 39.
STRENGTHS AND LIMITATIONS
These results should be interpreted cautiously because of some limitations. First, the cross-sectional design of our study does not allow to infer a causal relationship. A second significant limitation of this research is the small sample size. Despite these limitations, the main strengths of our study comprise the inclusion of potential confounders closely related with weight status and academic performance 10,40. Another strength was the use of the SCOFF questionnaire as an effective screening instrument (before its clinical manifestation) with excellent psychometric properties for the early detection of the risk of ED in Spanish adolescents 20,21. In addition, the SCOFF questionnaire has been validated in different languages making possible to establish comparisons between our data and other studies.
CONCLUSIONS
In conclusion, the results of this study suggest a negative association between the risk of ED and academic performance in adolescents, revealing higher risk of ED in overweight and obese adolescents. Early detection of youth at risk of developing ED is highly necessary not only to avoid academic failure, but also to limit its progression, since adolescents showing problematic eating behaviors are predisposed to ED later in life. Therefore, families, educators and policy makers should consider weight and the risk of ED assessments when designing programs aimed to improve academic outcomes.