Dietary behaviors are developed rapidly from infancy to adolescence, and they are influenced not only by individual factors and household characteristics [1
], but also by parent–child interactions and social interactions [2
]. In the adolescence period, dietary behaviors may be crucial, as they contribute to further behaviors in adulthood and to the resultant health risks [3
] associated with body mass [4
] and dieting [5
]. It is possible that appetitive traits may influence dietary behaviors [6
]. Food approach traits are associated with eating onset dietary behaviors, whereas food avoidance traits are associated with food offset dietary behaviors [7
]. Appetitive traits are defined as a set of persistent predispositions toward food [8
] that interact with environmental factors and influence dietary behaviors and their consequences [6
]. Appetitive traits are known to possess a strong genetic component [10
], and they can be perceived as stable traits [11
]. They are commonly measured in adults using the Adult Eating Behavior Questionnaire (AEBQ) developed by Hunot et al. [9
], which has been newly validated in adolescents [12
]. However, there is no Polish version of the AEBQ validated for adolescents, and to the best of our knowledge, no study completed with a Polish version of the AEBQ has been published thus far.
The lack of a reliable tool to measure appetitive traits in a Polish population of adolescents may be challenging, as such a tool may allow in-depth analysis of dietary behavior determinants. In the study conducted by Syrad et al. [7
], associations between appetitive traits and young children’s eating patterns were studied using the Child Eating Behavior Questionnaire [13
], from which the AEBQ was developed. The findings of this study emphasized that the behavioral expression of appetitive traits has been linked with the risk of obesity.
Appetitive traits have been associated with body mass index (BMI) in children and adults in numerous studies [6
]. Food approach subscales are generally positively correlated with BMI, whereas food avoidance subscales are negatively correlated with BMI [6
]. Because these traits can be measured across the life course and they have been shown to be continuous and stable traits [11
], there is a need for longitudinal studies to examine the continuity and stability of appetitive traits across the lifespan [6
Appetitive traits have been shown to be influenced by sex, as has been observed in both adults [16
] and children [17
]. This may be associated with the fact that females differ from males in appetitive traits that might make them more susceptible to eating in response to external food cues or less susceptible to feedback from internal satiety cues [18
]. Such problems are especially important during adolescence, where females have been found to be more sensitive to internal satiety cues, tend to eat more slowly, and show higher levels of emotional over-eating than males [12
]. Therefore, the possibility of assessing appetitive traits in adolescents may be an important public health goal, while also analyzing dietary behaviors.
Among the factors that may have influenced dietary behaviors and appetitive traits was the period of outbreak of coronavirus-19 disease (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which affected all aspects of daily life [19
]. A number of studies conducted during this period revealed changes in various dietary aspects, including overeating predispositions toward food. In the study conducted by Di Renzo et al. [20
], the authors found that the consumption of homemade products and dishes increased, even for products that may have been purchased at the grocery store. Similarly, the study by Sidor and Rzymski [21
] revealed that the majority of respondents reported eating more. Furthermore, in the study conducted by Phillipou et al. [22
], it was observed that both restricting and binge eating behaviors were increased. Changes in eating and physical activity behavior were found to be influenced by the COVID-19 pandemic, as demonstrated by overeating behavior in a British adult population [23
], while in an Australian adult population, 53.6% individuals reported overeating over the past two weeks [24
]. Moreover, higher scores for food approach subscales and lower scores for food avoidance subscales may be associated with a tendency toward higher food frequency and higher food consumption, respectively, during eating occasions [10
A number of studies conducted during the COVID-19 pandemic have revealed important determinants that might influence the dietary behaviors of adults [19
], but thus far, no such studies in a population of adolescents have been published. Despite the fact that the determinants of dietary behaviors are generally numerous and complex [1
], such dietary behaviors in adults during the COVID-19 pandemic can be explained by two major determinants: lockdown and stockpiling food at home [20
]. Both determinants were also experienced by children and adolescents in this period, and may have also influenced their dietary behaviors.
Therefore, there is a need to validate the Polish version of the AEBQ for adolescents to measure appetitive traits in a population of adolescents during the COVID-19 pandemic. Thus, the present study aimed to validate the AEBQ in a population-based sample of Polish secondary school students and to assess differences in appetitive traits between boys and girls within the Polish Adolescents’ COVID-19 Experience (PLACE-19) Study.
Descriptive characteristics of the study sample of Polish adolescents within the PLACE-19 Study are presented in Table 1
The results of the test–retest reliability of the Polish version of the AEBQ are presented in Table 2
. The analysis of weighted κ statistic values indicated fair to substantial agreement. According to the Landis and Koch [43
] criteria, values ≤ 0 indicated no agreement, and values of 0–0.20, 0.21–0.40, 0.41–0.60, 0.61–0.80, and 0.81–1.0 indicated slight, fair, moderate, substantial, and almost perfect agreement, respectively. For almost all subscales, at least moderate agreement was observed, whereas for Satiety Responsiveness, fair agreement was noted.
The results of standardized factor loadings within the CFA with WLS obtained for the Polish version of the AEBQ are presented in Table 3
. The results of standardized factor loadings within the CFA with WLS obtained for boys for the Polish version of the AEBQ are presented in Supplementary Table S2
. The results of standardized factor loadings within the CFA with WLS obtained for girls for the Polish version of the AEBQ are presented in Supplementary Table S3
Mean scores obtained for the AEBQ subscales and McDonald’s ω in this study of Polish adolescents are presented in Table 4
. The scores obtained for items of the AEBQ subscales are presented in Supplementary Table S4
. The analysis of McDonald’s ω values indicated fair to substantial agreement [36
Sex invariance analyses of the AEBQ in the study of Polish adolescents are presented in Table 5
. The CFA showed good model fit, with χ2
= 4826.105 (degrees of freedom (df) = 384), RMSEA = 0.069 (90% confidence interval (CI): 0.067, 0.070), CFI = 0.90, and SRMR = 0.08. The results revealed that, compared to the configural invariance model, the metric invariance model did not result in significantly decreased model fit, with ΔCFI = −0.002 and ΔRMSEA = −0.001, which were lower than the recommended cutoffs of 0.010 and 0.015, respectively. The scalar invariance model also did not result in significantly decreased fit of the model over the metric invariance model, with ΔCFI = −0.005 and ΔRMSEA = 0.000. This finding suggested a lack of response bias between boys and girls and allowed comparison of factor means across boys and girls.
Results from Table 6
show the comparison between mean AEBQ subscale scores for girls and boys in the study. The results revealed statistically significant differences, as girls reported higher levels of Food Responsiveness (p
< 0.0001), Emotional Over-Eating (p
< 0.0001), Satiety Responsiveness (p
< 0.0001), Emotional Under-Eating (p
< 0.0001), and Slowness in Eating than boys (p
< 0.0001). Their total AEBQ scores were also higher than those for boys (p
Correlations between AEBQ subscales in the sample of Polish adolescents are presented in Table 7
. Positive inter-correlations were observed between all food approach subscales. Positive inter-correlations were also observed between the majority of food avoidance subscales, except for Food Fussiness and Emotional Under-Eating, as their correlation was negative, and except for Food Fussiness and Slowness in Eating, as their correlation was not statistically significant. Within the correlations between food approach subscales and food avoidance subscales, both positive and negative correlations were observed, while Emotional Under-Eating correlated with all food approach subscales. The total AEBQ score was also positively correlated with the scores for all of the subscales.
Correlations between AEBQ subscales in the sample of Polish boys are presented in Supplementary Table S5
. Positive inter-correlations were observed between all food approach subscales. Positive inter-correlations were also observed between the majority of food avoidance subscales, except for Food Fussiness and Satiety Responsiveness, Food Fussiness and Emotional Under-Eating, and Food Fussiness and Slowness in Eating, as their correlations were not statistically significant. Within the correlations between food approach subscales and food avoidance subscales, both positive and negative correlations were observed, while Emotional Under-Eating was correlated with all food approach subscales. The total AEBQ score was positively correlated with the scores for all of the subscales.
Correlations between AEBQ subscales in the sample of Polish girls are presented in Supplementary Table S6
. Positive inter-correlations were observed between all food approach subscales. Positive inter-correlations were also observed between the majority of food avoidance subscales, except for Food Fussiness and Emotional Under-Eating, as well as Food Fussiness and Slowness in Eating, as their correlations were negative, and except for Food Fussiness and Satiety Responsiveness, as their correlation was not statistically significant. Within the correlations between food approach subscales and food avoidance subscales, both positive and negative correlations were observed, while Emotional Under-Eating was correlated with all food approach subscales. The total AEBQ score was also positively correlated with the scores for all of the subscales.
In the present study, a seven-factor, 30-item structured AEBQ without the Hunger items showed a satisfactory model fit. The results revealed that, compared to the configural invariance model, the metric invariance model did not result in significantly decreased model fit. The scalar invariance model also did not result in significantly decreased fit of the model over the metric invariance model. While comparing boys and girls, it was observed that girls reported higher levels of Food Responsiveness, Emotional Over-Eating, Satiety Responsiveness, Emotional Under-Eating, and Slowness in Eating than boys, while their total AEBQ scores were also higher. Positive inter-correlations were observed between all food approach subscales, as well as between Emotional Under-Eating and all food approach subscales for girls, boys, and the total sample; positive inter-correlations were also observed between the majority of food avoidance subscales.
The development study for the AEBQ [9
], the study conducted by Hunot-Alexander et al. [12
] for adolescents, and other studies, including the study of Mallan et al. [6
], suggested that the Hunger subscale could be excluded from the AEBQ; hence, this approach was also chosen in the present study. Similarly, as reported in other previous studies [6
], the original factor structure with the eight-factor model was compared with a seven-factor solution (excluding the Hunger subscale). The comparison revealed that the seven-factor model was suitable for the studied population of Polish adolescents. This finding corresponds with the results of Hunot-Alexander et al. [12
], who showed that in their population of adolescents, the model without the Hunger subscale exhibited better results than those including this subscale. Thus, it can be stated that the reliability and validity of the AEBQ may be improved by removing the Hunger subscale.
Hunot-Alexander et al. [12
] studied a sample of adolescents aged 11 to 18 years who were recruited from secondary schools in London; their results revealed similar associations between appetitive traits assessed by the AEBQ, as well as differences between girls and boys. In this study of English adolescents, the food approach subscales were positively inter-correlated, while Hunger was also positively correlated with Emotional Under-Eating [12
]; a similar result was observed in the studied sample of Polish adolescents in the present study. The differences observed between girls and boys were also similar in both populations; in the study of English adolescents, girls showed higher scores for Emotional Over-Eating, Satiety Responsiveness, and Slowness in Eating [41
], which was similar to that noted in the studied population of Polish adolescents. In the present study, girls showed higher scores for Food Responsiveness and Emotional Under-Eating, and similar to the study of Hunot-Alexander et al. [12
], boys did not obtain higher scores for any of the subscales.
Hunot-Alexander et al. [12
] also presented similar observations associated with correlations between appetitive traits assessed by the AEBQ, as well as differences between female and male respondents, and in this study, all of the food approach subscales were positively inter-correlated. In the study of Zickgraf and Rigby [16
], no significant sex-dependent differences were observed for other subscales. However, it is not surprising that observations are inconsistent, as the abovementioned study of Zickgraf and Rigby [16
] was conducted in a mixed sample of adolescents and adults, who were defined as patients pursuing bariatric surgery.
A comparison of the results of the AEBQ obtained for adolescents in the present study and in studies by other authors [12
] with the results obtained for adults showed similarities in terms of correlations observed between appetitive traits [6
], including Food Responsiveness and Satiety Responsiveness (correlation coefficient of 0.51), Satiety Responsiveness and Slowness in Eating (0.48) [6
], Enjoyment of Food and Food Responsiveness (0.39) [14
], Satiety Responsiveness and Slowness in Eating (0.38), and Emotional Under-Eating and Satiety Responsiveness (0.33) [15
], and differences between scores obtained by female and male respondents [14
]. Considering this aspect, it can be assumed that such observations may be typical for various populations.
In the present study, the AEBQ was administered during the unique period of the COVID-19 pandemic. It must be emphasized that the COVID-19 situation has a significant influence on mental health, as depression, anxiety, and stress are commonly observed psychological responses to the COVID-19 issue [44
]. This also indicates that the COVID-19 pandemic may have a substantial effect on adolescents’ mental health, as a poll conducted by United Nations International Children’s Emergency Fund (UNICEF) showed that 27% of adolescents reported anxiety and 15% reported depression associated with the current COVID-19 issue [45
Moreover, the COVID-19 pandemic may be associated with other problems related to the general interruption of routines and social interactions, which influence not only mental health [46
], but also dietary behaviors and emotional eating [47
]. Similarly, panic-buying or stockpiling of long-life foods, such as flour, sugar, dried pasta, rice, biscuits, and bottled and canned foods, may also contribute to increased consumption and overeating [24
]. Last but not least, sedentary behaviors and a decrease in physical activity during the COVID-19 period should be considered, as they may also promote increased consumption resulting from food cravings [49
The differences associated with sex and higher scores for female respondents than for male respondents for various AEBQ subscales may be explained by the general sex-dependent differences in food attitude. These observations indicate emotional eating, as confirmed by several studies, wherein the association between depression and emotional eating was found to be restricted to female respondents or was much stronger for female respondents than for male respondents [50
]. Thus, it can be considered that in the present study, female respondents showed higher levels of Emotional Over- and Under-Eating than male respondents in response to stress associated with the epidemiologic situation. These results are consistent with general observations, which indicate that, especially for adolescents, there is an important association between perceived stress, worries, tension, anxiety, and resultant emotional eating, which is observed in girls but not in boys [51
]. These observations are also noted in children, for example, in the International Study of Childhood Obesity, Lifestyle, and the Environment (ISCOLE) that studied a sample from the United Kingdom, girls showed a higher tendency for emotional eating than boys [52
]. A comparison of sex-dependent dietary behaviors revealed that females have higher emotional susceptibility to disinhibition, but they also show a higher level of eating-related self-determined motivation than males [53
]. Thus, women have a higher tendency toward overeating [54
] and undereating [55
], and these behaviors are transferred from parents to their daughters [56
]. In an American study of Striegel-Moore et al. [57
] conducted in a group of health organization members aged 18 to 35 years, a higher frequency of eating disorder symptoms was found in female patients than in male patients; furthermore, females diagnosed with binge eating disorders reported significantly higher body image dissatisfaction and drive for thinness than males [58
]. This situation is explained by a cultural expectation of thinness in women [59
], which results from internalized appearance standards [60
] and causes their weight-related concerns [61
]. It may also be associated with the influence of ovarian hormones [62
] and menstrual cycle [63
]; the mid-luteal phase increases emotional eating (as a result of ovarian hormone effects) [64
], which may impact the associations for women.
In the analysis of results of the present study obtained using the AEBQ, it must be emphasized that this questionnaire minimizes participant burden when compared with similar questionnaires [16
]. Moreover, the AEBQ measures eight appetitive traits, while other questionnaires, such as the Dutch Eating Behavior Questionnaire (DEBQ) [65
] or the Three-Factor Eating Scale (TFEQ) [66
], measure only two or three appetitive traits. In the present study, similar to other studies that analyzed the AEBQ in various populations [6
], some significant associations were noted between the analyzed appetitive traits; this may result from the fact that the assessed appetitive traits are associated with each other, as they are influenced by individual characteristics and emotional responses. Macht [67
] indicated that emotional overeating and emotional undereating are related to changes in eating behavior in response to emotional stress, including emotional control of food choice, emotional suppression of food intake, impairment of cognitive eating controls, eating to regulate emotions, and emotion-congruent modulation of eating.
Despite the fact that the present study is the first to use the AEBQ in a Polish population of adolescents and includes validation, some limitations of the study must be listed. The study was conducted during the COVID-19 pandemic, and the results may have been influenced by this specific period. Moreover, the study did not assess some potential confounders that may have influenced the obtained results, such as sociodemographic and lifestyle characteristics, household dietary practices, and health status, including diet-related diseases.