Introduction

Celiac disease (CD) is a chronic immune-mediated systemic disease characterized by inflammation and villous atrophy of the small intestine, with an estimated prevalence of 1:100 in Europe and in the US [1]. A strict, lifelong gluten-free diet (GFD) is currently the only treatment for CD.

Eating disorders (EDs) encompass a group of genetically determined psychiatric conditions, potentially leading to persistent failure to meet nutritional and metabolic needs, and often resulting in severe impairment in psychosocial functioning [2]. The broader term “disordered eating behaviors (DEBs)” encompasses a wide spectrum of eating pathologies, such as emotional eating, strict eating, disinhibited eating, night eating, weight and shape concerns and inappropriate compensatory behaviors (e.g., purging), at a frequency or severity that does not meet the criteria for the diagnosis of EDs. DEBs have a very important clinical significance as they constitute an increased risk of developing a clinical form of an ED [3,4,5], that if identified early could improve outcomes [6]. The prevalence of DEBs in adolescents is estimated as 10% in Western cultures [7]. In Israel, DEBs have been reported in 8.2% of adolescent females and in 2.8% of adolescent males [8].

Celiac disease, as well as other diet-treated chronic diseases, such as cystic fibrosis, type 1 diabetes mellitus and inflammatory bowel disease, necessitates adherence to prescribed dietary restrictions for life. This food restriction may distort attitudes toward eating and weight, and increase the risk of DEBs [9].

There are sparse data on the associations of CD with EDs and even less on that the between CD and DEBs. Studies examining the connection between CD with EDs consisted mainly of case reports [10,11,12,13,14] and two population-based works: one that comprised of only hospitalized CD patients [15], and a second nationwide study that described the link between CD and anorexia nervosa [16]. Reports that specifically explored the possible association of CD with DEBs consisted of structural psychological assessments [17] and/or questionnaires [18, 19]. All of these papers presented a positive connection of various degrees, between eating pathology and CD.

The impact of GFD adherence on DEBs is inconclusive. One study showed an association of increased GFD adherence with more frequent DEBs, particularly with DEBs that predominantly involve concerns about shape and weight [18]. In another study, individuals with eating pathology reported higher rates of non-adherence to their GFD [19]. The aims of the current study were to assess the prevalence and risk factors for DEBs among individuals with CD, and to examine the possibility of an association between adherence to GFD and DEBs.

Methods

Individuals with CD were recruited via the National Celiac Disease Organization network, between January 1 and February 28, 2015. Persons with CD join this organization from their own will. An invitation to participate in the study was sent by email to all individuals registered in the organization, approximately 3000. Those agreeing to participate were directed to an anonymous web-mediated survey. The respondents and in the case of minors under the age of 18 years, their legal guardians were informed that replying to the web-mediated survey provided informed consent to participate in the study. No compensation was offered for the time spent completing the survey. Participants were asked to fill in their demographic data including gender, current age, age at diagnosis of CD, weight, height and the presence of other diseases, as well as two self-rating questionnaires. Individuals aged 12–18 years were included in the analysis. Those under the age of 12 years were excluded to reduce the possibility of misdiagnosis of young children with eating problems that do not concern negative body image views. Those above the age of 18 years were excluded to enable a focus on adolescents and to avoid misdiagnosis of older ages for which loss of appetite might occur regardless of body image issues. To preclude misleading assumptions as to the source of DEBs, individuals who reported comorbidities or chronic illnesses in addition to celiac disease that necessitates a prescribed dietary regimen for a lifelong basis, such as inflammatory bowel disease, irritable bowel syndrome, type 1 diabetes mellitus and cystic fibrosis, were excluded from the analysis.

The Eating Attitudes Test-26 (EAT 26) [20]

This 26-item self-rating scale assesses a person’s pursuit of thinness, dieting behavior and control over eating. It is a key tool in the evaluation and screening of EDs and also DEBs, as it includes items that directly relate to eating behavior [21, 22]. An EAT-26 score of ≥ 20 is suggestive of an ED and/or may indicate disturbed eating behavior. We note that we instructed our respondents that the items presented in this questionnaire (such as “I engage in dieting behavior” or “I avoid food with high carbohydrate content”) do not relate to their diet management due to CD, but to their general tendencies towards food, in regard to losing weight.

The gluten-free diet (GFD) questionnaire [23, 24]

This self-rating questionnaire consists of four questions and is designed to evaluate GFD adherence of persons with CD. The questionnaire is easy to complete. Scores 0–1: the responder does not follow a strict GFD; score of 2: the responder generally follows a GFD but consumes gluten on occasion; score of 3: the responder follows a strict GFD, including declaring he or she has CD when visiting restaurants; score of 4: the responder eats only food that has been approved and labeled by the CD organization as gluten free.

Statistical analysis

Data analysis was performed by SAS for Windows version 9.4. Categorical variables were reported by their relative frequencies and compared by Pearson Chi-square test or Fisher’s exact test. A p value of 0.05 was considered significant. Univariate analysis was used to determine the relationships between two outcome variables and sociodemographic factors. The scores on both questionnaires were analyzed as dichotomous variables, i.e., EAT-26 score as greater or less than 20, and GFD score as 4 vs. 0–3. The continuous variables did not follow a normal distribution and were, therefore, reported by medians and interquartile ranges and compared using the two-sample Wilcoxon test. Differences in DEBs and GFD were analyzed using the two-sample Wilcoxon test, according to participant sex and age.

Ethical approval

The study protocol and its conduction were approved by the institutional review board (IRB) of the “Chaim Sheba Medical Center” in Tel Hashomer, Israel.

Results

Of the approximately 3000 individuals registered in the national CD network, 1215 responded (41%). Of them, 437 (36% of the respondents) fully completed the research questionnaires. Overall, 301 individuals were excluded from the analysis because their ages were not within the range determined for this study (257) or because they had comorbidities that necessitated specific dietary requirements [44]. Of the 136 participants included in the analysis, 93 (68%) were females and 43 (32%) were males. The median time for survey completion was 7.3 min.

The demographic and clinical characteristics of the study group are presented in Table 1. There was no difference between males and females in current age, disease-onset age, disease duration and BMI percentiles.

Table 1 Demographic and clinical characteristics of the study participants
Full size table

Adherence to GFD

Fifteen participants (11%) scored either 1 or 2 on the GFD questionnaire, representing poor adherence. Forty-four (32%) respondents scored 4 on the GFD questionnaire, representing the highest adherence level to GFD (Table 2). Comparing those with strict adherence (GFD grade 4) to participants with lower adherence (GFD grades 1–3), there was no significant difference between their age (p = 0.65), disease duration (p = 0.54) and age at the time of CD diagnosis (p = 0.72). Being overweight (BMI ≥ 85th percentile) was not correlated with adherence to GFD, according to the dichotomous classification (p = 0.4).

Table 2 Adherence of individuals with celiac disease to a gluten-free diet (GFD)
Full size table

Disordered eating behaviors

Twenty-one participants (15%) scored 20 or higher on the EAT-26 questionnaire. Eighteen (19%) females and 3 (7%) males had DEBs scores ≥ 20 (considered pathological); Characteristics that were associated with pathological EAT-26 scores included older current age (p = 0.04), being overweight (BMI ≥ 85th percentile) (p = 0.02) and female gender (p < 0.06) (Table 3). An EAT-26 score ≥ 20 was not correlated with disease duration (p = 0.54) nor with age of diagnosis (p = 0.1). A pathological EAT-26 score did not correlate with GFD adherence; EAT-26 score ≥ 20 was observed in 8/44 (18%) of the respondents with the highest level of adherence to GFD vs. 13/92 (14%) of the less adherent group (p = 0.52).

Table 3 Demographic and clinical characteristics of individuals with celiac disease according to pathological (EAT-26 ≥ 20) vs. normal (EAT-26 < 20) scores on the EAT-26 questionnaire
Full size table

Further analysis revealed that items in the EAT-26 questionnaire concerning dieting, food preoccupation and body image were scored in a higher percentage by those who had eating pathology, e.g., “I am preoccupied with a desire to be thinner” (76%), “I eat diet foods” (71%) and “I feel that food controls my life” (61%). Items of the oral control and binging-purging scales were not as prevalent, e.g., “I particularly avoid foods with high carbohydrate content” (28%) and “I display self-control around food” (38%).

Discussion

In our relatively large cohort, DEBs were found to be more prevalent in individuals with CD compared to the general population, both among males and females. To our knowledge, this is the first study to examine the occurrence of DEBs in adolescents with CD not performed in a hospital setting. Our results strengthen the supposition of an increased prevalence of DEBs in CD, a subject for which available information is scarce [10,11,12,13,14,15,16, 19].

We identified female gender, older age and being overweight as risk factors for DEBs in individuals with CD. These findings are comparable to other studies that investigated risk factors for DEBs in the general population in Israel and in Western countries [8, 19, 25,26,27].

In the current study, the risk for DEBs was 19% in females and 7% in males. We compared our results to those of a large study conducted in our country, which used the same questionnaires and comprised a diverse population of adolescents of the same ages, 12–18 years. In that study, DEB prevalence was 8% and 3% among females and males, respectively [8]. However, we note that this referenced work was published in 2014 and does not state its sampling period, a factor that could make the comparison less relevant due to time trends. Other studies investigated different age groups (25–65 years) [28], used other methods [e.g., the Eating Disorder Inventory (EDI) questionnaire, telephone interviews] [28, 29] or focused on particular populations [29,30,31,32,33,34], so comparison with our findings did not seem relevant. The increased frequency of DEBs occurrence in females and males with CD concurs with other reports [21, 35, 36].

Further analysis of the different items of the EAT-26 questionnaire, revealed that individuals with CD, who had a pathologic EAT-26 score, were mainly characterized with dieting, body image and food preoccupation whereas items reflecting oral control and binging-purging behavior were less common. These findings strengthen our findings that individuals with CD do not develop a specific eating pathology linked to their disease but rather a “common” form of DEB.

The etiologic factors of the development of eating pathology are mostly unknown. They include sociocultural and environmental influences together with a possible genetic predisposition [37, 38]. In an attempt to understand what influences may contribute to the association of CD with DEBs, we considered several causes. First, it is plausible that diet-treated chronic diseases enhance the risk of body image issues, and that these may increase the tendency for DEBs. This has been reported for other chronic illnesses that necessitate adherence to a lifelong dietary regimen, such as inflammatory bowel diseases, irritable bowel syndrome and type 1 diabetes mellitus [9, 39]. In CD, preoccupation with food can occur as a result of the general fear of abdominal symptoms associated with exposure to gluten, by the lack of gluten-free replacements for every food product, and by the uncertainty of gluten presence in food, at public places and social gatherings. Second, there has been evidence that immune-mediated processes, as in CD, may be linked to the establishment of eating pathology [15]. Finally, common molecular patterns have been found in AN, CD [40] meaning it is possible they share a genetic predisposition. Considering the above, we expected that higher GFD adherence would be associated positively with DEBs. Rather, we found no statistically significant difference in DEB occurrence among the “strict adherers” (GFD 4) compared to the other respondents (GFD 1–3). There are a number of possible explanations for our findings. First, in recent years, increased awareness to CD has facilitated accessibility to gluten-free products, and their variety has expanded. This might decrease the general preoccupation with food, which was associated with CD in the past. Second, “strict adherence” to a GFD may reduce anxiety levels and increase general psychosocial functioning. This was suggested by other studies that presented comparable findings [14, 18]. Third, our study group comprised a relatively small number of “non-adherers” (12.4%). In other studies of pediatric populations with CD, rates of non-compliance were similar to our results [41,42,43].

Interpretation of the findings reported here should take into consideration the limitations of the study design. Although surveys conducted via the internet have been shown to be as reliable as paper and pencil ones [44], a selection bias could arise due to non-representative characteristics of persons with CD who agreed to participate in the online survey. It is possible, for example, that those who answered are stricter about maintaining a gluten-free diet, and also more prone to developing disordered eating when requiring a restricted diet. However, we lacked information regarding the non-respondents of the survey, which would have helped verify this. Moreover, weight and height were reported and not measured. However, a high correlation between the validity of self-reported height and measured height has been described [45]. Lastly, the study design enabled the administration of only self-rating questionnaires and not standardized interviews.

Our study has, nevertheless, some important advantages. It is a hypothesis-generated study, and comprised a relatively large number of participants, using well-accepted tools. Moreover, the anonymous nature of the internet enabled communication with people who might have been hesitant to express themselves openly in a face-to-face or telephone interview [46].

In conclusion, this study should raise awareness for medical teams to the importance of closely monitoring adolescents with CD for DEBs, especially so if female, overweight or older; and if diagnosed with CD at a later age, regardless of their GFD adherence tendencies. Since individuals with DEBs are at increased risk of developing a clinical form of an ED, early identification and intervention may improve outcome.