Gender stereotypes in eating disorder recognition

One main problem with diagnosing and treating eating disorders (EDs) is that ED recognition rates are low. For example, a 2013 study by Sala and colleagues [1] found that only 41.8% of college students recognized anorexia nervosa (AN), 48.9% recognized bulimia nervosa (BN), and 24.3% recognized binge eating disorder (BED) in a series of vignettes describing individuals with EDs. Suggestion by peers that the individual seeks help from a health professional (i.e., perceived need for treatment) are also low. Sala and others [1] found that 77.1% of college students identified a probable need for treatment in the AN vignette, 37.5% in the BN vignette, and 56.5% in the BED vignette. The rates of low ED recognition and perceived need for treatment among the general population may contribute to the low treatment rates of EDs, where only 20–25% of individuals with EDs receive a diagnosis and/or seek specialized treatment [2]. Given that healthcare providers often fail to recognize EDs, it is particularly important that peers and community members are able to recognize EDs and suggest that treatment may be needed [3]. The low recognition of EDs is problematic because early detection and treatment are needed for a good ED prognosis [4], with treatment response being the greatest in the initial stages of EDs and an individual’s potential response to treatment diminishing as the disorder continues [5].

A challenge to ED recognition is the stereotypes that the general population has about who may physically look like they have an ED [6]. Some of these stereotypes may stem from how EDs are portrayed in the media as affecting White, teenage, Western women [5, 6]. Researchers have found race/ethnicity stereotypes in ED recognition, such that individuals are better able to recognize EDs in White women of higher socioeconomic status (SES) than minority women [6]. Weight stereotypes in ED recognition have also been found, with AN being more likely to be unrecognized or misdiagnosed by healthcare professionals when the presenting patient is of a higher bodyweight compared to underweight [8]. These stereotypes contribute to the low rates of ED recognition, accurate diagnoses, treatment-seeking, and perceived need for treatment in individuals who do not fit ED stereotypes [1, 5]. Although EDs occur in all groups (i.e., regardless of race/ethnicity, weight, and other types of minority status), studies have demonstrated lower rates of ED treatment among more diverse populations (e.g., individuals of non-White race ethnicity, individuals of higher weight, etc.) than among those who are White [5, 8].

However, little research to date has looked at gender stereotypes in ED recognition. While EDs are typically portrayed as affecting White women [6], EDs also occur among those identifying as men as well as non-binary individuals (i.e., individuals identifying outside of the gender binary) [10]. The DSM-5 estimates 10:1 female-to-male ratio [11] in AN. However, a 2007 study by Hudson and colleagues [12] found the number of men struggling from AN and BN to be closer to 25% of all AN and BN cases. Men with AN or BN symptomology report that they have difficulties admitting to having an ED because EDs are often seen as a woman’s problem [13]. The gender ratio for BED is even less skewed, with the prevalence for BED being 1.6% and 0.8% for women and men, respectively [11]. Prevalence rates for those identifying as non-binary are more difficult to determine due to limited research. There has been little research looking at EDs in gender nonconforming individuals, and in these studies, there has been no differentiation in prevalence rates between transgender individuals, or individuals whose gender identity is different from the sex they were assigned at birth, and those who identify as non-binary [14]. In a study examining prevalence of disordered eating in gender minorities (i.e., transgender men, transgender women, or non-binary individuals) researchers found that 26% of participants reported engaging in at least one ED behavior over the past 12 months [15]. In one study conducted at a large treatment center in the United States, 6% of adults identified as gender minorities at admission, a number that is ten times larger than that of gender minorities in the United States (0.6%) [10]. Overall, gender minority individuals appear to be at higher risk for eating disorders compared to their heterosexual peers [16] and experience perceived discrimination when seeking treatment [17].

Only a few studies to date have examined gender stereotypes in ED recognition. In one study, Schoen and colleagues found that college students were more likely to identify other specified and unspecified feeding and eating disorders in female than male characters [18]. In another study, Blackstone and colleagues found that college students were more able to identify AN and BN if the vignette target was a woman rather than a man, but that the gender of the vignette target did not impact ED recognition in BED. In another study, individuals shown the name and description of DSM-5 AN and BN rated both EDs to be feminine disorders [19]. However, these studies did not address the extent to which participants could identify EDs in non-binary individuals vs. women and men, nor differences in perceived need for treatment, perceived distress associated with having an ED, perceived acceptability (e.g., the extent to which it may not be too bad to have an ED), and perceived prevalence between women, men, and non-binary individuals. In addition to examining ED recognition, it is important to examine these other factors. Many individuals find EDs to be acceptable or desirable and not necessarily distressing, likely due to the thinness stereotypically associated with an ED [20, 21]. This is problematic, because even if an individual recognizes an ED but does not feel like it is a distressing problem, or thinks it is even desirable, it may preclude referral to treatment.

Overall, research suggests that EDs are prevalent in men, women, and individuals who identify themselves as non-binary [10, 13]. However, previous research has only examined the extent to which ED recognition differs between men vs. women. No research to date has examined the extent to which ED recognition differs in men/women vs. individuals identifying themselves as non-binary. In addition, no research to date has examined differences in perceived need for treatment, perceived distress, perceived acceptability, and perceived prevalence between women, men, and non-binary individuals. The objective of this study was to examine whether ED symptom recognition, perceived need for treatment, perceived distress, perceived acceptability, and perceived prevalence differ depending on the gender (female, male, non-binary) of the individual described. Participants read three different vignettes describing three individuals’ behaviors corresponding with symptoms of an ED (AN, BN, BED). The participants were randomly assigned to three different experimental conditions (female, male, and non-binary) of the individual in the vignette. Of note, we were unable to include a transgender condition given the difficulty of portraying it in a vignette (e.g., the inability to use pronouns to portray differences). We hypothesized that participants would be more likely to recognize the vignette description as a problem, recognize the ED symptoms, perceive a need for treatment, perceive it as more distressing, report higher acceptability, and perceive higher prevalence for: (1) Vignettes describing women than for vignettes describing individuals identifying themselves as non-binary or as men; (2) Vignettes describing individuals identifying themselves as non-binary than vignettes involving individuals identifying as men. We hypothesized that there would be no gender x diagnosis interactions.

Methods

Participants

Participants were recruited through listservs and social media (Twitter and Instagram). The only inclusion criteria was being 18 years of age or older. Eating disorders or gender were not mentioned in study advertisements. The initial sample consisted of 276 participants. Fifty-one participants were missing data (ranging from 13 to 89% complete). All 51 were removed from the data set because they had not responded to all three vignettes. The final sample consisted of 225 participants: 75 in the non-binary condition, 75 in the male condition, and 75 in the female condition. Participants had the opportunity to enter a drawing to win a $200 gift card for completion of the survey. The study was approved by the institutional review board and informed consent was obtained from all participants.

Measures

Vignette questionnaire Participants read each of the three vignettes (AN, BN, BED) (see Appendix). These vignettes were adapted from previous studies [1] and only included symptoms that a peer may witness. Participants were then asked whether they believed the individual described in the vignette had any problems, what specific problem they had, whether they should seek treatment for the problem, how distressing the problem was, perceived acceptability of the problem, and how many others they thought struggled with the same problem. The questions and corresponding response were also adapted from previous studies [1, 5], and options were as follows:

Problem recognition “Do you think [NAME] has a problem?” accompanied by a 5- point Likert scale with the following answer choices: (1) Definitely Yes, (2) Probably Yes, (3) Not Sure, (4) Probably Not, and (5) Definitely Not.

Specific problem “What issues, if any, do you think [NAME] is struggling with?” accompanied by a free response. We chose to include this item as a free response item instead of a multiple-choice item in order to avoid influencing participants perceptions (e.g., suggesting a possible ED could make participants consider the possibility of an ED when they wouldn’t have otherwise done so). Responses were coded in the following way: (1) being able to identify any eating pathology (i.e., their responses contained key words such as anorexia, bulimia, eating disorder, binge eating disorder, self-starvation, binge/purge, eating/food problem, compulsive eating, etc.); and (2) not being able to identify any eating pathology.

Perceived need for treatment “Do you think [NAME] should seek help?” accompanied by a 5- point Likert scale with the following answer choices: (1) Definitely Yes, (2) Probably Yes, (3) Not Sure, (4) Probably Not, and (5) Definitely Not.

Perceived distress “How distressing do you think it would be to have issues similar to [NAME]?” This question assessed perceived severity and was accompanied by a 5- point Likert scale with the following answer choices: (1) Not distressing at all (2) A little distressing, (3) Distressing, (4) Very Distressing, and (5) Extremely distressing.

Perceived acceptability ‘‘Have you ever thought that it might not be too bad to be like [NAME]? This question assessed perceived acceptability and was accompanied by a 5- point Likert scale with the following answer choices: (1) Definitely Yes, (2) Probably Yes, (3) Not Sure, (4) Probably No, and (5) Definitely No.

Perceived prevalence ‘‘What percentage of individuals in the community do you think might struggle with similar issues as [NAME] at any given point in time?” This question assessed perceived prevalence and was accompanied by a visual analogue scale with a movable slider rating from (0) to (100).

Procedure

Participants were randomly assigned to a gender condition and then completed the demographic questionnaire and the EDE-Q7. Next, participants read the vignettes corresponding to their condition (i.e., female, male, non-binary). In each condition, participants read three vignettes that described an individual with symptoms of: (1) AN; (2) BN, and (3) BED. The order of the AN, BN, and BED vignettes was randomized. Participants completed the set of vignette questions after reading each vignette. They were informed of the gender status of the participants via descriptive text and through the names of the participants and pronouns used.

Statistical analysis

The data were screened for outliers. Outliers were defined as values that were more than 3.29 standard deviations away from the mean of the participant group to which the value belonged. No outliers were identified on any of the dependent variables.

For each of the continuous outcome variables (problem recognition, perceived need for treatment, perceived level of distress, perceived acceptability, and perceived prevalence) separate 3 [Gender condition (between-subjects variable): non-binary vs. male vs. female] × 3 [ED type (repeated measures variable): BN vs. AN vs. BED] mixed ANOVAs were computed. Significant main effects were followed by Tukey-corrected pairwise comparisons. For the categorical outcome variables (i.e., accurately identify the problem described in the vignette as disordered eating), separate Chi-square analyses were conducted for each ED type.

Results

Sample characteristics

Demographic characteristics across the non-binary, male, and female gender conditions are presented in Table 1. Chi-square tests and one-way ANOVAs indicated that the participants in the three conditions did not differ significantly on any demographic variable.

Table 1 Demographic characteristics across participant groups
Full size table

Problem recognition

Contrary to the study hypothesis, there was no main effect of gender condition on problem recognition, F = 1.72, p = 0.18. The gender condition by ED type interaction was also non-significant, p > 0.05. There was a main effect of ED type, F = 18.74, p < 0.001. Participants, regardless of gender condition, rated the issue described in the BN vignette as more likely to be a problem than the issue described in the BED vignette. Participants also rated the issue described in the AN vignette as more likely to be a problem than the issue described in the BED vignette. Problem recognition did not differ significantly between the AN and BN vignettes.

ED recognition

For BN, the proportion of participants who classified the issue described in the vignette as some form of eating pathology (versus other) did not differ significantly across gender conditions, χ2 (2) = 2.98, p = 0.23. Similarly, for AN, the proportion of participants who classified the issue described in the vignette as some form of eating pathology (versus other) did not differ significantly across gender conditions, χ2 (2) = 1.97, p = 0.37. For BED, the proportion of participants who classified the issue described in the vignette as some form of eating pathology (versus other) did not differ significantly across gender conditions, χ2 (2) = 1.63, p = 0.44.

Perceived need for treatment

Contrary to the study hypothesis, there was no main effect of gender condition on perceived need for treatment, F = 0.82, p = 0.44. The gender condition by ED type interaction was also non-significant, p > 0.05. There was a main effect of ED type, F = 16.64, p < 0.001. Participants, regardless of the gender condition they were assigned to, were more likely to refer the individual described for treatment in the AN vignette than in the BN and BED vignettes. Perceived need for treatment did not differ significantly between the BN and BED vignettes.

Perceived level of distress

Contrary to the study hypothesis, there was no main effect of gender condition on perceived level of distress, F = 0.59, p = 0.56. The gender condition by ED type interaction was also non-significant, p > 0.05. There was a main effect of ED type, F = 10.29, p < 0.001. Participants, regardless of the gender condition they were assigned to, rated a higher perceived level of distress for the individual described for the AN vignette than for the BN and BED vignettes. Perceived level of distress did not differ significantly between the BN and BED vignettes.

Acceptability

Contrary to the study hypothesis, there was no main effect of gender condition on acceptability, F = 0.03, p = 0.98. The gender condition by ED type interaction was also non-significant, p > 0.05. There was also no main effect of ED type, F = 1.14, p = 0.32.

Perceived prevalence

Contrary to the study hypothesis, there was no main effect of gender condition on perceived prevalence, F = 1.04, p = 0.36. The gender condition by ED type interaction was significant, F = 2.34, p = 0.05. For the BN vignettes, perceived prevalence was significantly higher for the female condition than for the male and non-binary conditions. For the AN vignettes, perceived prevalence was significantly higher for the female and non-binary conditions than for the male condition. There were no significant differences between gender conditions for the BED vignettes. There was also a main effect of ED type, F = 7.72, p < 0.001.

Participants, regardless of gender condition, rated a higher perceived prevalence for the BN vignette than for the AN vignette. Perceived prevalence did not differ significantly between the AN and BED vignettes or the BN and BED vignettes.

Discussion

The purpose of this study was to examine whether ED recognition, perceived need for treatment, perceived distress, perceived acceptability, and perceived prevalence differed depending on the gender of the individual with the ED. We found that participants rated a higher prevalence of AN in women and non-binary individuals than men as well as a higher prevalence of BN in women than men and non-binary individuals. However, varying the gender of the individual described in the vignettes did not influence ED recognition, perceived need for treatment, perceived distress, perceived acceptability, or perceived prevalence. In addition, we found significant differences in the rates of ED recognition, perceived need for treatment, perceived distress, and perceive prevalence across the three different EDs.

Participants rated a higher prevalence of AN in women and non-binary individuals than men, as well as a higher prevalence of BN in women than men and non-binary individuals. This finding may in part be because of differences in the presentation of EDs in men vs. women [23]. The ED presentation in men is usually muscularity-oriented [24], with men focusing on attaining a more muscular and lean body type and experiencing a low drive for thinness [23]. Men are also more likely to use exercise as a compensatory behavior than women [23]. However, DSM descriptions of EDs are biased towards a female presentation (i.e., focusing on drive for thinness and body dissatisfaction instead of a desire to be more muscular). Until the most recent edition, the DSM even listed amenorrhea as a required criterion for AN, implying that only people who menstruate could have an ED. Furthermore, many current ED assessment tools are over-reliant on items that evaluate stereotypically feminine indicators of ED pathology [25] and EDs are depicted in the media as affecting mainly women [5, 18]. All of these factors may contribute to the perception of an AN and BN as female disorders, reinforcing the idea that it is not likely for men and/or non-binary individuals to have an ED. In non-binary individuals, disordered eating behaviors may be used to attain a body type that is in line with their identified gender [15]. The lower perceived prevalence of AN and BN in men and/or non-binary individuals may also relate to the higher rates of social stigma of EDs seen in non-female populations [17, 19]. The higher perceived prevalence of AN in females and non-binary individuals and the higher prevalence of BN in females is problematic because EDs also affect males and non-binary individuals. As of 2008, the NIMH estimated that roughly one million men struggle with EDs [23] and male EDs make up at least 25% of all AN and BN cases [12]. Furthermore, more and more men are seeking help or being identified in treatment [23]. It is thought that these estimated ED prevalence rates for men are underestimated due to the stigma associated with ED pathology in men and the potential for minimization of symptoms [28]. In contrast, we found that gender did not influence the perceived prevalence of BED. This may be in part be because the gender ratio for BED prevalence is less skewed than that of AN and BN [11].

Results suggest that the gender of the individual portrayed in the vignette did not influence problem recognition, ED recognition, perceived need for treatment, perceived level of distress, or acceptability. Our finding that problem and ED recognition did not differ among gender conditions stands in contrast to other research where participants were more able to identify AN and BN if the vignette target was a woman than a man [7]. This may be because of the difference in demographics of the participant samples; Blackstone and colleagues surveyed only college students. In contrast, our sample consisted of participants with a wider age range, a majority of who had already graduated college and were in the workforce. EDs are more common on college campuses than the larger community [29], which may impact the extent to which college students are able to identify EDs in different populations.

We found differences in the rates of ED recognition, perceived need for treatment, perceived distress, and perceived prevalence across the three different EDs. Participants were more likely to recognize BN and AN as a problem than BED. This finding may be related to low BED awareness, since it is a newer formal diagnosis in the latest version of the DSM [11]. Furthermore, AN and BN symptoms most closely fit the stereotypical image of a thin or underweight individual. In contrast, BED does not fit the perception of a stereotypic ED presentation (e.g., being underweight). Our findings are partially in line with past research where AN was more recognizable than BN and BED [1] as well as another study where BN was more recognizable than AN and BED [7]. We found higher perceived need for treatment and perceived level of distress for AN than BED and BN. This is in line with past research that found higher perceived need for treatment for AN than BED and BN [1]. Participants may have rated AN to be more distressing than BN and BED and were more likely to refer AN to treatment than BN and BED due to the severity of AN. For example, the mortality rate in AN is higher than in BED and one of the highest across all psychiatric illnesses [30]. Participants also rated a higher perceived prevalence for BN than for AN. This may be because population lifetime prevalence of BN (1.9% for women and 0.6% for men) is in actuality higher than that of AN (1.4% for women and 0.2% for men) [31]. In contrast, perceived acceptability did not differ across the three EDs, with participants describing acceptability as, on average, being “unsure if it would be too bad to have these problems” for all three EDs.

Strengths and limitations

One potential limitation is the homogeneity of the sample. Most participants were young White cisgender heterosexual females who were highly educated (e.g., completed at least college). Future studies might expand on these findings or arrive at different results if the study is distributed to a larger and less homogenous group. In addition, the vignettes did not portray any internal dialogue surrounding the disordered eating behaviors such as weight or body image. Although this was intentional as the focus of the study was on recognition of observable behaviors, thoughts related to food and body image are integral aspect of EDs. However, ED cognitions are rarely noticeable to others. Recognition and perceived need for treatment may have improved if the vignettes included the internal experiences consistent with AN, BN, and BED. Relatedly, the vignettes did not greatly vary between conditions to remain consistent. However, not varying the vignettes between conditions may pose as a limitation because of the differences in presentation of EDs in different genders [15, 16]. Another important limitation to consider is that there was no validation conducted for the measures used to assess problem/ED recognition, perceived need for treatment, perceived level of distress, acceptability, and perceived prevalence for these vignettes. However, the measures we used have been used in other studies [1, 5, 6]. Nevertheless, the wording of the questions may have been somewhat vague and may have been misinterpreted by participants. Finally, in describing the non-binary condition, we only stated that the individual identified as non-binary and used they/them/their pronouns, but did not have a definition provided to participants as to what nonbinary means. Some participants may not have known what a nonbinary identity is, particularly because most of the participants in our sample were cisgender.

What is already known on this subject?

The general population has several stereotypes about who may physically look like they have an eating disorder. Previous studies have identified various stereotypes, such as race/ethnicity and weight stereotypes, regarding eating disorder recognition, perceived need for treatment, perceived distress, perceived acceptability, and/or perceived prevalence. There is limited evidence regarding gender stereotypes, with findings from one study suggesting that college students are more able to identify anorexia nervosa and bulimia in women than men.

What this study adds?

Findings suggest that individuals believe that anorexia nervosa is more prevalent in women and non-binary individuals than men, and that bulimia nervosa is more prevalent in women than non-binary individuals and men.