Authors: Jason M. Nagata (Department of Pediatrics, University of California, San Francisco, San Francisco, California, USA), Karen Li (Department of Pediatrics, University of California, San Francisco, San Francisco, California, USA), Angela E. Kim (Department of Pediatrics, University of California, San Francisco, San Francisco, California, USA), Iris Yuefan Shao (Department of Pediatrics, University of California, San Francisco, San Francisco, California, USA), Christopher D. Otmar (Department of Pediatrics, University of California, San Francisco, San Francisco, California, USA), Kyle T. Ganson (Factor‐Inwentash Faculty of Social Work, University of Toronto, Toronto, Ontario, Canada), Alexander Testa (Department of Management, Policy and Community Health, University of Texas Health Science Center at Houston, Houston, Texas, USA), Jinbo He (Division of Applied Psychology, School of Humanities and Social Science, The Chinese University of Hong Kong, Shenzhen, Guangdong, China), Orsolya Kiss (Center for Health Sciences, SRI International, Menlo Park, California, USA), Jason M. Lavender (Military Cardiovascular Outcomes Research Program (MiCOR), Department of Medicine, Uniformed Services University of the Health Sciences, Bethesda, Maryland, USA; The Metis Foundation, San Antonio, Texas, USA), Fiona C. Baker (Center for Health Sciences, SRI International, Menlo Park, California, USA; School of Physiology, University of the Witwatersrand, Johannesburg, South Africa)
Categories: Regular Articles, adolescent, eating disorders, gender identity, gender minority, LGBTQ, transgender
Source: The International Journal of Eating Disorders
Doi: 10.1002/eat.24317
Authors: Jason M. Nagata, Karen Li, Angela E. Kim, Iris Yuefan Shao, Christopher D. Otmar, Kyle T. Ganson, Alexander Testa, Jinbo He, Orsolya Kiss, Jason M. Lavender, Fiona C. Baker
To assess associations between multiple dimensions of gender diversity with eating disorder symptoms in a national cohort of U.S. early adolescents.
This cross‐sectional study utilized data from the Adolescent Brain Cognitive Development (ABCD) Study (N = 10,092, M ~ age ~ = 12.9 years, 2019–2021). Gender diversity was measured using multiple dimensions, including categorical gender identity (e.g., transgender, cisgender), categorical and continuous felt gender (congruence between gender identity and assigned sex), ordinal gender non‐contentedness (dissatisfaction with one's gender), and ordinal gender expression (communication of gender through appearance and mannerisms). Multivariable logistic regression models were used to analyze the associations among gender diversity measures and eating disorder symptoms, adjusting for potential confounders.
Greater felt gender diversity was associated with self‐worth tied to weight (OR 1.30, 95% CI 1.11–1.53), binge eating (OR 1.24, 95% CI 1.06–1.46), and distress with binge eating (OR 1.32, 95% CI 1.09–1.59). Greater gender expression diversity was associated with self‐worth tied to weight (OR 1.16, 95% CI 1.02–1.33), distress with binge eating (OR 1.26, 95% CI 1.04–1.51), and characteristics of binge eating episodes (OR 1.33, 95% CI 1.06–1.66). Gender non‐contentedness was associated with self‐worth tied to weight (OR 1.38, 95% CI 1.20–1.58) and compensatory behaviors related to weight gain (OR 1.12, 95% CI 1.01–1.26). Transgender identity was not significantly associated with any eating disorder symptoms.
We found that greater gender diversity across multiple dimensions was associated with various eating disorder symptoms, and that measures beyond binary gender identity may be important to assess gender diversity in early adolescence.
Summary This study highlights the importance of studying multiple aspects of gender diversity in young adolescents and how various aspects of gender are associated with eating disorder symptoms.Results indicate that gender diverse adolescents may be more vulnerable to experiencing a variety of eating disorder symptoms, supporting the need for clinical and public health initiatives tailored to this population.
Eating disorders affect adolescents and young adults worldwide and present a growing public health concern (Silén and Keski‐Rahkonen 2022). A recent review estimated that the lifetime prevalence of all DSM‐5 eating disorder diagnoses was 5.5%–17.9% in young women (< 30 years) and 0.6%–2.4% in young men (Silén and Keski‐Rahkonen 2022). The mean age of onset is about 12.5 years for anorexia nervosa, bulimia nervosa, and binge eating disorder (Swanson et al. 2011). Although many studies have examined eating disorders in adolescents, few have investigated the relationship between gender diversity and eating disorders.
Studies of transgender and gender diverse adults indicate elevated rates of eating disorder diagnosis and disordered eating behaviors. A review of recent literature found that 20%–50% of transgender and gender diverse adults report engaging in disordered eating, and over 30% screen positive for eating disorder symptoms (Keski‐Rahkonen 2023). Several studies with college samples found that, compared to cisgender and sexual minority students, transgender students had higher rates of self‐reported anorexia nervosa, bulimia nervosa, and eating pathology‐specific academic impairment, as well as higher odds of engaging in compensatory behaviors such as diet pills, vomiting, or laxatives (Diemer et al. 2015; Simone et al. 2020). When examining possible explanations for this heightened vulnerability, one study found that discrimination trauma and internalized transphobia were significant predictors of eating pathology in transgender and nonbinary adults (Urban et al. 2023). Similarly, another study found that higher rates of harassment and discrimination experienced by transgender youth were linked to higher odds of engaging in binge eating and compensatory behaviors in the past year (Watson, Veale, and Saewyc 2017). Disproportionate rates of victimization and violence experienced by transgender and gender diverse adults may also play a role, as sexual assault is significantly associated with increased risk for disordered eating behaviors among gender minority young adults (Keski‐Rahkonen 2023; Eisenstadt et al. 2023). Taken together, studies in adults have generally shown links between gender diversity and eating disorders; however, literature in adolescents and youth is more limited.
Many youth encounter formative experiences related to their gender identity during adolescence, and gender diverse youth are disproportionately affected by eating disorders compared to their cisgender peers (Nagata, Ganson, and Austin 2020). Prior literature has proposed social stigma, lack of social support, and gender minority stress as possible factors contributing to eating disorder vulnerability in this population (Diemer et al. 2018). Moreover, gender diverse youth may experience unique issues surrounding body image, weight, or size secondary to gender dysphoria, manifesting in symptoms not captured by conventional assessments (Nagata, Ganson, and Austin 2020). Body dissatisfaction in particular may be a significant stressor contributing to eating disorders in transgender individuals (Nagata, Ganson, and Austin 2020).
Few studies on eating disorders in adolescents have explored gender identity using more nuanced approaches, instead commonly relying on binary or categorical distinctions (e.g., transgender vs. cisgender). Notably, Potter et al. (2021) found that 39.5% of participants in the Adolescent Brain Cognitive Development (ABCD) Study ages 9–10 (baseline) did not understand the meaning of the term “transgender.” Indeed, not identifying as transgender does not preclude having qualities that are gender diverse, and having gender diverse qualities does not necessitate identifying as transgender (Burke 2021). Adolescents may identify more closely with other terms, such as “gender‐fluid” and “gender‐expansive,” or they may not resonate with labels at all (Nagata et al. 2024). As such, approaching gender from many angles (expression, behavior, perception, dysphoria) rather than asking participants directly if they are transgender may be a more fruitful approach for studying this age range (Potter et al. 2021). Moreover, the developmental period of early adolescence (10–15 years of age) remains relatively understudied in general compared to adolescence (10–20 years of age) (Blum et al. 2014). Considering that 12.5 years is the average age at which eating disorder symptoms emerge, focusing on ages 10–15 may provide insight into early warning signs, enabling caregivers and providers to offer support sooner (Swanson et al. 2011).
A small but growing literature has addressed eating disorders in transgender or gender non‐conforming adolescents. One study of 91 transgender and non‐binary young people aged 8–22 years found that 40% engaged in at least one disordered eating behavior and 17% engaged in three or more disordered eating behaviors; however, there was no comparison group (Pham et al. 2023). Another study of transgender and nonbinary adolescents (mean age 15.5 years) reported notable rates of disordered eating behaviors among transmasculine adolescents (15% for binge eating), nonbinary adolescents (10% for caloric restriction), and transfeminine adolescents (2% for laxative use) (Roberts et al. 2022). We build upon these prior studies by focusing on early adolescence, analyzing a large national sample that reflects a broad gender spectrum, and assessing gender diversity across several ordinal, continuous, and categorical measures and domains.
The purpose of the current study was to examine the relationship between gender diversity across multiple measures and eating disorder symptoms using data from a large‐scale, diverse, longitudinal cohort study of early adolescents from across the United States. We hypothesized that greater gender diversity across ordinal, continuous, and categorical measures would be associated with greater eating disorder symptoms among early adolescents.
The current study utilized data from the Adolescent Brain Cognitive Development (ABCD) Study, a longitudinal cohort study of 11,875 adolescents from 21 research sites across the United States (baseline 2016–2018). Data from the Year 3 ABCD 5.1 release (2019–2021) were utilized as it was the most recent year with full exposure and outcome data. Additionally, eating disorder and gender diversity prevalence are expected to increase during early adolescence, so we chose the oldest adolescent age range with full data available. Informed consent was obtained from caregivers, and assent was secured from each participating adolescent. The University of California, San Diego institutional review board (IRB) approved the study, and local IRB approval was granted at each participating site. Further details on the structure and recruitment of participants are outlined elsewhere (Garavan et al. 2018). Individuals with missing gender or eating disorder data were excluded, resulting in a final dataset of 10,092 adolescents.
Gender diversity was assessed using two ordinal measures (gender expression and gender non‐contentedness), two categorical measures (transgender identity and felt gender), and one continuous measure (felt gender) of gender diversity. Categorical measures were coded such that higher scores indicated self‐identification of being gender diverse. The three continuous measures were based on the core gender constructs of the ABCD study's Youth Gender Survey, with higher scores reflecting higher gender conformity with the adolescent's sex assigned at birth. These continuous measures were then reverse‐coded such that higher scores indicate higher gender diversity. The Youth Gender Survey variables demonstrated adequate internal consistency in the ABCD Study sample (αs = 0.76 to 0.79) (Potter et al. 2022).
Felt gender was assessed as both a categorical measure and a continuous measure. For the categorical measure, responses to the questions “How much do you feel like a boy?” and “How much do you feel like a girl?” were coded according to the 2‐STEP definition of felt gender (A. Potter et al. 2021). This novel definition of felt gender was proposed using the ABCD Study and has been validated for use in early adolescents (Potter et al. 2021). Responses were given on a 5‐item Likert‐type scale from 1 (“Not at all”) to 5 (“Totally”). The 2‐STEP method is the recommended protocol for collecting gender identity data according to professional organizations such as the World Professional Organization for Transgender Health (Deutsch et al. 2013). Participants were classified into four groups (Cisgender, 1‐STEP, 2‐STEP, Minority) that increasingly move away from gender congruence. Detailed coding practices are outlined in Table A. Scores were then reverse coded to ensure consistent directionality such that higher scores indicate higher gender diversity and ranged from 1 to 5.
Gender expression was assessed as an ordinal measure based on responses to the question “How much have you dressed or acted as a girl during play?” for adolescents assigned male at birth. A similar set of questions was asked to those assigned female at birth (“How much have you dressed or acted as a boy during play?”). Responses were given on a 5‐item Likert‐type scale from “Always” to “Never” and reverse‐coded such that higher scores indicate higher gender diversity.
Gender non‐contentedness was measured as an ordinal measure based on the responses to the question “How much have you had the wish to be a girl?” for adolescents assigned male at birth. The question “How much have you had the wish to be a boy?” was asked to adolescents assigned female at birth. Responses were given on a 5‐item Likert‐type scale from “Always” to “Never” and reverse‐coded such that higher scores indicate higher gender diversity.
Transgender identity was assessed as a categorical variable using responses to the survey question “Are you transgender?” Responses were provided on a three‐point scale, with 1 representing “Yes,” 2 representing “Maybe,” and 3 representing “No.” Responses of “I don't know” or “Refuse to answer” were coded as missing.
The ABCD Study uses the Kiddie Schedule for Affective Disorders and Schizophrenia‐5 (KSADS‐5), with parents providing responses about the adolescents, to assess eating disorder symptoms. The KSADS‐5 has been shown to have strong reliability and good convergent validity against established clinical rating scales for assessing adolescent mental health (Townsend et al. 2020). This tool assesses the frequency, duration, and characteristics of eating disorder symptoms, and specific questions can be found in Table B. Specific eating disorder variables of interest that were evaluated were fear of obesity, self‐worth tied to weight, compensatory behaviors related to weight gain, presence of binge eating, distress with binge eating, frequency of binge eating, and specific characteristics of binge eating episodes. In particular, specific characteristics of binge eating episodes and the presence of binge eating were assessed separately in the KSADS‐5. These eating disorder variables were selected based on DSM‐5 criteria and represented symptoms that are of greatest relevance to adolescents (Cheng et al. 2023; Rozzell et al. 2019, 9). All eating disorder variables were analyzed as binary variables, with 1 representing the presence of the symptom and 0 representing the absence of the symptom. These cutoffs were proposed by the KSADS‐5 proprietary algorithm and were already dichotomized in the data released by the ABCD Study.
This study used age, sex assigned at birth, race/ethnicity, parental education, household income, and research site as key covariates. These characteristics were selected based on their potential associations with both adolescent gender identity and eating disorder symptoms (Kaltiala et al. 2023; Nagata, Smith‐Russack, et al. 2023b). Age was coded as a continuous variable. Sex at birth was categorized into “male” and “female.” Race/ethnicity was categorized as a six‐level variable with categories being White, Latino, Black, Asian, Native American, and Other. Parental education was defined as a binary variable with categories of “high school or less” and “more than high school.” Finally, household income was categorized into six levels, with the lowest category being a yearly income of less than 200,000 per year. All variables were collected from the baseline parental report of the ABCD Study. The research site was also included to control for any regional variations in the data.
Statistical analyses were performed using R version 4.3.3. Descriptive statistics, including percentages, means, and standard deviations, were calculated. Multivariable logistic regressions were fit using the survey package version 4.4‐2 which was used to evaluate the associations between gender diversity and eating disorder symptoms. Each regression model included covariate adjustment for age, sex assigned at birth, race/ethnicity, parental education, household income, and research site. Assumptions of logistic regression were checked, including a binary outcome, little or no multicollinearity (all variance inflation factors < 2), no extreme outliers or influential observations (all Pregibon's dbeta < 1), linearity of the independent variable and log odds (Box–Tidwell test p > 0.05 indicating no significant nonlinear terms), independent observations, and a large sample size. In sensitivity analyses, we examined the interaction between gender diversity and household income (binary cutoff at $75,000, which approximates the median US household income) and gender diversity and race (binary White vs. non‐White) on eating disorder symptoms. Additionally, to align our sample with demographic profiles represented in the American Community Survey and enhance the representativeness of our findings, all descriptive statistics and regression models incorporated the ABCD propensity scores (Heeringa and Berglund 2020).
In our sample of 10,092 adolescents (M
~
age
~ = 12.9 years, SDage = 1.0), 48.3% were assigned female at birth and 45.6% were from racial/ethnic minority groups (Table 1). Overall, early adolescents included in this sample reported high gender conformity as indicated by the lower scores for felt gender (1.24), gender expression (1.36), and gender non‐contentedness (1.23) (Table 1). The prevalence of each eating disorder variable is presented in Table 2, with characteristics of binge eating episodes having the highest prevalence of 14.0% and fear of obesity having the lowest at 1.0%.
The adjusted associations between various gender measures and eating disorder symptoms are shown in Table 3. Unadjusted associations are reported in Table C. We found significant associations between felt gender and self‐worth tied to weight (OR 1.30, 95% CI 1.11–1.53), binge eating (OR 1.24, 95% CI 1.06–1.46), and distress with binge eating (OR 1.32, 95% CI 1.09–1.59).
There also were significant associations found between felt gender individual categories and self‐worth tied to weight (OR 1.75, 95% CI 1.08–2.86), distress with binge eating (OR 2.20, 95% CI 1.22–3.98), and characteristics of binge eating (OR 2.42, 95% CI 1.27–4.62). Finally, significant associations were found between minority group felt gender adolescents and self‐worth tied to weight (OR 1.75, 95% CI 1.17–2.60), binge eating (OR 1.66, 95% CI 1.07–2.56), and distress with binge eating (OR 1.92, 95% CI 1.14–3.22). All of these associations were positive, indicating that more gender diverse identities were associated with higher odds of eating disorder symptoms.
Significant associations were also found between gender expression and self‐worth tied to weight (OR 1.16, 95% CI 1.02–1.33), distress with binge eating (OR 1.26, 95% CI 1.04–1.51), and characteristics of binge eating episodes (OR 1.33, 95% CI 1.06–1.66, Table 3).
Significant associations were found between gender non‐contentedness and self‐worth tied to weight (OR 1.38, 95% CI 1.20–1.58) and compensatory behaviors related to weight gain (OR 1.12, 95% CI 1.01–1.26, Table 3).
No significant associations were found between transgender identity and eating disorder variables (Table 3, Table D).
There were no significant interactions (all ps > 0.05) between gender diversity variables and household income or race/ethnicity and eating disorder symptoms.
In this nationwide sample of 10,092 adolescents mostly aged 12–13 years, we found that greater diversity in felt gender, gender non‐contentedness, and gender expression was significantly associated with greater odds of various eating disorder symptoms, including self‐worth tied to weight, binge eating, characteristics of binge eating episodes, distress with binge eating, and compensatory behaviors related to weight gain. Notably, however, there was not a significant association between transgender identity and eating disorder symptoms. This may be due to the categorical nature of the variable, with responses being limited to “Yes,” “No,” and “Maybe.” Some participants may not have understood what “transgender” meant and selected “No” by default. As a result, two types of respondents may not have been captured by this those who are gender diverse but do not identify with the term “transgender” and those who are not yet aware of their gender incongruence.
In contrast, using the 2‐STEP method for felt gender classification, in which participants were divided into four categories of increasing gender divergence, yielded significant results. Interestingly, the most gender diverse individuals (i.e., minority gender) also had significant associations with symptoms but less so than the moderate group. Despite both being categorical variables, the difference in findings between transgender identity and 2‐STEP felt gender suggests that more nuanced gender measures may be more apt in characterizing adolescents' experiences, especially at an age of significant identity formation.
All ordinal and continuous gender variables had significant associations with various eating disorder symptoms, indicating that nuanced assessments of gender diversity may be useful in the early adolescent population. A notable finding was that gender expression and gender non‐contentedness were significantly associated with self‐worth tied to weight. One possible explanation for this association is a social stigma attached to weight among adolescents (Pont et al. 2017). Adolescents may be influenced by various sources, including parental figures and media, tying their weight to feelings of self‐worth (Eisenberg et al. 2015; Puhl and Himmelstein 2018). The stigma attached to certain body weights and sizes may also heighten the gender dysphoria felt by gender diverse adolescents, contributing to eating pathologies (Nagata, Ganson, and Austin 2020). Another interesting finding was that gender non‐contentedness was the only gender variable significantly associated with compensatory behaviors such as laxative use, diet pill use, vomiting, and over‐exercising. The gender contentedness question asks, “How much have you wished to be a girl/boy?” It is possible that individuals could seek to cope with non‐contentedness through compensatory behaviors, which could alleviate their emotional and physical dissatisfaction. In general, the felt gender and gender expression variables were associated with binge‐eating variables. While speculative, it is possible that feelings and expressions about different genders could be manifested through dysregulated eating behaviors. As eating disorders affect both physical and mental health, providing strong family support may reduce the risk of adverse psychosocial and medical outcomes, especially for transgender and gender diverse adolescents (McGregor et al. 2024).
Notable strengths of this study are the large, national, and racially/ethnically diverse sample of early adolescents in the United States and the inclusion of multiple measures of gender diversity, including ordinal and continuous scores for felt gender, gender expression, and gender non‐contentedness. Relying solely on binary or categorical gender identity variables increases the risk of false positive responses from cisgender participants who misread the question or do not understand what is being asked (National Academies of Sciences, Engineering, and Medicine et al. 2022). Using continuous and ordinal variables allows for the operationalization of individual variability in gender identity, making the data more reflective of participants' lived experiences with gender (Miller, Willson, and Ryan 2021). As such, future investigations should continue to include the use of multi‐dimensional and ordinal and continuous measures of gender identity in adolescents.
This study also had several limitations. First, this is a cross‐sectional study using Year 3 ABCD data, and thus, causality cannot be determined. Additionally, given the sensitivity of the subject matter and self‐reported nature of the gender survey, there is a possibility of recall bias where adolescents do not answer questions truthfully, either intentionally or because they do not fully understand the question (i.e., “Are you transgender?”). The lack of significant association between transgender identity and eating disorder symptoms may be partially attributed to this bias. Misreporting of gender has also been observed in other investigations assessing sexual and gender minority identity (Hall, Dawes, and Plocek 2021). Furthermore, in this study, eating disorder symptoms were analyzed categorically as a binary variable rather than on a continuous scale, which may not capture the nuance of disordered eating in early adolescents. Finally, adolescent eating disorder symptom variables were based on parent reports; as such, there is the possibility of misclassification since parent and adolescent reports may differ because parents may not be present for all meals and may be unaware of certain subjective experiences of their child.
The research findings have several potential clinical and public health implications. Given the association between gender diversity and eating disorders, it may be important for clinicians caring for adolescents with eating disorders to create safe and welcoming environments for people of all genders so that gender diverse adolescents are not discouraged from seeking care. Primary care clinicians caring for early adolescents who endorse gender diversity, especially across gender expression, gender non‐contentedness, or felt gender, may consider screening for eating disorder symptoms. Given unique body image considerations for gender diverse adolescents, it is important to ensure that assessment tools and interventions developed for the general population are appropriate for gender diverse adolescents. Recent studies have validated eating disorder assessment tools (Nagata, Compte, et al. 2023a) and supported the efficacy of certain interventions, such as family‐based treatment (Baker et al. 2024), for gender diverse populations.
This study highlights the association between early adolescent gender diversity and eating disorder symptoms, with our findings suggesting that adolescents with greater gender diversity have higher odds of reporting various eating disorder symptoms. Specifically, it demonstrates the importance of using multiple measures of gender diversity to accurately capture the nuances of gender among early adolescents, particularly those with eating disorders. However, further research is needed to explore associations between gender diversity and eating disorders longitudinally across a wider range of ages and developmental stages to inform clinical guidelines and public health policies for gender diverse individuals. Future research could also examine intersectional associations with gender diversity and race/ethnicity or socioeconomic status and eating disorder symptoms.
Jason M. Nagata: conceptualization, formal analysis, supervision, writing – original draft, writing – review and editing. Karen Li: formal analysis, writing – original draft, writing – review and editing. Angela E. Kim: writing – original draft, writing – review and editing. Iris Yuefan Shao: conceptualization, formal analysis, writing – review and editing. Christopher D. Otmar: conceptualization, writing – review and editing. Kyle T. Ganson: writing – review and editing. Orsolya Kiss: conceptualization, writing – review and editing. Alexander Testa: writing – review and editing. Jinbo He: writing – review and editing. Jason M. Lavender: writing – review and editing. Fiona C. Baker: conceptualization, data curation, methodology, writing – review and editing.
The University of California, San Diego (UCSD) provided centralized institutional review board (IRB) approval and each participating site received local IRB approval.
Written informed consent was obtained from the parents/caregivers of adolescents, and written assent was obtained from adolescents.
The authors declare no conflicts of interest.