Introduction

The adolescent period is one of rapid and transformative development. While the majority of adolescents manage the sometimes turbulent transition from childhood to adulthood, there is an emerging concern that far too many adolescents may not achieve their full potential [1]. Adolescence is also a period fraught with many threats to health and well-being, sometimes leading to substantial premature impairment and disability [1]. Significant contributors to adverse health consequences experienced by adolescents are, to a larger extent, the result of health risk behaviors [2, 3].

Of particular concern among adolescents in the general population are health risk behaviors associated with disordered eating and dieting (e.g., excessive calorie restriction and purging, including self-induced vomiting, laxative misuse and excessive exercise to lose weight), which are associated with a wide range of adverse medical and psychological consequences, some potentially life-threatening [4, 5]. Although disordered eating behaviors do not classify as clinical eating disorders, they often develop into eating disorders [6, 7]. Consequently, disordered eating behaviors are often studied as early indicators of eating disorders, such as anorexia nervosa and bulimia nervosa, which are associated with a range of serious comorbidities [4,5,6,7,8,9]. In 2013, 13% of United States high school students reported recent fasting, 5% reported recent use of diet pills or other weight-loss products, and 4.4% reported recent self-induced vomiting or laxative use [8].

Health problems associated with disordered eating affect individuals during critical physiological and psychological developmental periods and disproportionally affect marginalized subgroups of adolescents, in particular sexual minority youth (i.e., lesbian, gay, bisexual adolescents, and youth who self-identify as heterosexual but who have had same-sex sexual contact and are questioning/unsure of their sexual orientation) [10]. For males who self-identify as gay or bisexual, a growing body of research suggests an association with increased susceptibility to eating disorders [11, 12]. Research on sexual minority adolescents [10,11,12,13,14] has suggested a relationship between disordered eating and poor body image. Differences in poor body image among adolescent sexual minorities and heterosexuals may be rooted in different perceptions of ideal physical appearance [13, 14]. Evidence suggests that gay and bisexual males experience greater pressure to emulate popular men in movies and on television than do their heterosexual peers, with particular attention to muscle tone, muscle definition and thinness, whereas lesbian and bisexual females are less likely to experience or internalize these pressures relative to their heterosexual peers [13,14,15,16].

The research literature on sexual minority youth consistently notes two theories that assist in explaining the association between disordered eating and poor body image. The first, Objectification Theory, suggests that women experience greater social pressure to appear thin and attractive, leading to a female gender bias for disordered eating [17,18,19]. In addition, Objectification Theory notes that internalization of unrealistic social standards can result in body shame and consequently, disordered eating. Because men tend to value physical attractiveness more than women, gay and bisexual men may face social and cultural pressures similar to heterosexual women [18,19,20]. Researchers have noted that peer pressure associated with body dissatisfaction and physical appearance is greater for gay men than for heterosexual men [19].

An additional theory that attempts to explain the relationship between sexual identity and disordered eating is Minority Stress Theory, which suggests minority groups such as lesbian, gay, bisexual, and questioning (LGBQ) individuals are at increased risk for mental health challenges resulting from stress caused by social stigma [20]. For LGBQ individuals, this stress manifests as external, objective stressors, such as antigay discrimination; the expectation of prejudice, resulting in vigilant monitoring; and internalization of stigma and prejudice (e.g., internalized homophobia). These factors (not identity itself) account for higher rates of mental and emotional health challenges in LGBQ adolescent populations [20, 21].

From an overall perspective, the body of research on LGBQ women and disordered eating depicts mixed results. A small number of early studies suggest that lesbians experience lower rates of body dissatisfaction and eating disorders compared to heterosexual women [21,22,23], with other studies finding no differences between these two groups of women [11, 18, 24,25,26]. A small number of recent studies suggest that LGBQ females, similar to LGBQ males, may be at an increased risk for eating disorders and disordered eating during adolescent development and emerging adulthood [27,28,29,30].

Researchers have increased their recognition of bisexual persons as a distinct subgroup for investigation, although individuals who identify as “questioning” or “unsure” as to their sexual orientation have been generally ignored. A review of the literature suggests that researchers who do collect data on “questioning” and/or “unsure” adolescent subgroups often choose to exclude them from data analysis [16, 27, 28, 31] even with existing evidence that suggests, similar to their LGB peers, that “unsure” and/or questioning youth are at an increased risk of psychological challenges including suicidality [32, 33]. Owing to the fact that sexual identity development is imperative in the adolescent and emerging adult developmental process, studies with unsure and/or questioning youth subsamples are important for future research.

In turn, the purpose of this study was to investigate the relationship between selected disordered eating behaviors and self-reported sexual minority status (gay/lesbian, bisexual, and unsure) among a representative sample of high-school adolescents. In view of the extant literature, we hypothesized that: (1) sexual minority students would experience increased odds of disordered eating compared to heterosexuals, (2) findings would be stronger for males, and (3) students unsure of their sexuality would not be at increased risk compared to other sexual minorities.

Methods

The dataset from the 2013 Centers for Disease Control and Prevention’s (CDC) Connecticut Youth Risk Behavior Survey (YRBS) was selected because questions querying participants about sexual orientation were available for analysis [34]. Schools (with the exception special education schools) were selected systematically with probability proportional to enrollment in grades 9–12 using random starts to obtain a representative sample of Connecticut public high schools and students in classes. All required subject classes or classes meeting during a particular period of the day, depending on the school, were included in the sampling frame. Survey procedures were designed to protect the privacy of students by allowing for anonymous and voluntary participation. Local parental permission procedures were followed before survey administration [8].

Of the 55 eligible schools, 46 participated (85% school response rate). A total of 2429 (of 3064) students submitted questionnaires (78% student response rate), for an overall response rate of 67% (0.85 × 0.78) [8], meeting CDC’s criterion for weighted data (60%). The YRBS questionnaire has demonstrated adequate test–retest reliability [35].

Instrumentation

The independent variable was formed by combining a question asking participants about their gender (male/female) and a second question asking participants about their sexual orientation (gay or lesbian, bisexual, or not sure). Dependent variables were: “During the past 30 days, did you exercise, eat less food, fewer calories, or foods low in fat to lose weight or to keep from gaining weight?” (responses = yes/no) and “During the past 30 days, did you go without eating for 24 hours or more (also called fasting), vomit, or take laxatives, any diet pills, powders, or liquids (without a doctor’s advice) to lose weight or to keep from gaining weight?” (responses = yes/no).

Adjusted analyses controlled for both demographic and covariates. The demographic covariates of age and race were selected because disordered eating behaviors vary by age and race [9]. The behavioral covariates of cigarette use (past 30 days), binge drinking (5+ drinks at one sitting during the past 30 days), depression (last 12 months), and organized school activity participation (past 7 days) were selected based on previous research [36] and research demonstrating adolescents who engage in disordered eating behaviors are at increased risk for each [37, 38]. The current study also included body mass index (BMI) as an additional covariate. Covariates were entered simultaneously into the adjusted models and not re-coded for analysis. Models were computed separately for males and females.

Statistical analysis

Data were analyzed in SAS version 9.3 using methods accounting for the complex survey design of the CDC YRBS. PROC SURVEYLOGISTIC compared each sexual minority group against the referent heterosexual group separately for males and females.

Results

Sample characteristics are located in Table 1. After removing participants with incomplete data on the study’s independent and dependent variables (n = 187, 7.7%), the final sample was 2242 (49.4% female). Approximately 64% (n = 1441) of the sample self-identified as being non-Hispanic white, 6.8% (n = 152) as non-Hispanic black, 17.4% (n = 391) as Hispanic or Latino, 3.8% (n = 86) as Asian, and the remaining 7.7% (n = 172) were self-identified as multiple ethnicities. Among females, 85.3% of the sample self-identified as heterosexual, while 1.5, 8.8, and 4.4% self-identified as lesbian, bisexual, or not sure of their sexuality, respectively. Among males, 91.5% of the sample self-identified as heterosexual, while 2.7, 2.9, and 2.9% self-identified as gay, bisexual, or not sure of their sexuality, respectively. These estimates are small, but largely consistent with national estimates of adolescents [39], with the exception of the percentage of females self-identifying as bisexual when compared to males. However, studies suggest that females are more likely to self-identify as bisexual than males [40, 41]. Mean BMI for the sample was 22.95 kg/m2 (SD = 4.62).

Table 1 Sample characteristics, frequency and percent by gender
Full size table

No differences were detected between participants not reporting their gender and sexual orientation for exercising or consumed less food to lose weight, χ2 (1, N = 2279) = 1.91, p = 0.17 or fasted, vomited, or took pills to lose weight, χ2 (1, N = 2271) = 0.06, p = 0.81.

Regression analyses

Sexual minority males

Unadjusted and models adjusted for the covariates are located in Table 2, but only significant findings are reported here. Results suggest that gay males were 12.63 times more likely to report fasting, vomiting, or taking pills to lose weight (past 30 days) compared to heterosexual males in the unadjusted models (OR 12.63; CI 4.94–32.27) and 5.67 times more likely in the adjusted models (OR 5.67; CI 1.28–26.12). These associations were statistically significant at the p < 0.01 and p < 0.05 levels respectively.

Gay males were also 2.37 times more likely to report exercising or eating less to lose weight (past 30 days) compared to heterosexual males in the unadjusted models (OR 2.37; CI 1.01–5.69); however, no increased odds were determined in the adjusted models. This association was statistically significant at the p < 0.05 level.

Table 2 Unadjusted and adjusted odds ratios (OR) and confidence intervals (CI) for reporting eating behaviors utilizing heterosexuals as the referent
Full size table

Sexual minority females

Bisexual females were 2.02 times more likely to report fasting, vomiting, or taking pills to lose weight (past 30 days) compared to heterosexual females in the unadjusted models (OR 2.02; CI 1.19–3.45); however, no association was determined in the adjusted models (Table 2). The association in the unadjusted model was statistically significant at the p < 0.05 level.

In addition, bisexual females were significantly less likely to report exercising or eating less to lose weight (past 30 days) compared to heterosexual females in the adjusted models (OR 0.48; CI 0.26–0.90). The association was statistically significant at the p < 0.05 level.

Discussion

Disordered eating in adolescence places young people at high risk for a number of health problems associated with significant premature morbidity, premature mortality, and increased medical costs [4, 42, 43]. Identifying adolescent groups at increased risk is imperative to informing targeted and effective interventions designed to reduce the burden of disordered eating.

In this study with a large, representative sample of high school students from Connecticut we hypothesized that: (1) sexual minority students would experience increased odds of disordered eating compared to heterosexuals, (2) findings would be stronger for males, and (3) students unsure of their sexuality would not be at increased risk compared to other sexual minorities. Results of this study support our hypotheses in that we found gay males at increased risk to engage in disordered eating and sexual minority females not at increased risk in the adjusted models. Strength in separating analyses by gender and sexual minority group is further exemplified when comparing all sexual minorities (gay/lesbian, bisexuals and unsure) against heterosexuals, where bisexual females reported significantly decreased odds of exercising or eating less to lose weight and gay males reported significantly greater odds of fasting, vomiting, or taking diet pills to lose weight in adjusted models.

For the most part, our findings for sexual minority males and females are consistent with previous studies [10, 13, 14, 27, 30]. One exception is the Matthews-Ewald et al. [36] study which includes a lack of significance for bisexual males and finding bisexual females significantly less likely to report exercising or eating less to lose weight. Differences in question wording, smaller sexual minority sample sizes, and no available stress covariate may have contributed to these differences and are study limitations. However, we also cannot discount potential developmental differences between high school and college students as possible explanations, since most disclose their sexual identity in late high school or college [44].

Our findings are also consistent with previous studies that suggest homosexual orientation is related to an increased risk of disordered eating behaviors in males and decreased risk of disordered eating behaviors in females [18,19,20, 22]. Our results are consistent with the hypothesized differential importance placed on physical attractiveness in the gay and lesbian cultures [18,19,20, 22, 45–47]. Rationale for the differential emphasis on physical appearance among the different subcultures is not well understood. One proposed hypothesis suggests that both gay males and heterosexual females desire to make themselves attractive to men [17, 22, 45, 47]. Men tend to evaluate potential romantic partners on the basis of physical appearance to a greater extent than do women [17, 22, 45]. In turn, both gay men and heterosexual women become concerned with their physical appearance to attract male partners [17, 22, 47]. Thus, differences in weight concern, body dissatisfaction, and weight control behaviors can be explained by differences in the degree to which people experience “sexual objectification” by their romantic partners [17, 22, 47]. Objectification Theory [17, 47] deserves further empirical study among adolescents where the proposed mediator (i.e., perceived objectification or perceived instrumental value of having an attractive body shape vis-a-vis attracting a romantic partner) is actually measured [14].

It is uncertain why bisexual females in our study were twice as likely as heterosexual females to have fasted, vomited, used diet pills to lose weight (past 30 days) in the unadjusted models. A somewhat recent study suggests that adolescents with disordered eating experience feelings of shame and disgust regarding puberty and sexuality [48]. In addition, bisexual individuals face unique types of prejudice and stigma not attributed to lesbians or gay persons. Results from a recent study suggest that, in particular, bisexuals experience three types of prejudice: (1) sexual orientation instability (i.e., bisexuality is not a legitimate identity); (2) sexual irresponsibility (i.e., bisexuals are more promiscuous); and (3) general interpersonal hostility (i.e., being disliked as a result of their sexual identity) [49]. In view of this prejudice/stigma, it is possible that bisexual females are learning that their sexuality is “not okay” and use disordered eating as a coping mechanism for negative feelings. Bisexual individuals have reported experiencing prejudice from gay/lesbian communities as well [49]. In addition, bisexual women are more likely to be victims of sexual violence than lesbians or heterosexual women [50] and sexual violence has been associated with disordered eating among adolescent females [51]. The Minority Stress Theory helps explain this prejudice and discrimination against bisexuals even within the gay/lesbian community [20]. In contrast, our results found that bisexual females were significantly less likely to report exercising or eating less to lose weight (past 30 days) compared to heterosexual females, suggesting a protective association. There is research suggesting that gay community involvement plays a protective role against disordered eating in women. For example, Heffernan [24] speculates this protective effect may be related to the self-acceptance individuals feel when surrounded by others similar to oneself. In this research, “coming out” was also associated with positive changes in self and body esteem [24]; however, additional research is necessary to sort these differences.

Disordered eating is posited to occur at higher rates than clinically diagnosed eating disorders [9], and can include behaviors such as laxative use, vomiting, fasting or skipping meals, and diet pill use with the explicit purpose of controlling or losing weight [52, 53]. Among high school adolescents, disordered eating is estimated to occur at rates anywhere from 7.3 to 31% for males and between 15.8 and 57% among females [9, 52] and is thought to be the intermediary step between dieting and the development of clinically diagnosed eating disorders [54]. Whereas disordered eating has been positively associated to feelings of loss of control [55], self-efficacy has been negatively associated with perceived loss of control [56]. Defined as an individual’s perceived confidence in performing a behavior that leads to a desired outcome [57], self-efficacy has been shown to be predictive of the engagement in a variety of positive behaviors [57]. For example, high school students with positive emotional well-being are less likely to engage in disordered eating behaviors [9].

Therefore, results from this study have implications for disordered eating prevention and risk reduction interventions. Regarding interventions, adolescent health experts suggest necessary skills for navigating the challenges of the immediate social environment [58,59,60]. For example, effective learning for improved self-efficacy and emotional well-being in regard to disordered eating risk reduction is skill-development to the degree where an adolescent can begin to trust their ability to reach their goals when faced with emotion-laden interactions with others. In turn, increasing self-efficacy appears to be an important component for interventions designed to prevent adolescent engagement in risky emotion driven behaviors [61,62,63] often associated with the risk factors for adolescent disordered eating [64,65,66].

In addition, results also suggest future disordered eating prevention interventions should be gender, sexual orientation, and behavior specific, as well as model-based, and theory-driven. Programs such as those that include Social and Emotional Learning (SEL) should be considered for adaptation and ultimate use with adolescent sexual minorities experiencing disordered eating. SEL programs are associated with reductions in adolescent health risk behaviors [67], and involve “the systematic development of a core set of social and emotion identification and regulation skills that help children more effectively handle life challenges and thrive in both their learning and their social environments” [68]. The SEL model is based on a number of well-established theories, including theories of emotional intelligence, social and emotional competence promotion, social developmental model, social information processing, and self-management [69]. The SEL model also integrates important aspects of several other behavior change models, including the Health Belief Model, the Theory of Reasoned Action, Problem Behavior Theory, and Social-Cognitive Theory [70, 71]. The SEL model emerged from research on resiliency and teaching social and emotional competencies to children and adolescents [71]. SEL programs use social skills instruction to address behavior, discipline, safety, and academics to help youth become more self-aware, manage their emotions, build social skills, build friendship skills, and decrease their engagement in health risk behaviors [72,73,74,75]. Specific content targets include: anger, empathy, perspective-taking, respect for diversity, attitudes supportive of being assertive, coping, intentions to intervene for others, communication, and problem-solving skills [70, 71].

Study limitations are noted here. First, question wording is a limitation, owing to the collapsing of multiple behaviors into single questions. For example, fasting, diet pill use, and vomiting to lose weight are often separate questions on national surveys (including the national YRBS) which are disordered eating behaviors. Conversely, exercising and eating foods lower in fat may be considered positive weight control behaviors when done in moderation. Hence, the finding that bisexual females are significantly less like to report exercising or eating less to lose weight is difficult to interpret, as the questions do not specifically ask about the degree to which respondents engaged in these behaviors. Second, these data represent public school students from Connecticut and should not necessarily be considered generalizable to other states. Third, study data were cross-sectional, and therefore, preclude conclusions on causality. Fourth, even though estimates of self-identified sexual minorities are consistent with national estimates, the small sample sizes produced some wide confidence intervals; therefore, results should be interpreted cautiously. However, the lack of differences between females and males unsure of the sexuality and bisexuals (males) and lesbians for the disordered eating behaviors question is a significant finding, suggesting it may be appropriate to collapse those unsure of their sexuality with other sexual minorities in disordered eating research.

Conclusion

These findings illuminate the importance of recognizing distinct subgroups of sexual minority youth in research and practice when considering the physical and psychological of disordered eating. Health and other professionals should be cognizant of the dynamic of sexual minority youth having inaccurate perceptions of their exercise and dieting behaviors [76, 77]. In addition, these findings underscore the importance of screening sexually minority adolescents for excessive exercise, extreme dieting, purging, and the use of diet pills and related products. Early detection and referral to treatment for many sexual minority youth may prevent loss of bone density, adverse growth effects, and future reproductive function associated with potential eating disorders [78, 79]. Future school, community, and clinical research needs to carefully examine how primary and secondary prevention strategies are effectively utilized among three distinct sexual minority youth subgroups to improve health and academic outcomes.