Eating Disorders & Sports Participation

Abstract:

Eating disorder research is scarce due to the stigma surrounding the condition. Much of the research has focused on the influence of societal norms and social media, while many patients attribute their relationship with physical activity as problematic. This study focused on self-reported symptoms potentially indicative of an eating or body image disorder, even if a diagnosis is not present and how those symptoms differed across athletes and non-athletes. The 11 eating disorder questions, which were coded on a scale from 1 to 5, included how much people are worried about their physical appearance and how much exercise is perceived to be a tool for controlling and altering one’s physique. Differences in self-reported symptoms were also considered between genders. Predicting severity of eating and body image issues from sports participation alone proved to be inconclusive, but a strong association between gender and eating disorders existed. Additionally, predicting eating disorders from sports participation when accounting for gender was more statistically-significant, demonstrating unique psychological pressures on women, whether involved in athletics or not, that predispose them to an eating or body image disorder that are not as apparent in men.

Keywords: eating disorders, body image issues, athletes, sports, gender, anorexia

Introduction

Eating disorders have the highest mortality rate of any mental health disorders (“Eating Disorders and Suicide - Neda”) What is especially problematic is that very little is known about the causes of severe eating and body image issues, and eating disorders can lurk in the shadows due to stigma surrounding eating disorders (Salafia et. al, 2015).

In particular, research has primarily investigated the relationship between media and culture onset of eating and body image issues, while very little research has occurred regarding sports and physical activity and eating disorder onset. According to Salafia et. al, this research discrepancy informs public attitudes, as those without eating disorders were more likely to attribute media and culture as contributing to eating disorders and ignore sports participation at a rate significantly higher than individuals suffering from eating or body image disorders (Salafia et. al, 2015). Past research has quantified eating disorders using self-reported data. However, this approach assumes all participants have access to a mental health professional ooor are in a position to get a diagnosis. Given that the rate at which individuals suffering from eating disorders seek treatment is approximately 10%, it is highly probable participants could be possessing the pathology of an eating disorder without an official diagnosis.

Examining sports participation, psychologists have mostly used self-reported data and evaluated the influence of factors such as sports type, environment and miscellaneous characteristics. For example, Bratland-Sanda et. al. defined sports participation influence as bodily needs of the sport (ie aesthetic, endurance etc.), duration of participation, rules and regulations of the sport (ie weight restrictions in wrestling), and presence of injuries (Bratland-Sanda et. al., 2012). Currie et. al. observed that aesthetic sports and sports that required maintenance of a lean physique or a physique within strict guidelines had the greatest prevalence of eating and body image disorders (Currie et. al., 2019). Additionally, athletes who started their sport at a younger age or incurred an injury were more likely to develop eating or body image disorders, indicating a multifaceted, positive association between sports participation and eating and body image disorders. One current pitfall according to the same study is that many sports programs have frameworks in place for dealing with physical injuries, but none for eating disorders and their related comorbidities (Currie et. al., 2010).

One major critique, however, is that gender could be skewing the result of past research, according to Blair et. al., which suggested that since women are more likely to have eating disorders, female athletes will be documented as having higher rates of eating disorders, leading to the impression that female-dominated sports have a high prevalence of eating disorders due to the nature of the sport, as opposed to being composed of more women than men (Blair et. al., 2017). Gender has been defined using biological sex as opposed to current gender identity, with women being overrepresented: for example, Striegel-Moore et. al in trying to understand gender-based differences in eating disorders among athletes conducted a study with approximately 1,800 men and 3,700 women (Striegel-Moore et. al., nda).

Blair et. al. noted how past literature has concluded that female athletes are at greater risk of eating disorders than male athletes. However, because women are naturally at a greater risk of eating disorders than men, it is unclear whether being an athlete serves as an additional risk factor (Blair et. al., 2017). Another confound of gender is that certain sports have a greater prevalence of women than men (ie aesthetic sports like dance), which could falsely contribute to the perception that sports type predicts onset of an eating disorder. To assess this, participants will be split based on gender and sports participation (4 groups), and the data will be compared based on sports participation without focus on the actual sport, ensuring that the study focus is not diluted. Finally, the measurement of using only male and female excludes a substantial portion of the gender spectrum.

The goal of this study is to provide insight into the role sports participation plays in contributing to an eating disorder (an area of eating disorders research not currently studied extensively), while also determining if eating disorder symptoms manifest at different rates betweensexes (or if outside factors contribute to the rate at which different genders receive a diagnosis). For this study, the hypotheses tested are the following:

Table 1: Hypotheses (null & alternative) for study

Methods

This survey was uploaded to the class Discord as well as club organizations, sports teams, labs, and places of employment. To increase visibility for participants whom I may not know, participants were asked to send the survey to people that they know if they wished. 23 participants successfully completed the survey with no participants excluded. The mean age was 22.09 with a standard deviation of ~5.66 and a range of 18 to 41. The participant pool was majority biologically-female, with 7 biologically-male participants and 16 biologically-female participants. The sports participation was evenly split among participants, with 11 participants stating that they play sports, and 12 participants stating that they do not.

In this study, participants first read a consent form detailing the possible risks of completing the survey, a general background on the research study, and mental health resources that could be utilized if necessary. After agreeing with the terms and conditions of the consent form, participants provided their age, gender identity, and whether or not they participated in a sport or intensive physical activity regularly (defined as > 5 hours per week). Finally, participants completed a modified version of the EAT-26 scale to discern symptoms congruent with eating or body image issues.

Presence of symptoms indicative of an eating or body-image disorder was assessed through an adapted version of the EAT-26 scale containing 11 questions. The scale was converted into a numerical format with 5 items as opposed to 6. The EAT-26 relies on a series of positively-coded items (i.e. “I am terrified about being overweight”, “I avoid eating when I am hungry” from a 1 (indicating never) to 5 (indicating always)) to assess the likelihood of an eating or body image disorder. The average value selected across all questions was calculated for each participant and represented their severity of an eating and body image issue. Due to the stigma surrounding eating disorders, there is potential pressure for participants to minimize their symptoms, which could lead to artificially lower values for this item. However, the movement away from listing the term “eating disorder” survey and rephrasing it as “eating or body image issues” is an effort to mitigate this stigma.

Gender identity was assessed through a multiple choice question, where participants could select “male”, “female”, “nonbinary”, or “other”. No participants selected “nonbinary” or “other”. There is no incentive for this question to lie or for repeated administration of the survey to generate different responses, so the validity and reliability of this measure is high.

Sports participation was assessed through a multiple choice question, where participants were asked if they played a sport or physical activity that required at least 5 hours per week. The participants could select “yes” or “no”. Engagement in physical activity is considered desirable among many individuals, so participants might be tempted to inflate their engagement in physical activity. To attempt to minimize the likelihood of this, “athletes” must engage in physical activity for the key number of 5 hours, providing an objective measure to assess this variable.

Table 2: Descriptive Statistics for Variables in Study

Figure 1: Severity of eating & body image issues among participants Figure 2: Gender composition of participants

Figure 3: Participants’ sports participation Figure 4: Age composition of participants

Results

Hypothesis 1

H0: There is no relationship or a negative relationship between one’s participation in a sport and presence of an eating or body image disorder.

HA: There is a positive relationship between one’s participation in a sport and presence of an eating or body image disorder.

The linear model for Hypothesis 1 drew a relationship between participation in a sport (DV) and onset of an eating or body image disorder (IV). The results were the following: β = -0.1439, 95% CI = (-1.145017, 0.8571383), R^2 = -0.0391, t-value = -0.415, p-value = 0.6824. Contradicting previous research, sports participation actually negatively predicted having an eating or body image disorder, with athletes having a mean severity of eating and body image issues ~0.34 points lower than non-athletes. However, the ability of sport’s participation to predict (either positively or negatively) is rather weak, demonstrated by the R^2 value of -0.04. Additionally, the difference between athletes and non-athletes is statistically insignificant, with a p-value of 0.6824. Further, the difference between the actual slope and the estimated standard error is small, demonstrated by the t-value of 0.415. Hence, while the data did not support the hypothesis, the likelihood that these results were obtained due to chance is to reject the null hypothesis (sports participation does not predict or positively predicts eating disorders).

Figure 5: Linear model between participation in a sport and severity of eating & body image issues

Hypothesis 2

H0: There is no relationship or a negative relationship between one’s gender and presence of an eating or body image disorder.

HA: There is a positive relationship between one’s participation in a sport and presence of an eating or body image disorder.

The linear model for Hypothesis 2 drew a relationship between gender (DV) and onset of an eating or body image disorder (IV). The results were the following: β = 0.7321, 95% CI = (0.1899822, 1.654268), R^2 = 0.1394, t-value = 2.136, p-value = 0.04. This data supported past research, with being female predicting an increase in the severity of eating or body image issues and being male predicting a decrease (women had a mean severity of eating and body image issues ~0.88 points higher than men). Severity of eating and body image issues was positively and more closely correlated with gender, as demonstrated by the R^2 value of 0.1394. Additionally, the difference between women and men is statistically-significant, with a p-value of 0.04. Further, the difference between the actual slope and the estimated standard error is large, demonstrated by the t-value of 2.136. Hence, there is enough evidence to reject the null hypothesis and suggest that gender can effectively predict severity of eating and body image issues.

Figure 6: Linear model between gender and severity of eating & body image issues

Hypothesis 3

H0: There is no relationship or a negative relationship between one’s participation in a sport and presence of an eating or body image disorder after controlling for gender.

HA: There is a positive relationship between one’s participation in a sport and presence of an eating or body image disorder after controlling for gender.

The linear model for Hypothesis 3 drew a relationship between participation in a sport (DV) and onset of an eating or body image disorder (IV), while controlling for the independent variable gender. Gender can influence sports participation: for example, conditioning of women to be timid and submissive frequently leads to women shying away from masculine, aggressive sports or avoiding sports entirely (“Examination…”, 2016). Therefore, certain sports or sports in their entirety might appear to predict symptoms of an eating or body image disorder, when in actuality gender composition is the hidden variable.

The results were the following: β =-0.2306, 95% CI (0.770430, 1.679367), R^2 = 0.119, t-value sports = 0.717, p-value = 0.1.

Beta sports decreased, indicating a suppressor effect on sports. Accounting for gender also increased the R^2 value to 0.119 and decreased the p-value to 0.1, indicating that accounting for gender generates better data. This data still contradicted past research, with sports participation still having a negative association with the onset of an eating or body image disorder, but this time with a stronger link between the independent and dependent variables and more statistically-significant differences between the groups.

Table 3: Inferential Statistics Across Bivariate Models (Models 1 & 2) & Multivariate Models (Model 3)

Discussion

Sports participation without accounting for gender did not prove to be an effective way to predict severity of an eating or body image issue. Not only were differences between non-athletes and athletes insignificant, but the minute differences contradicted data previously generated. In contrast, gender predicted eating or body image issues much more strongly and in line with prior research, and any differences between men and women proved to be significant. However, reexamining eating disorders and sports participation while accounting for gender proved to be more effective than sports alone.

In sports where eating disorders are prevalent, like running and swimming, female athletes suffer at a greater rate from eating disorders than their male compatriots despite partaking in the same sport (Becker, nda), This opens up the possibility for researching how athletic performance expectations within the same sport differ across genders. Although this sample lacked non-cisgender individuals, future studies should investigate the pressure non-cisgender athletes face in comparison to their cisgender colleagues within the same sport. One major limitation was the age distribution of participants, with most participants being between the ages of 19 and 25. Different eating disorders have distinct ages of onset, with a median onset of 18.9 years for anorexia nervosa, 19.7 years for bulimia, and 25.4 years for binge-eating disorder (Rohde et. al, 2017). Additionally, each of these major eating disorders plateaus in prevalence at a different speed and age range: anorexia nervosa, with the earliest onset, also plateaus quickly at 26 years (Rohde et. al, 2017). In contrast, bulimia and binge-eating disorder persist much longer throughout the lifespan, with plateau ages of 47 and 70 respectively (Rohde et. al, 2017). Given the age of participants, it is likely anorexia nervosa would be overrepresented and binge-eating disorder was underrepresented. Future research should repeat this study methodology but with an older population with a median age in the thirties (as opposed to the low twenties).

A particularly interesting direction would be analyzing the relationship between participation in a sport and onset of binge eating disorders in individuals with a median age of 30 in an experimental context by having athletes first fill out an eating disorder likert scale catered to binge eating disorders, stopping or pausing their sport for an extended period of time, and subsequently completing the likert scale again. Past research has focused on the prevalence of eating disorders characterized by partial or full purging (anorexia nervosa and bulimia) in younger athletes (especially highschool and college-aged), so this new direction will expose an untapped area of study. Additionally, previous studies have only used self-reporting data without altering any independent variables directly, so this would provide a new perspective. This study demonstrated that the difference in eating disorder symptoms and body image issues is statistically-significant, demonstrating the need for societal change to improve the mental health of women, but actually showed that people who do not participate in sports have more severe symptoms, calling into question the rationale behind people with eating disorders articulating that sports are a big contributor to their psychopathology.

Literature Cited

Becker, Kendra. “Eating Disorders in Female Athletes.” Mass General Brigham, www.massgeneralbrigham.org/en/about/newsroom/articles/eating-disorders-in-female-ath letes. Accessed 9 Dec. 2024.

Blair, L., Aloia, C. R., Valliant, M. W., Knight, K. B., Garner, J. C., & Nahar, V. K. (2017). Association between athletic participation and the risk of eating disorder and body dissatisfaction in college students. International journal of health sciences, 11(4), 8–12.

Blodgett Salafia, E.H., Jones, M.E., Haugen, E.C. et al. Perceptions of the causes of eating disorders: a comparison of individuals with and without eating disorders. J Eat Disord 3, 32 (2015). https://doi.org/10.1186/s40337-015-0069-8

Bratland-Sanda S, Sundgot-Borgen J. Eating disorders in athletes: overview of prevalence, risk factors and recommendations for prevention and treatment. Eur J Sport Sci. 2013;13(5):499-508. doi: 10.1080/17461391.2012.740504. Epub 2012 Nov 13. PMID: 24050467.

Currie A. (2010). Sport and eating disorders - understanding and managing the risks. Asian journal of sports medicine, 1(2), 63–68. https://doi.org/10.5812/asjsm.34864

“Eating Attitudes Test.” EAT26, 1 Aug. 2022, www.eat-26.com/downloads/.

“Eating Disorders and Suicide - Neda.” National Eating Disorders Association, 22 Mar. 2024, www.nationaleatingdisorders.org/eating-disorders-and-suicide/#:~:text=The%20eating% 20disorder%20anorexia%20nervosa,ranging%20from%203%20to%2029.7%25.

“Examination of Gender Equity and Female Participation in Sport.” The Sport Journal, 29 Feb. 2016, thesportjournal.org/article/examination-of-gender-equity-and-female-participation-in-spo rt.

Rohde P, Stice E, Shaw H, Gau JM, Ohls OC. Age effects in eating disorder baseline risk factors and prevention intervention effects. Int J Eat Disord. 2017 Nov;50(11):1273-1280. doi: 10.1002/eat.22775. Epub 2017 Aug 31. PMID: 28861902; PMCID: PMC5745064.

Striegel-Moore RH, Rosselli F, Perrin N, DeBar L, Wilson GT, May A, Kraemer HC. Gender difference in the prevalence of eating disorder symptoms. Int J Eat Disord. 2009 Jul;42(5):471-4. doi: 10.1002/eat.20625. PMID: 19107833; PMCID: PMC2696560.

Appendix:

EAT-26 Items:

• I am terrified about being overweight.
• I avoid eating when I am hungry.
• I find myself preoccupied with food.
• I have gone on eating binges where I feel that I may not be able to stop.
• I am aware of the calorie content of foods that I eat.
• I particularly avoid food with a high carbohydrate content (i.e. bread, rice, potatoes, etc.)
• I vomit after I have eaten.
• I feel extremely guilty after eating.
• I am preoccupied with a desire to be thinner.
• I think about burning up calories when I exercise.
• I feel that food controls my life.

ED_Sports

ED_df <- data.frame (ED_Sports$ED1,

ED_Sports$ED2,

ED_Sports$ED3,

ED_Sports$ED4,

ED_Sports$ED5,

ED_Sports$ED6,

ED_Sports$ED7,

ED_Sports$ED8,

ED_Sports$ED9,

ED_Sports$ED10,

ED_Sports$ED11)

names(ED_Sports)

EAT <- rowMeans (ED_df)

alpha(ED_df)

mean(EAT)

sd(EAT)

range(EAT)

mean(ED_Sports$Gender)

sd(ED_Sports$Gender)

range(ED_Sports$Gender)

mean(ED_Sports$Sport)

sd(ED_Sports$Sport)

range(ED_Sports$Sport)

mean(ED_Sports$Age)

sd(ED_Sports$Age)

range(ED_Sports$Age)

sportsmod <- lm (EAT ~ Sport, data = ED_Sports)

hist (EAT,

main = "Prevalence of Eating Disorders",

xlab = "Severity of Eating & Body Image Issues",

xlim = c(0,5),

ylab = "Count",

col = "gold")

hist(ED_Sports$Gender,

main = "Gender Distribution of Participants",

xlab = "Gender, 0 = male, 1 = female",

ylab = "Count",

col = "purple")

hist (ED_Sports$Sport,

main = "Sports Involvement of Participants",

xlab = "Sports Involvement, 0 = no, 1 = yes",

ylab = "Count",

col = "red")

hist (ED_Sports$Age,

main = "Age Distribution of Participants",

xlab = "Age",

ylab = "Count",

ylim = c(0,10),

col = "seagreen1")

plot (EAT ~ Sport, data = ED_Sports)

plotmeans (EAT ~ Sport,

data = ED_Sports,

main = "Eating Disorder Prevalence Among Athletes",

xlab = "Sport Participation, 0 = no, 1 = yes",

ylim = c(1,5),

ylab = "Severity of Eating & Body Image Issues")

summary(sportsmod)

coef(sportsmod) + 1.96 * sd(ED_Sports$Sport)

coef(sportsmod) - 1.96 * sd(ED_Sports$Sport)

gendersmod <- lm (EAT ~ Gender, data = ED_Sports)

summary(gendersmod)

coef(gendersmod) + 1.96 * sd(ED_Sports$Gender)

coef(gendersmod) - 1.96 * sd(ED_Sports$Gender)

plotmeans (EAT ~ Gender,

data = ED_Sports,

main = "Eating Disorder Prevalence Based on Gender",

xlab = "Gender, 0 = male, 1 = female",

ylim = c(1,5),

ylab = "Severity of Eating & Body Image Issues")

combmod <- lm (EAT ~ Sport + Gender, data = ED_Sports)

summary(combmod)

coef(combmod) + 1.96 * sd(ED_Sports$Gender)

coef(combmod) - 1.96 * sd(ED_Sports$Gender)

coef(combmod) + 1.96 * sd(ED_Sports$Sport)

coef(combmod) - 1.96 * sd(ED_Sports$Sport)