Introduction

Disordered eating is defined as “a spectrum between normal eating and an eating disorder and may include symptoms and behaviours of eating disorders, but at a lesser frequency or lower level of severity”1. Disordered eating attitudes and behaviours range between eating (restricting eating, binge eating, purging), exercise (excessive exercise and muscle building), and body concerns and cognitions [body dissatisfaction, cognitive restraint, negative attitudes toward obesity (although “obesity” is a commonly used term in medical fields and is evidenced to predict health-related outcomes, social science research cautions against its use because of stigma and discrimination often associated with it.)]. These components are described by experts in disordered eating research2 in Table 1.

Table 1 Description of disordered eating components according to the EPSI2.
Full size table

In the past, disordered eating has been linked to athleticism or sports affiliation3. These links are especially relevant for high-level weight-sensitive sports such as dance4. Recently, however, research in a general population of young adults indicated that sport participation was not significantly associated with disordered eating, specifically anorexic attitudes and behaviouors5, and may serve as a protective factor in college athletes, compared to non-athletes6, which may be due to their higher levels of self-esteem and body image satisfaction7,8. A meta-analysis examining moderating variables on body image between athletes and non-athletes indicated that college athletes had greater body image satisfaction than club/recreational athletes, but no significant differences in effect sizes between gender (i.e., men and women), sport type (i.e., aesthetic, endurance, ball games), age, and BMI8. Further, some researchers argue that lower levels of physical activity could indirectly lead to the development of detrimental weight regulation behaviours, such as dieting and self-induced vomiting9.

Researchers have attempted to examine how different types of physical activity may influence the development of disordered eating. For example, some researchers have focused on the impact of engaging in “lean sports,” which emphasize body composition, on the development of disordered eating, especially in women, as these activities are often included in patterns of strict dieting and fasting10,11. “Lean-weight sports” include cheer, swimming, volleyball, and cross-country running, whereas “non-lean-weight sports” consist of basketball, softball, soccer, and golf12. These studies serve to inform on preventive interventions in women playing sports12 because female athletes are 14–19% more at risk for engaging in disordered eating than their male counterparts13; however, this physical activity categorization does not encompass muscular-driven disordered eating attitudes and behaviours in men or women.

Although research has indicated that sport participation in men increases the risk of disordered eating behaviours, including binge eating, vomiting, and using laxatives, diuretics, and use of diet pills14, few studies have used male samples to further examine the relationship between involvement in specific physical activity types and weight control behaviours, ranging from exercise to fasting/dieting15. Some of the existing literature examining disordered eating and sports affiliation has several limitations, including methodological issues such as single-source sampling and using questionnaires with no pre-established psychometric properties16. One proposed sport categorization focuses on the comparison between “lean” versus “non-lean sports;” however, this does not adequately represent gender differences as women tend to have a greater drive for thinness, whereas men have a higher drive for muscularity17, which could impact the choice of activities. For example, women inclined towards thinness may favour calorie-burning exercises, while men desiring muscularity may opt for resistance training8. Nevertheless, the drive for muscularity and thinness is not mutually exclusive; studies suggest that men and women typically experience a combination of both desires17.

Petrie and their colleagues proposed a sport categorization that was inclusive to all genders: endurance (cross-country, track, swimming), aesthetic (cheerleading, diving), weight dependent (wrestling), ball game (basketball, baseball, soccer, volleyball, hockey, lacrosse), power (football, downhill skiing), and technical (fencing, golf) sports15. Although this categorization has potential, too few participants were included in the aesthetic and weight-dependent sports categories to make statistically meaningful comparisons; thus, more research is needed to examine physical activity categorization in a sample that includes an adequate number of males and females15.

Due to the inconsistent results regarding the role of physical activity involvement in disordered eating in men and women12, as well as the lack of research comparing physical activity type across specific weight control behaviours15, the main objective of this study was to corroborate existing physical activity categorization (e.g., ball game, aesthetic, endurance, and weight-class) across disordered eating components in a sample of men and women, and (b) further close the gap in the literature by examining disordered eating attitudes and behaviours in men and women separately.

Our first research objective examined the relationships between participation in (yes vs. no) and frequency of physical activity. We hypothesized that active respondents who engaged more frequently in physical activity would have BMIs within the normal weight range and report greater levels of some components of disordered eating. The second research objective explored whether the type of physical activity influenced the prevalence of disordered eating components. The physical activities that participants reported engaging in were categorized using guidelines in the published  literature[e.g.,12,15] and, when reported activities were not included in previous categorizations, through deductive reasoning. Our final physical activity categories included aesthetic, ball games, endurance, and weight-class activities. We hypothesized that weight-sensitive physical activities, such as aesthetic and weight-class physical activities, would be associated with greater disordered eating symptoms. Although we did not have a hypothesis about gender differences, the effect of gender was also included in the analyses of this research objective.

Method

Participants

Participants with more than 30% missing data were removed from further analyses. For each retained participant, the mean and total scores were computed only if at least 80% of the values were available. This approach aimed to preserve the integrity of the data and minimize the impact of missing values on the accuracy of the analysis. The remaining sample consisted of 749 participants (ages 16–62 years), with 539 females (Mage = 27.98, SD = 10.94) and 209 males (Mage = 28.26, SD = 9.36). The final sample consisted of mostly Caucasian (n = 586, 78.2%), heterosexual (n = 605, 80.8%), and post-secondary students (n = 500, 66.8%). Based on a yes/no question, the participants were categorized according to whether they were physically active (n = 537, 72.6%) or inactive (n = 196, 26.5%) to examine differences in BMI and across components of disordered eating. Further, the participants listed the physical activity, including sports, they participated in, which were subsequently categorized into aesthetic, ball games, endurance, and weight-class physical activities12,15.

Measures

Demographic questionnaire

The internally generated demographic questionnaire included questions to collect information about age, gender (“what is your gender?”), weight, and height. Due to a limited number of participants identifying with another gender identity (n = 10), analyses of gender differences were restricted to men and women. Both physical activity and sports were collected to comprehensively assess the spectrum of physical activities individuals engage in, spanning from lower-end competitive levels (e.g., walking) to higher-end involvement (e.g., organized sports): “Do you participate in any sports and/or physical activity?” and “What sports and/or physical activity do you participate in?” We asked participants, on average, how many days a week they participated in physical activity and/or sports as a measure of frequency.

Eating pathology symptoms inventory

The eating pathology symptoms inventory (EPSI)18 measures disordered eating attitudes and behaviours within the past month. The EPSI is divided into eight subscales that demonstrated good internal consistency in the current study: body dissatisfaction (e.g., “I did not like how clothes fit the shape of my body;” α = 0.87), binge eating (e.g., “I ate until I was uncomfortably full;” α = 0.88), Cognitive restraint (e.g., “I counted the calories of foods I ate;” α = 0.72), purging (e.g., “I used diet pills;” α = 0.89), restricting (e.g., “I skipped two meals in a row;” α = 0.86), excessive exercise (e.g., “I exercised to the point of exhaustion;” α = 0.86), negative attitudes toward obesity (e.g., “I felt that overweight people are lazy;” α = 0.88), and muscle building (e.g., “I thought my muscles were too small;” α = 0.79). The 45 items are answered using a five-point Likert scale ranging from 0 (never) to 4 (very often). The EPSI has good psychometric qualities in non-clinical samples of both men and women19. The EPSI was derived from the DSM-5 and has indicated high convergent and divergent validity across gender and weight status18.

Procedure

This study was approved by the University of New Brunswick Research Ethics Board (REB File #056–2020). The procedure follows all guidelines outlined by the Canadian Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans (TCPS 2) and the University of New Brunswick research ethics board. Participants were recruited from social media via Facebook (www.facebook.com) and from SONA, an online tool facilitating students enrolled in psychology courses at the University of New Brunswick to participate in research. Social media participants had the opportunity to enter a draw prize for one of six $25 Amazon gift cards. The University of New Brunswick participants received bonus points applicable to selected psychology courses. Prior to completing the questionnaire package, all participants read a description of the study and provided their informed consent by reading and acknowledging the informed consent form. After providing consent, participants completed the questionnaires through Qualtrics (www.qualtrics.com). In this study, the demographic questions were presented first, followed by the randomized questionnaire package.

Results

Prior to data analyses, the assumptions underlying the statistical analyses were examined. The linearity of the data distribution was confirmed using scatter plots, and the normality of data distribution was evaluated based on skewness and kurtosis, which indicated a normal distribution (skewness <|1.5|; kurtosis <|3.0|; The extreme outliers ( >|3| SD) were winsorized. The assumption of homogeneity of variance was examined using Levene’s homogeneity test for independent samples t-tests. The assumption of multicollinearity was assessed for each linear regression. The levels of the variance inflation factor and the predictors’ tolerance were appropriate, below 10 and above 0.20, respectively20.

The data analytic plan included the calculation of descriptive statistics to provide a breakdown of participant demographics and participation rates in different physical activities. Independent samples t-tests were used to compare BMIs between inactive participants and those participating in aesthetic, ball game, endurance, and weight-class activities. Pearson correlation coefficients assessed the relationship between the frequency of physical activity, BMI, and each disordered eating component. Multiple response crosstabulations were used to analyze the association between participation in multiple categories of physical activities and disordered eating components, with EPSI subscales dichotomized into lower and higher scores based on quartiles. Separate multiple linear regressions were conducted for men and women, predicting each disordered eating component (EPSI subscales) from participation in different types of physical activities. Results were deemed significant at the threshold of p < 0.05.

Physical activity versus inactivity

Compared to inactive participants, individuals who indicated participating in physical activity had lower BMIs and greater levels of cognitive restraint, excessive exercise, and muscle building with small to medium effect sizes based on previously established guidelines21 of small (d = 0.20), medium (d = 0.50), and large (d = 0.80) effect sizes (see Table 2). Greater frequency of physical activity (i.e., more days/week participating in physical activity) was associated with lower levels of body dissatisfaction (r =  − 0.16, p < 0.001), purging (r =  − 0.19, p < 0.001), restricting (r =  − 0.20, p < 0.001), and binge eating (r =  − 0.17, p < 0.001), as well as higher levels of cognitive restraint (r = 0.16, p < 0.001) and excessive exercise (r = 0.38, p < 0.001).

Table 2 Means of active and inactive participants on disordered eating (EPSI) subscale scores.
Full size table

Physical activity type comparisons

Table 3 presents the number of participants who reported engaging in different activities. Participants engaged in a variety of different activities, with most participants reporting activities in multiple categories; the most popular physical activity reported by participants was walking, followed by running, yoga, weightlifting, and cycling. Overall, participation in physical activity in more than one of the categories, regardless of their type, was associated with higher levels of cognitive restraint (r = 0.16, p < 0.001), excessive exercise (r = 0.34, p < 0.001), and muscle building (r = 0.08, p = 0.041). Participation in multiple types of physical activity was not associated with body dissatisfaction, binge eating, purging, restricting, or negative attitudes towards obesity. Independent samples t-tests revealed that inactive participants had a significantly higher BMI (M = 27.11 kg/m2), which fell within the overweight range (BMI = 25–29.9 kg/m2), than respondents who played aesthetic, t(423) = 3.25, p = 0.001, d = 0.31, ball game, t(438) = 3.67, p < 0.001, d = 0.34, and endurance physical activities, t(645) = 2.91, p = 0.004, d = 0.23 (see Table 3).

Table 3 All physical activities reported by participants.
Full size table

Multiple response crosstabulations were calculated because the participants could indicate that they participated in more than one physical activity, resulting in respondents who are included in more than one physical activity category22. For this purpose, the EPSI subscales were dichotomized according to their first and third quartiles, indicating lower and higher subscale scores (see Table 4). Most aesthetic physical activity participants had higher body dissatisfaction scores rather than lower, whereas most ball game, endurance, and weight-class participants had lower body dissatisfaction. In all physical activity categories, the majority of participants had higher cognitive restraint scores than lower ones. Physical activity category membership was not associated with higher levels of purging, as most participants reported lower scores on this subscale. Unexpectedly, the majority of ball game players had higher restricting and negative attitudes toward obesity scores than lower; in contrast, the opposite was true for all other physical activity categories. Engaging in any type of physical activity was related to greater excessive exercise. Lower scores in binge eating were only found in participants engaged in endurance physical activities. Most ball game and weight-class participants reported higher muscle building scores, and most aesthetic and endurance athletes had lower scores.

Table 4 Crosstabulations of dichotomized disordered eating components (quartile cutoff scores) across physical activity categories.
Full size table

Gender differences across physical activity types

A series of linear regressions were conducted to determine if physical activity type membership (dummy coded) was a significant predictor of disordered eating variables above and beyond the variance accounted for by other physical activity type membership and no membership. The series of linear regressions were calculated in the women-only (see Table 5) and men-only samples (see Table 6). Each regression consists of one block with aesthetic, ball game, endurance, and weight-class physical activities as the predictors (0 = does not participate in that physical activity type; 1 = does participate in that physical activity type). The predicted criterion consisted of the EPSI subscales: body dissatisfaction, cognitive restraint, purging, restricting, excessive exercise, negative attitudes toward obesity, binge eating, and muscle building. R2 effect size can be interpreted as very weak (R2 < 0.02), weak (0.02 ≤ R2 < 0.13), moderate (0.13 ≤ R2 < 0.26), and substantial (R2 ≥ 0.26)21.

Table 5 Linear regressions of physical activity type membership on disordered eating in women.
Full size table
Table 6 Linear regressions of physical activity type membership on disordered eating in men.
Full size table

In women, physical activity type significantly accounted for significant variance in the following disordered eating components: cognitive restraint, F(4, 523) = 7.43, p < 0.001, R2 = 0.05; purging, F(4, 523) = 3.97, p = 0.004, R2 = 0.03; restricting, F(4, 523) = 3.58, p = 0.007, R2 = 0.02; excessive exercise, F(4, 521) = 37.01, p < 0.001, R2 = 0.22; and muscle building, F(4, 522) = 13.68, p < 0.001, R2 = 0.10. In contrast, physical activity type did not statistically account for variance in body dissatisfaction, F(4, 521) = 1.37, p = 0.244, R2 = 0.01, negative attitudes toward obesity, F(4, 523) = 0.69, p = 0.595, R2 = 0.00, and binge eating, F(4, 522) = 0.53, p = 0.533, R2 = 0.00. The most at-risk physical activity type was ball game, which emerged as a statistically significant predictor of purging, restricting, excessive exercise, and muscle building. Weight-class women athletes were more likely to engage in cognitive restraint, excessive exercise, and muscle building. Women participating in endurance physical activities tended to engage in cognitive restraint and excessive exercise. Participation in all physical activity categories were significant predictors of excessive exercise. None of the physical activity types were a significant predictor of body dissatisfaction, negative attitudes toward obesity, and binge eating.

In men, physical activity type significantly accounted for variance in the following: body dissatisfaction, F(4, 196) = 1.75, p = 0.141, R2 = 0.04; cognitive restraint, F(4, 197) = 3.00, p = 0.020, R2 = 0.06; purging, F(4, 197) = 7.03, p < 0.001, R2 = 0.13; restricting eating, F(4, 197) = 9.31, p < 0.001, R2 = 0.16; and excessive exercise, F(4, 197) = 3.09, p = 0.017, R2 = 0.06. Conversely, physical activity type did not statistically account for variance in negative attitudes toward obesity, F(4, 196) = 0.57, p = 0.684, R2 = 0.01, binge eating, F(4, 196) = 1.32, p = 0.263, R2 = 0.03, and muscle building, F(4, 197) = 1.98, p = 0.099, R2 = 0.04. Men who engaged in aesthetic physical activities were at a higher risk of engaging in cognitive restraint, purging, and muscle building. Men who engaged in weight-class physical activities tended to have higher excessive exercise but lower body dissatisfaction and purging. Men ball players tended to have greater levels of restricting. Endurance physical activities were not a significant predictor of any disordered eating component. None of the physical activity types were significant predictors of negative attitudes toward obesity and binge eating.

Discussion

The scarce literature that includes contradictory findings demonstrates the importance of further understanding the role of physical activity involvement in disordered eating in adult men and women, comparing physical activity type across specific disordered eating behaviours and attitudes12,15. Overall, in this study, physically active participants had higher levels of cognitive restraint, excessive exercise, and muscle building. Further, our results indicated that increased exercise frequency is associated with some, but not all, components of disordered eating; increased exercise frequency was associated with higher cognitive restraint and excessive exercise but lower body dissatisfaction, purging, restricting, and binge eating. Thus, our results partially confirm research suggesting that disordered eating attitudes and behaviours are not associated with5.

In the current study, individuals who engaged in sports and/or physical activity had an average BMI of 24.96 kg/m2, within the healthy weight range. In contrast, participants who reported not engaging in sports and/or physical activity had a significantly greater average BMI of 27.11 kg/m2, which is considered overweight. In sum, the present findings suggest that being active in physical activity is generally associated with higher levels of excessive exercise, which was captured in this study by items pertaining to engaging in strenuous exercise nearly every day and to the point of exhaustion; however, it was also related to a BMI in the normal range. Maintaining a positive balance of physical activity can pose challenges for some people, as it may inadvertently result in adverse outcomes, including disordered eating, particularly with excessive exercise. Following recommended guidelines, such as 150 min of moderate-intensity aerobic activity and two days of muscle-strengthening activity per week 23, can improve physical and psychological well-being while preventing excessive exercise; however, this may not be feasible for athletes engaging in competitive sports who have strict training routines. Based on the current findings, we suggest the following additional recommendations: (a) developing and implementing individualized training programs that balance performance goals with the prevention of disordered eating behaviours; (b) regularly monitoring athletes for signs of excessive exercise, including routine assessments by coaches, trainers, and healthcare professionals; and (c) providing education for athletes and coaches about the risks of excessive exercise and the importance of healthy eating and exercising behaviours.

Due to the multiple burdens associated with all eating pathologies, including medical complications, stigma and discrimination, and nutritional deficiencies, this study aimed to corroborate existing sports categorization in a sample of men and women across disordered eating components and further mend the gap in the literature on disordered eating attitudes and behaviours in men. Participation in various types of physical activity significantly predicted disordered eating attitudes, behaviours, and cognitions differently across genders (see Table 7). For women, purging and restricting eating were more strongly associated with being involved with a ball game physical activity compared with participation in other physical activities. Further, for women, muscle building was more associated with ball games and weight-class physical activities. Each type of physical activity significantly contributed to the variance in excessive exercise; thus, all types of physical activity are relevant factors in predicting excessive exercise behaviour. Although physical activity type, as a whole, was a significant predictor of cognitive restraint, individual physical activity types did not uniquely contribute to the variance in this variable. For men, body dissatisfaction and excessive exercise were associated with participating in a weight-class physical activity. Cognitive restraint and muscle building were more related to aesthetic physical activities than others. Purging was associated with aesthetic and weight-class physical activities, whereas restricting eating was related to the latter and ball game physical activities.

Table 7 Summary of significant predictors of disordered eating components across genders.
Full size table

Previous literature supports that women who play ball games and men who engage in weight-class physical activities are more likely to engage in disordered eating attitudes, behaviours, and cognitions24. Although participating in any physical activity was consistently associated with most components of disordered eating according to the overall regression models, the present findings suggest that, when comparing the relative contribution of each type of physical activity, differences in physical activity type were not a reliable predictor across genders. Clinicians should develop and implement targeted interventions for individuals who participate in various types of physical activity, focusing on the unique pressures and challenges they face that may contribute to disordered eating.

With the possible exception of men who engage in aesthetic physical activities, women involved in physical activities were more at risk for developing disordered eating attitudes and behaviours compared to men involved in activities. This finding is consistent with well-documented prevalence rates of disordered eating, which are higher in women compared to men25. Given that women are generally more at risk for disordered eating, there should be a greater emphasis on screening and preventive measures for female athletes across all physical activities. Special attention should also be given to men in aesthetic physical activities, as they may also be at a higher risk for disordered eating.

When comparing each physical activity category without accounting for gender, higher body dissatisfaction was found in those who played aesthetic physical activities, whereas higher body dissatisfaction and muscle building were related to endurance physical activities. Restricting and negative attitudes towards obesity were higher among ball game players compared to other physical activities. Cognitive restraint and excessive exercise were related to participation in all physical activity categories; binge eating was associated with all physical activities except endurance. Altogether, ball game physical activities had prevalence proportions in favour of disordered eating more frequently than other physical activity categories. Aesthetic and weight-class physical activities were associated with greater disordered eating on half of the subscales, whereas endurance physical activities were only once related to a higher disordered eating score.

These findings are only partially consistent with the two studies used to model our sports categorization to compare disordered eating rates across sport types. In the first study12, compared to “non-lean sports,” such as ball games, “lean sports,” including aesthetic and endurance sports, were associated with greater levels of disordered eating. In the second study15, sport type (i.e., endurance, ball game, and power) was not associated with disordered eating. The finding that physically active people tended to have greater disordered eating tendencies compared to inactive individuals is consistent with previous literature26; however, even in a sample including males and females, sports emphasizing “leanness” and those that were weight-dependent (e.g., wrestling) had a higher risk of disordered eating than ball game sports.

In contrast, Rosendahl and colleagues27 found that non-athletes are at greater risk of disordered eating than athletes and that “leanness sports” (endurance, aesthetic, weight-class, and antigravitation sports) were associated with greater disordered eating compared to “non-leanness sports” (power, technical, ball game sports) in boys but not girls; however, they were using an adolescent sample. Some researchers concluded that men in a weight-sensitive sport are especially susceptible to disordered eating compared to ball games relative to women27,28,29. The discrepancies among the findings in the existing literature could be the result of differences in samples and/or sport categorization (e.g. some studies classified endurance as weight-sensitive sports [e.g.,27], whereas others did not [e.g.,29]). The inconsistent results across studies imply the presence of unaccounted variables. For example, although some studies link ball games with increased disordered eating, including this one, others suggest a decrease. Therefore, additional factors, such as the level of competitiveness, warrant further investigation.

A few limitations to the present study were noted. First, the data was collected at the beginning of the COVID-19 pandemic (January–April 2020), significantly impacting the general population’s eating habits and physical activity participation30,31. Second, the EPSI was used as a comprehensive measure of various disordered eating attitudes, thoughts, and behaviours, including excessive exercise and muscle-building scales, which are often omitted from other popular scales (e.g., eating disorder examination questionnaire [EDE-Q] 32; and eating attitudes test [EAT-26] 33;). Despite its use with athletic samples in previous research34, some researchers remain concerned that the EPSI may inflate scores in athletic populations due to the intensive exercise programs required for high-level competition. Future research could address these concerns by evaluating the influence of competition level on EPSI scores, thereby providing deeper insights into its validity and applicability within athletic settings. Third, the non-probability, cross-sectional sampling does not allow inferences about causality. Fourth, since participants could choose multiple responses, finding an appropriate inferential analysis to examine differences among physical activity types was challenging. A series of linear regressions were conducted to determine if physical activity type membership (dummy coded) was a significant predictor of disordered eating variables above and beyond the variance accounted for by other physical activity type membership and no membership.

Due to only partially consistent findings, future research is needed to corroborate further existing sports categorization in a sample of men and women across disordered eating components. Examining the motivation to engage in disordered eating could elucidate whether athletes engage in such attitudes and behaviours due to body appearance or functionality, as seen in a study with an adolescent sample35.