Psychological risk indicators of disordered eating in athletes.

OBJECTIVES: This project examined risk factors of disordered eating in athletes by adapting and applying a theoretical model. It tested a previously proposed theoretical model and explored the utility of a newly formed model within an athletic population across gender, age, and sport type to explain disordered eating.
DESIGN: The design was cross-sectional and the first phase in a series of longitudinal studies.
METHODS: 1,017 athletes completed online questionnaires related to social pressures, internalisation, body dissatisfaction, negative affect, restriction, and bulimia. Structural equation modelling was employed to analyse the fit of the measurement and structural models and to do invariance testing.
RESULTS: The original theoretical model failed to achieve acceptable goodness of fit (χ2 [70, 1017] = 1043.07; p < .0001. CFI = .55; GFI = .88; NFI = .53; RMSEA = .12 [90% CI = .111-.123]). Removal of non-significant pathways and addition of social media resulted in the model achieving a parsimonious goodness of fit (χ2 [19, 1017] = 77.58; p < .0001. CFI = .96; GFI = .98; NFI = .95; RMSEA = .055 [90% CI = .043-.068]). Invariance tests revealed that the newly revised model differed across gender, age, level, competition status, and length of sport participation.
CONCLUSION: This study showed that the formation of disordered eating symptomology might not be associated with sport pressures experienced by athletes. It revealed that disordered eating development varies across gender, competition level, sport type, and age, which must be considered to prevent and treat disordered eating in athletes.

Introduction

Participation in competitive sports has the potential to increase the risk of eating disorders and disordered eating in athletes [1, 2]. Disordered eating and eating disorders affect the psychological and physical health of millions worldwide [3]. Understanding, this particular pathology in athletes is especially intriguing as regimented diet and intense exercise is often part of competitive sport, but can also be symptoms or maintenance factors of disordered eating and eating disorders [4]. In 2007 (and later re-released in a second edition in 2012), Petrie and Greenleaf published a theoretical etiological model that outlined potential risk factors for the development of disordered eating in athletes as a series of mediators and moderators (see Fig 1). This model was chosen for additional study in this research due to its detailed nature and because it is the only disordered eating model that has been created specifically for athletes rather than a clinical model that has been adapted for use with athletes [5]. Furthermore, as dictated by the original authors of the theoretical etiological model, more testing in athlete samples is needed. This theoretical etiological model is based on the dual-pathway model that posits that social pressure and the internalisation to be thin lead to body dissatisfaction, which in turn leads to negative affect and dietary restraint resulting in eating pathology [6, 7]. The theoretical etiological model also uses evidence from research that explored psychological and social risk factors for disordered eating in athlete and non-athlete samples that consisted primarily of female participants [7-10]. The theoretical etiological model outlines eight factors/mediators that are deemed to be risk factors or causal risk factors based on previous experimental or longitudinal research in the topic area [6]. The factors that are included in the model are (1) sport pressures, (2) societal pressures, (3) internalisation, (4) body dissatisfaction, (5) negative affect, (6) restrained eating, (7) modelled behaviours by peers and family, and (8) binge eating and bulimia. The model also includes five groupings of moderators that affect the intensity and directionality between factors, however this research will focus on the eight risk factors outlined in detail below.

Fig 1

Original theoretical etiological model from petrie and greenleaf, 2007, 2012.

Sport pressures

In the original chapter that accompanies the theoretical etiological model, the higher prevalence of disordered eating in athletes was determined to result from three elements: being an athlete versus a non-athlete, competition level, and type of sport (lean/non-lean)[11]. These three elements that influence prevalence are used in the current study to operationalise sport pressure alongside three additional relevant elements: the number of years of sport participation, training hours per week, and whether an athlete is currently competing or in off-season. These are applied to help to determine whether participating for longer in sport and actively competing creates additional pressure that may relate to the development of disordered eating behaviour.

Previous research findings have been largely consistent in indicating that for lean sports, in which performance or success is influenced by a lean body shape, such as gymnastics or diving, disordered eating has a higher prevalence rate for both men and women (e.g., [12-18]). The level at which the athlete competes is also a relevant sport pressure. However, while some research has posited that elite athletes have higher levels of disordered eating than recreational ones, other research has found the opposite [2, 12, 19].

The current study also includes the number of hours spent training per week as a sport pressure, which has rarely been included in previous research, as much of the previous work has been conducted using National Collegiate Athletic Association athletes in the U.S. (whose hours of training are capped at 20 per week) [20]. This study is relatively unique in its investigation of the amount training hours per week as a sport pressure and will build on previous work which claims that level and training regime should be investigated [21]. The original work by Petrie and Greenleaf stated that increased exercise and training is inversely connected with body dissatisfaction and is thus a relevant factor to be included [22]. Therefore, increased hours could be assumed to mean an increased amount of pressure on the athlete with the caveat that not all sports require more than 20 hours of training per week. Furthermore, there is a positive correlation between level and hours per week, with more hours generally required for high-level sports; hence, these two variables tend to be related. The present study also determines whether an athlete’s status as currently competing or currently in ‘off-season’ plays a role as a sport pressure. Each sport has a time of year or season designated for competition. For example, rugby players compete in the winter, while major cycling events are held during the summer. Although athletes continue to train during the off-season, it has been suggested that measuring disordered eating without accounting for whether those athletes are currently feeling the pressures of competition may have resulted in equivocal findings [23, 24].

Societal pressures

Societal pressures affect athletes, and much like the rest of the population in the Western world, female athletes feel pressure to fit a thin ideal, and men feel pressure to fit a muscular one [25]. However, for athletes, these ideals can compound the sport pressures already exacted upon them, or an opposing tension can arise [26-28]. For example, for a female powerlifter, gaining mass may be advantageous for her sport but not for meeting societal ideals. Societal pressures can stem from family, teammates, the media, and more recently, from social media and the internet [29, 30, 21, 31]. Discerning the degree to which these pressures are internalised—the extent to which the pressures are incorporated into one’s values and beliefs—is key to understanding the impact of sport and societal pressures on athletes. It is often only when these pressures are internalised that athletes’ body satisfaction is damaged and disordered eating behaviours become likely to develop [32, 33].

The current study updates the model to fit the demands of the 21st century by including social media pressure as one of the societal pressures faced by athletes. The majority of popular social media involves photographs, and research has shown that viewing photographs decreases self-evaluations in women and is linked to the increased prevalence of eating disorder symptoms, including excessive exercise [33-37]. Social media creates a community in which harmful behaviours associated with disordered eating and eating disorders can be reinforced or seen as commonplace [35]. In fact, those people considered to be part of the online health community—a source of ‘fitspiration’—scored higher on disordered eating questionnaires and presented more symptoms of compulsive exercise than those whose online presence was focused on travel [38]. Research indicates that repeated exposure to photographs of women who possess the athletic-ideal body type increases body dissatisfaction, and as athletes are likely to follow other athletes on social media, the implications of social media usage warrant attention [39].

Internalisation and body dissatisfaction

Internalisation, in this context, is the incorporation of an external, often unattainable body shape ideal into how one measures one’s own self-worth [11]. When sport and societal pressures give rise to an ideal body standard that is then internalised, if an athlete perceives his or her own body to not match to this ideal, body dissatisfaction and disordered eating behaviours can occur [40-42]. This body dissatisfaction has been found to be directly related to the formation and maintenance of eating disorders and disordered eating [8]. Evidence specifically suggests that body dissatisfaction in athletes is strongly related to the development of disordered eating [14, 43]. However, discrepancies regarding which type(s) of athletes experience the most body dissatisfaction and whether athletes experience more dissatisfaction than the general population remains the subject of debate [17, 44–46]. When comparing levels of body dissatisfaction in athletes to nonathlete controls and lean sport athletes to nonlean sport athletes only equivocal conclusions can be drawn; however, despite the debate, there is consensus that body dissatisfaction contributes to disordered eating [14, 17].

Negative affect

While negative affect is presented in the original model as a link between body dissatisfaction and bulimic behaviours based on research in the general population, few studies have examined negative affect in athletes in relation to disordered eating [11, 47]. Research has found that negative affect in athletes, especially constructs such as fear and guilt, influenced bulimic behaviour when tested in conjunction with increased body dissatisfaction and dietary intent [48-50]. Negative affect has been linked with disordered eating in athletes via increased levels of compulsive exercise [51, 52]. As athletes have a readily available outlet of sport and exercise, over-exercise may be a tempting avenue through which athletes alleviate negative affect [4].

Modelled behaviours

Modelled behaviours are those adopted behaviours that have been reinforced as the norm in a group and those modelled behaviours pertaining to eating psychopathology have been shown to increase disordered eating symptomology [53, 54]. The mediator that is termed modelled behaviour by peers and family is one that much of the literature has overlooked, and research has yet to operationalise or validate a measurement tool for this potential risk factor. Therefore, the present research begins to close this gap in the literature by including questions relating to modelled behaviours of disordered eating among athletes.

Restrained eating, and binge eating and bulimia

Restrained eating refers to the attempted and successful behaviour of limiting food intake in terms of quantity and type [55]. Binge eating and bulimia refer to the symptoms of overeating and compensatory behaviours [56]. These two mediators are conceptually considered part of disordered eating but are examined separately as dictated by the theoretical model. In the model, restrained eating is thought to mediate the relationship between negative affect and binge eating and bulimia [11, 47]. Restrained eating, as well as bulimia and bingeing, have been linked to negative affect resulting from stressful situations [57].

Previous structural testing of the model

Previous studies have tested several elements of the model [58]. However, only a few have used structural equation modelling that allows for the simultaneous analysis of relationships, which is required to move the field forward. First, Anderson, Petrie, and Neumann (2011) tested the eight factors depicted in the model in female collegiate gymnasts, swimmers, and divers. They found that modelled behaviours did not fit the model and that several pathways required adjustment in that sport pressure directly impacted body dissatisfaction and dietary restraint rather than being mediated by internalisation. Several years later, De Sousa Fortes and colleagues (2015) considered the model in a multi-sport sample of male adolescent Brazilian athletes. The researchers originally predicted that the sport pressures of training regimes at the competitive level, body fat percentage, and sociocultural pressures would lead to body dissatisfaction, which in turn may promote disordered eating in male athletes. However, they found that only the sociocultural factors and body dissatisfaction predicted disordered eating. These two studies, the prevalence factors taken from the original 2007 chapter, and the findings of a literature review were all used to guide the operationalisation in the present study to create measurement consistency throughout the research area [2, 59].

Demographic correlates of disordered eating

The current study also aims to test this model between gender and age. Previous studies have tested the elements of the model in single-gender samples [21, 48, 60, 61]. Therefore, in an effort to advance the field this study has included both genders for analysis. The literature on disordered eating has shown higher prevalence rates of disordered eating and eating disorders in female athletes, but a co-ed sample is needed to re-test this element [2]. Due to convenience sampling, the ages of the athletes in previously conducted studies have often been of traditional US university age (c.f. [28, 62]). This study instead considers a broader range of ages to ascertain how age may influence disordered eating development. With regard to this demographic information, elements that are intrinsic to each individual will help to enhance how this research can be utilised in applied care.

Aims

The existing literature on athletes and disordered eating has presented significant inconsistencies with respect to the utilised tools, methodologies, and general findings. There is also a lack of longitudinal studies in the literature. As a result, it is difficult to identify the underlying causal factors for disordered eating in this population [63, 59]. The original theoretical etiological model is comprehensive in its inclusion of factors and its sport-specific focus, thus providing an ideal starting point. In summary, this study has three aims. The first is to use structural equation modelling (SEM) to determine the utility of the original theoretical etiological model by testing all eight mediators in a diverse sample that includes both men and women and a wide range of sports and levels of participation. The second aim is to develop and test a revised model that includes social media. The third and final aim is to test the newly revised model and determine whether it is equivalent across groups such as gender, sport type, competition level, if currently competing, years of participation, and age.

Methods

Procedure

This study received ethical approval from the Clinical Educational and Health Psychology Department at University College London, Reference for this approval: CEHP/2018/573. Participants were recruited using online social media campaigns as well as by word-of-mouth. Those who identified themselves as athletes and performed at least ten hours of training per week were invited to participate. The questionnaire consisted of 241 questions and took just over 30 minutes to complete. It was administered using Opinio [64] and following informed consent, participants could start the questionnaire and return to it at a later point within a seven-day window. The questionnaire was open from January 27, 2019 to February 24, 2019. Participants who completed the questionnaire were given a £5 Amazon voucher as compensation for their time.

Participants

The inclusion criteria that were applied for this study were that participants had to be over 18; there was no upper cap on the age of potential participants. Participants also had to consider themselves to be an athlete, and a minimum of ten hours a week participating in their sport as well as have been actively competing, thus meaning that competitive sport—rather than zealous exercise—was a significant part of their daily experience. After providing their informed consent, 1,208 participants started the online questionnaire, of whom 1,084 completed it. Only completed questionnaires were kept for analysis. Seventeen participants were excluded due to not fitting the inclusion criteria and another three were deleted as they were based outside of the UK, as this study did not have ethical approval to recruit internationally. Thirty-five more with matching email addresses and exact matching answers were removed due to suspicion from the researcher that the answers were duplicates completed to obtain the £5 gift voucher. A final two were removed as when asked, they did not identify as an athlete. Therefore, the final sample consisted of N = 1017 athletes, of whom 56% were male and 44% were female. A wide range of sports was represented in the 1,017 participants, with swimming (18%), tennis (8%), football (soccer) (11%), basketball (18%), volleyball (10%), dancing (7%), and various athletics (track and field) events (19%) making up the majority of the sports represented. Other less-represented sports included golf, boxing, equestrian, cycling, ping-pong, orienteering, rugby, rowing, and race walking. Slightly over half (58.3%) of the participants were classified as competing in lean sports, with 41.7% classified as non-lean sport participants. Several other demographic variables as well as clinical scores were captured and are presented in Table 1.

Table 1

Additional demographic information.

		Percent of Total
Age	18–26	85.7%
	27+	14.3%
Individual or Team	Individual	52.4%
	Team	47.6%
When in Season	Currently Competing	75.2%
	Not Currently Competing	24.8%
Years participated in their sport	1–3 Years	6.4%
	4–8 Years	52.7%
	9–15 Years	40.1%
	16+ Years	0.8%
Hours/Week	10–15 hours/week	17.9%
	16–25 hours/week	60.4%
	26–40 hours/week	20.9%
	40+ hours/week	0.8%
Level	NonElite	82.5%
	Elite	17.5%
BMI (Range)	All Participants	14.1–32.4
BMI (Mean; SD)	All Participants	21.2; 2.5
EDE-Q (Range)	All Participants	0.39–5.49
EDE-Q (Mean; SD)	All participants	2.6; 0.8

Measures

The materials used to operationalise the relevant variables were a combination of existing validated questionnaires and other new measures created for the purpose of this study. It is important to note that the reliability and validity of these measures were largely established in non-athlete samples.

Demographics

Participants’ age, gender, sport type information, years spent competing, the teams for which they had competed, and self-reported height and weight were collected.

Sport pressure

Sport pressure was operationalised as lean (where weight and shape are integral to performance success) or non-lean sport participation, the level at which the participants competed, the hours per week that they spent training, their years of participation in that sport, and their current status as competing or in off-season. The categorisation of lean and non-lean sports was decided by the first author, based on previous classifications of sports in the existing literature or based on the first author’s applied experience. The hours trained per week were categorised into 10–15, 16–25, 26–40, and 40+ hours. The levels of sport participation were grouped into ‘non-elite’, which was comprised athletes competing at a county or regional level, and ‘elite’, which consisted of those participating at a national or international standard. Athletes determined whether they were part of an individual or team sport, and in the instances where the athletes chose ‘both’, the authors used their discretion to categorise that response appropriately. Athletes could choose from the following categories of the years for which they had taken part in their sports: 1–3 years, 4–8 years, 9–15 years, and 16+ years. Finally, athletes could indicate whether they were currently competing or in their off-season. In the case that N/A was chosen for the answer, it was assumed that those athletes were consistently competing throughout the year and thus were categorised with those who indicated that they were currently competing.

Societal pressure

Societal pressure was measured using the nine-item Information and seven-item Pressures subscales from the Sociocultural Attitudes Towards Appearance Questionnaire-3 (SATAQ-3) [65]. The subscales were summed for a total societal pressures score. An example of a statement from the pressure subscale is “I’ve felt pressure from TV or magazines to lose weight”, and from the information subscale, “Pictures in magazines are an important source of information about fashion and ‘being attractive.’” Cronbach alphas were high for these subscales as well as for the entire SATAQ scale: Information (α = .94), Pressures (α = .94), and global score (α = .94) [66]. Statements related to social media usage were also utilised with similar questions to those that were asked in the SATAQ about TV or magazines, simply by replacing these words with “social media”. For example, “I’ve felt pressure from social media to be thin.” This brief part of the questionnaire consisted of five items, was created for this study, and has not been formally validated; however, reliability calculations for the current study showed an alpha of .75.

Internalisation

Internalisation was also measured by using subscales of the SATAQ-3, specifically the five-item Internalisation-Athlete (α = .89) and the nine-item Internalisation-General (α = .92) [66]. As above, these two subscales were summed for ease of analysis. An example item from the Internalisation-General is “I compare my body to the bodies of people who are on TV,” and from the athlete scale, “I wish I looked as athletic as sports stars.”

Body dissatisfaction

The nine-item body dissatisfaction subscale (EDI-BD) of the Eating Disorder Inventory (EDI-2) was utilised to measure body dissatisfaction (α = .88) [67, 68]. An example of a statement from this subscale is “I think that my stomach is too big.” Notably, this scale shows lower alpha coefficients for men compared to women [69].

Negative affect was measured using the 10-item subscale (α = .85; 95%, CI = .84–.87) from the Positive and Negative Affect Schedule (PANAS), in which participants had to rate the extent to which they felt various emotions, such as “guilty”, “scared”, and “nervous: from “Very slightly or not at all” (1) to “Extremely” (5). Alpha reliability for this scale was .87.

Restrained eating

Restrained eating was operationalised by using the Restraint Subscale of the Eating Disorder Examination Questionnaire (EDE-Q-R) [70, 71]. This subscale had high internal consistency in this sample (α = .78) and a high test-rest reliability in other mixed-gender samples (r = .81) [72]. An example question from this subscale is “Have you tried to exclude from your diet any foods that you like in order to influence your shape or weight (whether or not you have succeeded)?.” It is important to note that one question from this subscale was omitted due to human error, hence, as recommended by the authors of the scale, mean calculations were used with items included. The item missing was “Have you wanted your stomach to be empty?”. It is also key to note that controversy over the content validity of this scale exists, with research indicating that it more accurately measures restriction rather than restraint [55, 73]. The global EDE-Q score was also calculated for each participant to better understand prevalence rates and create clinical relevance (α = .91).

No previous research has included a validated measure for modelled behaviours of peers and family, so this current study used seven questions to attempt to capture these sentiments. Example statements included “My friends diet or use weight control behaviours” and “My teammates diet or use weight control behaviours.” Reliability analysis for the current study found the alpha to be .61. In the current study, questions were designed based on the qualitative results that had previously discussed the harmful role that modelled behaviours of teammates can have in the development of eating disorders [54].

Binge eating and bulimia

Binge eating and bulimia were measured by using the bulimia subscale of the 64-item EDI-2 (EDI-B) [67]. This subscale is scored from “Never” to “Always” on a six-item Likert scale. An example statement is “I have gone on eating binges where I have felt that I could not stop.” This subscale also has a high internal consistency (α = .76).

Data analysis

SEM was conducted using AMOS. Several indicators of model fit were utilised during the analysis: χ2 significance, the Comparative Fit Index (CFI), Goodness of Fit (GFI), Normed Fit Index (NFI), and Root Mean Square Error Approximation (RMSEA). In general, a nonsignificant χ2 value shows that the data fits the model well, however, with large samples it is unlikely to obtain a p value < .05 [74]. For CFI, GFI, and NFI a value >.90 is considered to indicate acceptable fit and for RMSEA a value < .08 is needed for acceptable fit [75].

Results

Analysis was conducted using SPSS version 25 and the SPSS add-on AMOS. Excel Macros was also utilised. When analysing the data for normality, it was determined that all variables had distributions within the acceptable range for skewness and kurtosis [75].

Descriptive statistics

Table 2 outlines the descriptive statistics and intercorrelations for all the variables involved in the model. While some of the correlations did show significant relationships between variables the correlations were all weak, suggesting that variables were measuring relatively independent underlying constructs.

Table 2

Table of correlations and means and standard deviations.

	Lean/NonLean	Hours/Week	Elite/Nonelite	Competing/Offseason	Years done sport	SATAQ-Pressures	SATAQ-Information	Social Media Pressures	Internalisation-General	Internalisation-Athlete	EDI-BD	Modelled Behaviour	EDE-Q Restraint Sore	EDI-B
Lean/NonLean	1	.183**	0.051	.069*	.096**	−0.024	0.061	0.001	0.028	0.033	−0.013	.073*	0.048	0.014
Hours/Week		1	.256**	−.321**	.296**	−0.003	−0.002	0.014	0.052	−.072*	0.015	−.154**	−.209**	.107**
Elite/Nonelite			1	−0.013	.163**	0.023	.084**	−0.032	.090**	0.027	−0.049	−0.005	−.087**	.097**
Competing/Offseason				1	−.191**	−0.016	0.038	−0.050	−0.051	−0.005	−0.031	.275**	.282**	−0.025
Years done sport					1	−.113**	−0.049	−0.057	0.027	−.119**	.069*	−.077*	−0.061	0.013
SATAQ-Pressures						1	.372**	.530**	.479**	.454**	−0.014	−0.016	−0.020	.395**
SATAQ-Information							1	.310**	.337**	.193**	−.069*	0.008	0.009	.207**
Social Media Pressures								1	.416**	.423**	−0.005	−0.008	0.005	.276**
Internalisation-General									1	.312**	−0.001	−0.013	−0.028	.230**
Internalisation-Athlete										1	−0.016	0.001	−0.030	.303**
EDI-BD											1	−0.028	0.017	0.051
Modelled Behaviour												1	.564**	−0.022
EDE-Q Restraint													1	−0.015
EDI-B														1
M						22.33	28.04	17.67	28.18	16.15	6.26	15.03	3.03	4.33
SD						3.33	3.07	3.19	3.14	2.35	2.72	3.27	.94	3.57

* p < .05;

**p < .001

Clinical information

The gold standard for clinically diagnosing an eating disorder is by using the EDE-Q global along with the accompanying Eating Disorder Examination Interview [76]. Therefore, this study cannot clinically diagnose as it only reports the EDE-Q global, however it can give clues to prevalence rates among athletes with 5.7% of the current sample scoring above the suggested clinical cut off of 4.0 [77, 78]. Additional information about the EDE-Q global scores can be found in Table 1.

Testing the theoretical etiological model

Testing the theoretical etiological model involved creating a hybrid model, a cross between a path model and a measurement model in AMOS (Fig 2). Results revealed that this model did not fit the data well. χ2 (70, 1017) = 1043.07; p < .0001. CFI = .55; GFI = .88; NFI = .53; RMSEA = .12 (90% CI = .111–-.123). Regression weights are indicated with Beta’s to show the strength of the relationship between factors.

Fig 2

Original theoretical etiological model in athletes.

To enhance the model, social media was added as an observed variable as part of social pressures. This did improve the fit of the model but not to an acceptable standard. χ2 (70, 1017) = 646.09; p < .0001. CFI = .74; GFI = .92; NFI = .72; RMSEA = .09 (90% CI = .084–-.096). Thus the original etiological model with social media was rerun followed by the deletion of all nonsignificant pathways, followed by other pathway alternations as indicated by the modification indices. This resulted in the following revised model with excellent parsimonious goodness of fit (Fig 3). This was the model used for invariance testing. χ2 (19, 1017) = 77.58; p < .0001. CFI = .96; GFI = .98; NFI = .95; RMSEA = .055 (90% CI = .043–-.068).

Fig 3

Revised model with best fit for athletes, used for all invariance testing.

Standardised regression weights, or direct effects are shown for this newly revised model in the figure (Fig 3). Indirect effect analysis along with bootstrapping 200 samples at a 90% confidence interval revealed that Societal Pressures (β = .45, p = .002) and Body Dissatisfaction (β = .08, p = .01) both had significant indirect effects on Bulimia.

Testing for invariance across groups

The revised model created (Fig 3) had the best fit of all the models tested. Therefore, tests of invariance across several groups including gender, age, competition level, if currently competing, and years done sport, using this revised model are outlined below. As sport pressure had been removed from the model due to a lack of fit, the model could be tested for differences across these pressures as well. When testing for invariance, two methods were employed. First, AMOS multigroup analysis was utilised to examine the Comparative Fit Index to see if the constrained models differed from the unconstrained model by an amount equal or larger than .01, which is then taken as indication that the model differs across groups [74, 79]. Secondly, a Chi Square difference test using AMOS analysis and Excel Macros was also used to give additional insight. While still showing good fit for the vast majority of groups, the model showed that variance was present depending on gender, age, level of athlete, whether athlete was currently competing, and how long that athlete had participated in sport. The model was invariant for lean and nonlean sport types.

Multigroup invariance testing revealed differences between genders (see Table 3). The model fit better for men χ2 (19,567) = 54.46; p < .0001; CFI = .96; GFI = .98; NFI = .95; RMSEA = .06 (90% CI = .040–-.076) than for women χ2 (19, 450) = 46.70, p< = .0001; CFI = .95; GFI = .98; NFI = .93; RMSEA = .06 (90% CI = .037–-.078). Excel macros used in conjunction with AMOS revealed that gender differences were explained by differences in the pathway between body dissatisfaction and negative affect at a 99% confidence interval (CI).

Table 3

Gender model fit indices across various model constraints.

Model	χ2	DF	CFI	GFI	NFI	RMSEA	CI for RMSEA
Unconstrained	101.170	38	.960	.977	.938	.040	.031	.050
Measurement weights	107.601	42	.959	.975	.935	.039	.030	.048
Structural weights	130.870	45	.946	.969	.920	.043	.035	.052
Structural covariances	131.040	46	.946	.969	.920	.043	.034	.051
Structural residuals	132.603	49	.947	.969	.919	.041	.033	.049
Measurement residuals	149.089	55	.941	.965	.909	.041	.033	.049

Invariant testing between those athletes under age 26 (young) and those above age 26 (mature) also showed a significantly different fit (see Table 4). The model fit better for younger athletes χ2(19, 872) = 79.81, p < .0001; CFI = .96, GFI = .98, NFI = .94, RMSEA = .061 (90% CI = .047–-.075) than mature ones χ2(19, 145) = 36.98, p = 0.008; CFI = .93, GFI = .94, NFI = .87, RMSEA = .081 (90% CI = .041–-.120). Further testing using excel macros revealed that age specifically moderated the pathways between internalisation and bulimia and negative affect and bulimia at a 99% CI thus explaining the different fit for the two age groups.

Table 4

Age model fit indices across various model constraints.

Model	χ2	DF	CFI	GFI	NFI	RMSEA	CI for RMSEA
Unconstrained	116.904	38	.951	.973	.930	.045	.036	.055
Measurement weights	138.229	42	.940	.968	.917	.048	.039	.056
Structural weights	160.925	45	.928	.963	.904	.050	.042	.059
Structural covariances	163.515	46	.927	.964	.902	.050	.042	.059
Structural residuals	167.979	49	.926	.962	.899	.049	.041	.057
Measurement residuals	178.351	55	.924	.961	.893	.047	.039	.055

The model better explained disordered eating for nonelite athletes χ2(19, 839) = 78.56, p < .0001; CFI = .95, GFI = .98, NFI = .94, RMSEA = .061. (90% CI = .047–-.076) than elite athletes χ2(19, 178) = 57.68, p < .0001; CFI = .88, GFI = .93, NFI = .84, RMSEA = .107 (90% CI = .076–-.139) with elite athletes being the only group in which the CFI dipped below the level required for adequate fit. See Table 5 for invariant test results. Excel macros analysis showed that for these groups, the difference could be explained by the path from societal pressures to internalisation at a 90% CI and from internalisation to bulimia at a 95% CI.

Table 5

Elite vs Nonelite model fit indices across various model constraints.

Model	χ2	DF	CFI	GFI	NFI	RMSEA	CI for RMSEA
Unconstrained	136.387	38	.939	.969	.919	.051	.042	.060
Measurement weights	155.302	42	.930	.964	.908	.052	.043	.060
Structural weights	156.395	45	.931	.963	.907	.049	.041	.058
Structural covariances	160.276	46	.930	.963	.905	.049	.041	.058
Structural residuals	164.555	49	.929	.963	.902	.048	.040	.056
Measurement residuals	208.583	55	.906	.958	.876	.052	.045	.060

Those currently competing and those in off-season were variant in a multi-group analysis test (see Table 6) but only showed a very slight difference in fit for the GFI, but no difference in fit for the CFI with those currently competing showing a fit of χ2 (19, 252) = 41.40, p = .002; CFI = .95, GFI = .96, NFI = .91, RMSEA = .069 (90% CI = .040–-.097) and those in offseason with a fit of χ2 (19, 765) = 73.02, p < .0001; CFI = .95, GFI = .98, NFI = .94, RMSEA = .061 (90% CI = .047–-.076). Excel macros showed that the difference in fit can be explained by currently competing or being in offseason moderating the pathways from internalisation to bulimia and negative affect to bulimia at a 95% CI and body dissatisfaction to negative affect at a 99% CI.

Table 6

Currently competing vs out of season model fit indices across model constraints.

	χ2	DF	CFI	GFI	NFI	RMSEA	CI for RMSEA
Unconstrained	114.454	38	.952	.974	.931	.035	.036	.054
Measurement weights	121.960	42	.950	.972	.926	.034	.035	.052
Structural weights	139.598	45	.941	.969	.916	.037	.038	.054
Structural covariances	141.597	46	.940	.968	.915	.037	.037	.054
Structural residuals	145.959	49	.939	.967	.912	.036	.036	.053
Measurement residuals	164.633	55	.932	.964	.901	.037	.037	.052

Finally, the model fit increasingly well the longer an athlete had participated in his/her sport (see Table 7). Those doing sport for one to three years showed a fit of χ2 (19, 65) = 33.47, p = .02; CFI = .84, GFI = .89, NFI = .72, RMSEA = .109. (90% CI = .042–-.169), those doing it for four to eight years showed a fit of χ2 (19, 536) = 106.22, p < .0001; CFI = .91, GFI = .96, NFI = .89, RMSEA = .093 (90% CI = .076-.110), and those who had done sport for nine or more years χ2 (19, 416) = 76.75, p < .0001; CFI = .92, GFI = .96, NFI = .90, RMSEA = .086 (90% CI = .066–-.106). (An insufficient number of athletes had done their sport for over 16 years and so the 16+ category was combined with the 9–15 years category for analysis). Excel macros exposed that years participated in sport influenced the pathways from internalisation to bulimia at a 99% CI and negative affect to bulimia at a 95% CI and body dissatisfaction to negative affect at a 90% CI.

Table 7

‘Years spent participating in sport’ model fit indices across model constraints.

Model	χ2	DF	CFI	NFI	GFI	RMSEA	CI for RMSEA
Unconstrained	216.700	57	.910	.883	.952	.053	.045	.060
Measurement weights	357.694	65	.835	.808	.922	.067	.060	.073
Structural weights	403.791	71	.813	.783	.909	.068	.062	.075
Structural covariances	410.205	73	.810	.779	.910	.067	.061	.074
Structural residuals	461.764	79	.784	.752	.903	.069	.063	.075
Measurement residuals	520.471	91	.758	.720	.891	.068	.063	.074

Discussion

The purpose of this research was to assess the applicability of the theoretical etiological model proposed by Petrie and Greenleaf in 2007 and 2012 in a large, multi-sport, mixed-gender sample. This study specifically considered relationships between the eight mediators described in the theoretical model: sport pressure, societal pressure, internalisation, body dissatisfaction, negative affect, modelled behaviours, restrained eating, and binge eating and bulimia, as well as the addition of social media as a societal pressure. This study aimed to expand upon previous research that tested this theoretical model by adding social media in order to create a newly revised parsimonious model with high goodness of fit. It was hoped that by using a larger sample of mixed gender athletes that was not constrained by the convenience sampling of US universities would mean a revised, better fitting model would be revealed. The revised model’s utility was also assessed across groups to test for invariance.

The analysis showed poor fit for the original theoretical etiological model, although this fit improved slightly with the inclusion of social media. The relevance of social media as a source of societal pressure has been shown in a range of recent research that has indicated how intimately sociocultural experiences are tied to the online world [80]. The next step was to remove non-significant pathways, which resulted in the creation of a revised model that had good fit on the CFI, GFI, and other relevant indices.

This newly revised model showed that societal pressures (including social media), mediated by internalisation, are associated with binge eating and bulimia, while body dissatisfaction leads to bulimia and binge eating, mediated by negative affect. It is important to note that restrained eating was eliminated from the model and, therefore, this model predicts binge eating and bulimia as only one facet of disordered eating in athletes. Predicting only bulimia symptomology matches with the findings of the dual-pathway model that was the original clinical basis for all modelling by Petrie and Greenleaf [57].

As this study is cross-sectional, causal claims cannot be made. However, these results demonstrate that the formation of binge eating and bulimia is not associated with the sport pressures experienced by athletes, as theoretically suggested. These findings add to the growing body of research that indicates that sport pressures may not directly influence the development of disordered eating symptoms of binge eating and bulimia [5, 21]. The applied implication of this specific finding may be that prevention techniques and interventions utilised in non-athlete samples may be applicable. However, it is still important to note that several aspects of sport pressure created variance in the model, demonstrating that while sport pressure does not fit within the model, various sport pressures may still be tangentially relevant to the development of disordered eating symptoms of bulimia and binge eating in athletes.

Invariance testing across several groups showed that the newly created model differed across several groups, and the results showed in which pathways these differences arose. This variance is something that must be taken into account when considering the prevention and treatment of disordered eating in athletes. With sport pressures removed from the model, invariance testing was also conducted across the factors that were originally described as measuring sport pressure.

While the newly revised model showed very good fit for both males and females, tests of gender differences revealed that the model fit better for male than for female athletes, which is notable, as the original model was designed based on literature that included mainly females. However, research has found increasing evidence that significant numbers of male athletes suffer from disordered eating [1, 81]. The model also fit better for young athletes, specifically those younger than age 27, which is also noteworthy, as the vast majority of research on athletes originates from convenience sampling in US universities, where the ages are often limited to 18–22 years. The model also fit better for those athletes of a non-elite level (i.e., those at a regional or county level), which indicates that these athletes, who are serious yet aspiring, may not have adequate support in coping with the pressure that they experience, whereas elite athletes have access to such support. The model fit better for athletes who completed the questionnaire while currently in their off-season, something that is rarely taken into account when surveying athletes. This finding may indicate that when training intensifies around competition time, it provides an outlet for athletes that allows them to fully fuel their bodies without internalising societal pressures or experiencing negative feelings around food. This finding is in line with previous work that has found athletes score higher on DE measures in the pre-competition time of the season than when currently competing [82]. Finally, invariance testing showed that the longer an athlete had participated in a sport, the better the model fit for that athlete, indicating an increasing pressure on athletes who had participated in a sport for over nine years. Notably, the model was invariant across lean and non-lean sport types. This final lack of variance may be due to societal pressures playing a larger role in the creation of disordered eating symptomology rather than the sport classification as lean or nonlean. In other words, perhaps it is society’s image of an athlete that creates pressure for weight and shape to be managed in a way that defies the physiological demands of the sport (Stoyel, Shanmuganathan-Felton, Stoyel, & Serpell, Under Review).

Invariance testing gives indications of how lived experiences in sport and therefore disordered eating symptoms can fluctuate sport to sport, level to level, and season to season. Noticing the variance between demographics of athletes acknowledges that formation, prevention, and treatment of disordered and eating disorders in this population is not one size fits all, and gives early clues as to how to best tailor care.

Limitations

This study is limited by its cross-sectional nature; however, future longitudinal phases of this project on currently underway. The measurement scales used in the current study were the ones most commonly found in eating disorder and disordered eating literature; however, none were specifically designed for athletes. Perhaps this use of general scales can explain why restrained eating also no longer fit. Additionally, in the absence of other acceptable measures, the scales for social media usage and modelled behaviours were purpose-built for this study and thus were not formally validated in previous research and so had lower Cronbach alpha coefficients than would normally be acceptable. However, the sample, the range of different sports, and the fact that this study did not rely on a convenience sample of US university athletes from a similar age-range make it a unique study in this area of research. Finally, while clinical information was collected and superficially analysed, this research is limited in its ability to make sound assumptions on clinical implications.

Future directions

Future research should use a longitudinal design to determine whether the relevant factors are able to predict future disordered eating in an athlete sample, something that this research group aims to do over the course of the next year. Repeating this research with athletes younger than 18 years old would also provide additional insight as it is a high-risk population for the development of eating disorders. A qualitative investigation to further understand the differences between the original etiological model and the revised model will also be undertaken.

Conclusion

In conclusion, this study provides a large amount of empirical data to further test the original theoretical model with a more diverse sample than previously used. In doing so, a newly revised model was created that can begin to explain the presentation of disordered symptomology specifically related to binge eating and bulimia in a wide range of male and female athletes. Future research should study samples in repeated measures designs so that it is possible to begin to predict the development of disordered eating symptoms in athletes.

(XLSX)

Click here for additional data file.

27 Feb 2020

PONE-D-19-34507

Psychological Risk Indicators of Disordered Eating in Athletes

PLOS ONE

Dear Hannah,

Thank you for submitting your manuscript to PLOS ONE. The manuscript was revised by two reviewers. They both recommended minor revisions and therefore I invite you to re-submit a revised version of this manuscript.

We would appreciate receiving your revised manuscript by 26/03/2020. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Valentina

Valentina Cardi

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For more information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially sensitive information, data are owned by a third-party organization, etc.) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for inviting me to review this manuscript, examining psychological risk factors of disordered eating in athletes. The project tested a theoretical model proposed by Petrie and Greenleaf in a large, multi-sport and mixed gender sample using structural equation modelling. It also tested the utility of a newly formed model including the addition of social media as a risk factor. The newly revised model was then tested to see if it is equivalent across groups including gender, age, sport type and competition level.

The study found a poor fit for the original theoretical model, which improved slightly with the addition of social media. The newly revised model had a better fit for males than females, young athletes and those at a non-elite level.

Strengths of this manuscript include its large sample size, mixed gender sample, the clarity in writing and the importance of the research topic. Limitations include the use of research scales which have not been validated by past research and its cross-sectional design meaning that causation cannot be inferred (as noted by the authors).

My recommendation is to accept this manuscript for publication following minor amendments, as follows:

1) Can the authors provide more details on the studies that previously undertook structural testing of the model? For example, I think it would be helpful for the reader to know which pathways required adjustment in the study by Anderson et al. Further, in De Sousa Fortes and colleagues study, what sports did the male Brazilian athletes compete in? I think this information would help the reader to further understand the background to the present study, to then aid their interpretation of this study’s findings.

2) What led the authors to suspect that there were duplicates in the completion of the survey? Was this based on email addresses or IP addresses?

3) Further information on the demographic details would be useful. For the age range, 27+, what was the maximum age recruited in the study? What was the mean BMI for the study? I also think it is important to include the total N for each of the different sports that the athletes took part in so that future research could replicate this study.

4) In the limitations it should also be mentioned that due to human error one of the questions from the restrained eating measure was omitted. Which item was omitted from the scale? Please include this information in the manuscript.

5) There are a few minor typos in the article that require editing, otherwise the article is very well written. Please amend: line 57: ‘disordered eating in’; line 235 there is a % sign missing; line 226, please remove either 1,017 or N=1017 as repetition is not needed; line 357 ‘due to a lack of fit’; line 186 a full stop is missing; line 92 please spell out the acronym NCAA in full.

6) The discussion section of the article would benefit from elaboration. For instance, drawing on psychological theory why did the model fit better for certain groups and how could the findings be taken into account when considering the prevention and treatment of disordered eating in athletes?

Reviewer #2: This study entitled “Psychological Risk Indicators of Disordered Eating in Athletes” aimed at assessing the applicability of the theoretical etiological model proposed by Petrie and Greenleaf (2007; 2012) in a large sample of athletes. Overall, the study addresses a gap in the current literature and I believe that it offers an interesting contribution, but would be strengthened by addressing the following concerns.

Introduction:

General feedback: The authors need to spend more time/space explaining why is important to investigate disordered eating and eating disorders in athletes. Although I am familiar with the literature, the authors assume a lot of knowledge about this topic from the reader. I suggest explaining why this topic is important before describe the Petrie and Greenleaf (2007; 2012) model.

1) Page 3, line 57: The sentence starting with "Participation in competitive sports has the potential to increase the risk of eating disorders and disordered eating in" do not make sense. Perhaps the authors mean: in athletes?

2) Page 3, lines 63-64: Please explain briefly the dual-pathway model.

3) Page 3, line 73: Authors should clearly state that they are going to explain all the eight factors/mediators included in the model.

4) Pages 3 and 4, lines 78-83: Why the number of years of sport participation and whether an athlete is currently competing or in his/her off-season are elements that may relate to the development of disordered eating? Other variables might have been chosen. Indeed, later (lines 90-101) authors introduced the number of hours spent training per week as a relevant factor of spot pressure. Please clarify because it is misleading.

5) Page 4, lines 84-85: Please provide examples of lean sports. Again, authors assume a lot of knowledge about this topic from the reader.

6) Page 4, line 86: Please include the bracket after the number 15.

7) Page 4, line 92: Do not use the acronym (NCAA) the first time you mentioned a word.

8) Page 5, line 113: I think authors should include also the coach as a source of societal pressure (along with teammates).

9) Page 6, lines 140-142: Authors should explain deeply the debate pertaining to which type(s) of athletes experience the most body dissatisfaction and whether they experience more dissatisfaction than the general population.

10) Page 6, lines 146-148: The sentence "Research has found that negative affect in athletes, especially constructs such as fear and guilt, influenced bulimic behaviour when tested in conjunction with increased body dissatisfaction, dietary intent, and dietary intent and that certain elements of negative affect " do not make sense. Please clarify.

11) Page 8, line 186: Please include a full stop after the word analysis.

Methods & Results:

General feedback: The authors should have assessed for the presence of psychological disorders and, specifically, for EDs (and eventually removed participants with EDs) or, at least, include the lack of assessment for psychological disorder within the Limitation section.

Indeed, the range of BMI (Table 1) was comprised between 14.1 and 32.4; both a BMI of 14.1 and 32.4 may be indicative of an ED/disordered eating. Therefore, the etiological model was tested in a sample of athletes where the prevalence of EDs and disordered eating was uncertain. Authors should include this information in the limitation section.

Furthermore, authors should include internal consistency reliabilities for the current study instead of from others' investigations (see Measures section).

1) Page 9, lines 217-219: Why 10 hours of training should be an index of significant involvement in sport activity (instead of a lower/higher number of hours)?

2) Page 9, lines 227-229: It could be useful for readers the inclusion of % for each sport included in the study.

3) Page 10, lines 253-245: It is not clear to me why authors operationalized sport pressure with participation in an individual/team sport. They did not include any reference pertaining to this topic in the introduction.

4) Page 12, line 299: Include a full stop after the question mark.

Discussion:

General feedback: The discussion largely repeats the findings of the analyses, which is helpful to the reader. However, the authors do not give enough attention to theoretical explanations of the findings (i.e. page 21, lines 466-467: please provide explanations for the invariance of the model across lean and non-lean sports). In general, the authors need to spend less time re-stating their results and more time discussing them. Why are the important? What is the next step?

Furthermore, authors should discuss the clinical implications of the current study. Why your results pertaining to the Petrie and Greenleaf (2007; 2012) model are important in terms of clinical implications?

1) Page 20, line 440: Please include a full stop.

2) Page 20, line 444: Authors should rephrase the discussion in accordance with the results they found. They should refer only to binge eating and bulimia symptoms instead of talking about disordered eating in general. See also page 21, line 488.

3) Page 21: Authors should include in the future directions section the inclusion of athletes younger than 18 yeas old, given that this population was overlook by authors but represents an high-risk population for the development of EDs.

Figures:

Please include the number and the legend for each figure.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: Review.docx

Click here for additional data file.

22 Mar 2020

The authors thank the reviewers for their careful consideration of this manuscript. All of the suggested changes have been incorporated into the manuscripts. Responses can be found on this document as well as in the track changes and comments into the original manuscript.

Reviewer #1: Thank you for inviting me to review this manuscript, examining psychological risk factors of disordered eating in athletes. The project tested a theoretical model proposed by Petrie and Greenleaf in a large, multi-sport and mixed gender sample using structural equation modelling. It also tested the utility of a newly formed model including the addition of social media as a risk factor. The newly revised model was then tested to see if it is equivalent across groups including gender, age, sport type and competition level.

The study found a poor fit for the original theoretical model, which improved slightly with the addition of social media. The newly revised model had a better fit for males than females, young athletes and those at a non-elite level.

Strengths of this manuscript include its large sample size, mixed gender sample, the clarity in writing and the importance of the research topic. Limitations include the use of research scales which have not been validated by past research and its cross-sectional design meaning that causation cannot be inferred (as noted by the authors).

My recommendation is to accept this manuscript for publication following minor amendments, as follows:

1) Can the authors provide more details on the studies that previously undertook structural testing of the model? For example, I think it would be helpful for the reader to know which pathways required adjustment in the study by Anderson et al. Further, in De Sousa Fortes and colleagues study, what sports did the male Brazilian athletes compete in? I think this information would help the reader to further understand the background to the present study, to then aid their interpretation of this study’s findings.

The reviewer is thanked for their insight, additional information has been added to this section of the manuscript.

2) What led the authors to suspect that there were duplicates in the completion of the survey? Was this based on email addresses or IP addresses?

They were removed based on email addresses and exact matching answers, this information has been added to the manuscript. The authors thank the reviewer for their attention to detail.

3) Further information on the demographic details would be useful. For the age range, 27+, what was the maximum age recruited in the study? What was the mean BMI for the study? I also think it is important to include the total N for each of the different sports that the athletes took part in so that future research could replicate this study.

The author thanks the reviewer for these suggestions, they have now been incorporated into the manuscript.

4) In the limitations it should also be mentioned that due to human error one of the questions from the restrained eating measure was omitted. Which item was omitted from the scale? Please include this information in the manuscript.

The reviewer is thanked for this idea and the information has been added to the manuscript.

5) There are a few minor typos in the article that require editing, otherwise the article is very well written. Please amend: line 57: ‘disordered eating in’; line 235 there is a % sign missing; line 226, please remove either 1,017 or N=1017 as repetition is not needed; line 357 ‘due to a lack of fit’; line 186 a full stop is missing; line 92 please spell out the acronym NCAA in full.

The reviewer is thanked for their spectacular attention to detail. All of these changes have been made to the manuscript.

6) The discussion section of the article would benefit from elaboration. For instance, drawing on psychological theory why did the model fit better for certain groups and how could the findings be taken into account when considering the prevention and treatment of disordered eating in athletes?

The reviewer is thanked for this contribution. Additional information has now been added to manuscript to address this highlighted issue. The additions can be found throughout the discussion section.

This study entitled “Psychological Risk Indicators of Disordered Eating in Athletes” aimed at assessing the applicability of the theoretical etiological model proposed by Petrie and Greenleaf (2007; 2012) in a large sample of athletes. Overall, the study addresses a gap in the current literature and I believe that it offers an interesting contribution, but would be strengthened by addressing the following concerns.

Introduction:

Reviewer #2 General feedback: The authors need to spend more time/space explaining why is important to investigate disordered eating and eating disorders in athletes. Although I am familiar with the literature, the authors assume a lot of knowledge about this topic from the reader. I suggest explaining why this topic is important before describe the Petrie and Greenleaf (2007; 2012) model.

The reviewer is thanked for this feedback, additional background information has now been included in the early stages of the manuscript.

1) Page 3, line 57: The sentence starting with "Participation in competitive sports has the potential to increase the risk of eating disorders and disordered eating in" do not make sense. Perhaps the authors mean: in athletes?

The reviewer is thanked for their attention to detail, this mistake has been rectified.

2) Page 3, lines 63-64: Please explain briefly the dual-pathway model.

The reviewer is thanked for this idea, more explanation has been added to the manuscript.

3) Page 3, line 73: Authors should clearly state that they are going to explain all the eight factors/mediators included in the model.

The reviewer is thanked for the insight. This has change is now included in the manuscript.

4) Pages 3 and 4, lines 78-83: Why the number of years of sport participation and whether an athlete is currently competing or in his/her off-season are elements that may relate to the development of disordered eating? Other variables might have been chosen. Indeed, later (lines 90-101) authors introduced the number of hours spent training per week as a relevant factor of spot pressure. Please clarify because it is misleading.

The reviewer is thanked for highlighting this issue, additional information and therefore clarification has been added throughout this section.

5) Page 4, lines 84-85: Please provide examples of lean sports. Again, authors assume a lot of knowledge about this topic from the reader.

The reviewer is thanked for their insight, relevant examples have been added.

6) Page 4, line 86: Please include the bracket after the number 15.

Thank you for this mention, a bracket has been added.

7) Page 4, line 92: Do not use the acronym (NCAA) the first time you mentioned a word.

This acronym has now been spelled out, many thanks for the identification of the error.

8) Page 5, line 113: I think authors should include also the coach as a source of societal pressure (along with teammates).

The reviewer is thanked for this insight. While the research does not have the scope to touch on the role of the coach, it is an insightful idea for future work.

9) Page 6, lines 140-142: Authors should explain deeply the debate pertaining to which type(s) of athletes experience the most body dissatisfaction and whether they experience more dissatisfaction than the general population.

The authors thank the reviewer for their insight. Additional information has been added to the manuscript to satisfy this comment.

10) Page 6, lines 146-148: The sentence "Research has found that negative affect in athletes, especially constructs such as fear and guilt, influenced bulimic behaviour when tested in conjunction with increased body dissatisfaction, dietary intent, and dietary intent and that certain elements of negative affect " do not make sense. Please clarify.

The reviewer is thanked for this remark, the sentence has been clarified.

11) Page 8, line 186: Please include a full stop after the word analysis.

Thank you for your attention to detail, a full stop has been added.

Methods & Results:

General feedback: The authors should have assessed for the presence of psychological disorders and, specifically, for EDs (and eventually removed participants with EDs) or, at least, include the lack of assessment for psychological disorder within the Limitation section.

Indeed, the range of BMI (Table 1) was comprised between 14.1 and 32.4; both a BMI of 14.1 and 32.4 may be indicative of an ED/disordered eating. Therefore, the etiological model was tested in a sample of athletes where the prevalence of EDs and disordered eating was uncertain. Authors should include this information in the limitation section. Furthermore, authors should include internal consistency reliabilities for the current study instead of from others' investigations (see Measures section).

The reviewer is thanked for the detailed and helpful insight. More information on the EDE-Q global scores have been added to both the results section and limitations. Additionally, reliability calculations have been added.

1) Page 9, lines 217-219: Why 10 hours of training should be an index of significant involvement in sport activity (instead of a lower/higher number of hours)?

The reviewer is thanked for their question, additional information has been provided in the manuscript. See page 10 for the additions.

2) Page 9, lines 227-229: It could be useful for readers the inclusion of % for each sport included in the study.

The reviewer is thanked for this suggestion. Additional information on the percentages of each sport included in the study have been added to the manuscript. Please see page 10 for additions.

3) Page 10, lines 253-245: It is not clear to me why authors operationalized sport pressure with participation in an individual/team sport. They did not include any reference pertaining to this topic in the introduction.

The reviewer is thanked for this remark as it is a mistake that this operationalisation was included as it was not part of the analysis.

4) Page 12, line 299: Include a full stop after the question mark.

The reviewer is thanked for their attention to detail, this omission has been rectified.

Discussion:

General feedback: The discussion largely repeats the findings of the analyses, which is helpful to the reader. However, the authors do not give enough attention to theoretical explanations of the findings (i.e. page 21, lines 466-467: please provide explanations for the invariance of the model across lean and non-lean sports). In general, the authors need to spend less time re-stating their results and more time discussing them. Why are the important? What is the next step?

Furthermore, authors should discuss the clinical implications of the current study. Why your results pertaining to the Petrie and Greenleaf (2007; 2012) model are important in terms of clinical implications?

The author is thanking the reviewer for this recommendation. It has been fully notes and several additional sentences have been added to the discussion of the manuscript.

1) Page 20, line 440: Please include a full stop.

The reviewer is thanked for their attention to detail, this omission has been rectified.

2) Page 20, line 444: Authors should rephrase the discussion in accordance with the results they found. They should refer only to binge eating and bulimia symptoms instead of talking about disordered eating in general. See also page 21, line 488.

The author thanks the reviewer for this suggestion, it has been noted and carefully implemented throughout the discussion.

3) Page 21: Authors should include in the future directions section the inclusion of athletes younger than 18 yeas old, given that this population was overlook by authors but represents an high-risk population for the development of EDs.

The reviewer is thanked for their insight, this recommendation has been added to the manuscript.

Figures:

Please include the number and the legend for each figure.

The Reviewers are thanked for this comment, number and legends have been added to the figures.

Submitted filename: PLOS ONE Response to Reviewers.docx

Click here for additional data file.

27 Apr 2020

Psychological Risk Indicators of Disordered Eating in Athletes

PONE-D-19-34507R1

Dear Mrs Stoyel

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Valentina Cardi

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Both reviewers found your revisions satisfactory and have approved publication of the manuscript.

Reviewers' comments:

Reviewer's Responses to Questions

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: (No Response)

Reviewer #2: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: (No Response)

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

Reviewer #2: This is a revised version of a manuscript that I previously reviewed. The authors have adequately addressed my prior concerns.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

1 May 2020

PONE-D-19-34507R1

Psychological Risk Indicators of Disordered Eating in Athletes

Dear Dr. Stoyel:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Valentina Cardi

Academic Editor

PLOS ONE