A systematic review, meta-analysis, and meta-regression of the prevalence of self-reported disordered eating and associated factors among athletes worldwide

Background The purpose of this meta-analysis was to provide a pooled prevalence estimate of self-reported disordered eating (SRDE) in athletes based on the available literature, and to identify risk factors for their occurrence. Methods Across ten academic databases, an electronic search was conducted from inception to 7th January 2024. The proportion of athletes scoring at or above predetermined cutoffs on validated self-reporting screening measures was used to identify disordered eating (DE). Subgroup analysis per country, per culture, and per research measure were also conducted. Age, body mass index (BMI), and sex were considered as associated/correlated factors. Results The mean prevalence of SRDE among 70,957 athletes in 177 studies (132 publications) was 19.23% (17.04%; 21.62%), I2 = 97.4%, τ2 = 0.8990, Cochran's Q p value = 0. Australia had the highest percentage of SRDE athletes with a mean of 57.1% (36.0%-75.8%), while Iceland had the lowest, with a mean of 4.9% (1.2%-17.7%). The SRDE prevalence in Eastern countries was higher than in Western countries with 29.1% versus 18.5%. Anaerobic sports had almost double the prevalence of SRDE 37.9% (27.0%-50.2%) compared to aerobic sports 19.6% (15.2%-25%). Gymnastics sports had the highest SRDE prevalence rate, with 41.5% (30.4%-53.6%) while outdoor sports showed the lowest at 15.4% (11.6%-20.2%). Among various tools used to assess SRDE, the three-factor eating questionnaire yielded the highest SRDE rate 73.0% (60.1%-82.8%). Meta-regression analyses showed that female sex, older age, and higher BMI (all p < 0.01) are associated with higher prevalence rates of SRDE. Conclusion The outcome of this review suggests that factors specific to the sport affect eating behaviors throughout an athlete's life. As a result, one in five athletes run the risk of developing an eating disorder. Culture-specific and sport-specific diagnostic tools need to be developed and increased attention paid to nutritional deficiencies in athletes. Supplementary Information The online version contains supplementary material available at 10.1186/s40337-024-00982-5.


Introduction
Eating disorders (EDs) are serious, all too often chronic, mental illnesses that usually begin in adolescence or young adulthood [1,2].Approximately thirty million people globally suffer from EDs, disorders that are frequently misdiagnosed and under (ineffectively) treated [3].There is general agreement that a variable and complicated combination of biological, psychological, social, and cultural factors cumulatively increases the risk of ED [4].
In the general adult population, anorexia nervosa (AN) and bulimia nervosa (BN) are clinically recognized EDs that, at least in part, reflect the extremes of a continuum of traits that include negative body image, food restraint, and preoccupation with weight and musculature in the formation of self-identity [5,6].According to longitudinal risk factor studies [5,6], negative body image and "disordered eating" (DE), and "screening for at-risk" individuals is where this continuity assumption is most visibly present [7,8].Continuity assumption in this context refers to the assumption that DE lies on a continuum from normal eating to clinical disorders [9].
Negative body image and DE are the strongest predictors of the development of full-blown EDs [10].Together, they constitute an "at risk" state [11].While usually associated with women, and attributed to occupations such as modeling or ballet where weight must be restricted, EDs are also prevalent in males, and especially in both male and female sports [12,13].They are most likely to develop in athletes who compete in sports that a) promote leanness as a way to improve performance, b) divide team eligibility according to weight, or c) receive points based on appearance [14].Athletes' risk of acquiring EDs varies according to their gender, sport, and also level of competition [15].Self-reported measures for DE et al., (2022) indicated that overall, 35% (1%-62%) of Jordanian athletes were at substantial risk for ED [20].As mentioned earlier, several factors explain discrepancies among reports and account for the broad prevalence range cited by Ghazzawi et al. [20][21][22][23].
Recent studies point to the promotion of a healthy body image as a helpful foundation for addressing athletes' DE [24,25].Athletes may be susceptible to EDs because the prevalent idealized body image in sports emphasizes leanness or low body weight, particularly so in endurance, aesthetic, and weight class competitions [15,24,26].Other contributing factors are: early sport-specific training, participation in weight-class sports, the engrained habit of frequent dieting and subsequent weight fluctuation [27], the wearing of skimpy sportswear, coaching pressure, and involvement in elite-level competition [20,26,28].Female athletes are considered more vulnerable than males to concerns regarding body mass and shape; they may, therefore, use more extreme weight-control techniques than men [29].[29]Athletes who base their identity and, hence, their self-esteem, on their love of the game they play have less trouble with SRDE than those whose identity is based on their personal game skills and successes (which fluctuate).Low self-esteem is a significant risk factor leading to body image problems and subsequent ED [15].
Athletes may not, however, be more at risk for ED than the general population.Some meta-analyses [21] have shown a higher frequency of risky eating behaviors in non-athletes.Either way, the evidence for an elevated risk of EDs among athletes is not conclusive [22,30,31] In an attempt to help resolve the debate and to extend our previous meta-analytic reviews of SRDE in athletes [5,6,32], we conducted a meta-analysis of its global prevalence.To the best of our knowledge, based on searches of the literature and various registration platforms, this is the first such meta-analysis of DE and potential moderators in this population.In this meta-analysis, we aimed to synthesize the available literature to provide a pooled estimate of the prevalence of SRDE in athletes.

Materials and Methods
The study's methodology was submitted to Open Science Framework (OSF; https:// doi.org/ 10. 17605/ OSF.IO/ EJ2QN) in September 2022.OSF is a freely available platform that enables researchers to share their research ideas with peers and to receive assistance over the duration of a study.
The review follows the guidelines for reporting, as recommended by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA2020) [33].The Meta-analysis of Observational Studies in Epidemiology (MOOSE) [34] procedure was followed in relation to the statistical analyses and to the reporting of results.

Search strategy
In September 2022, two authors (HJ and HG) conducted an online literature review using ten databases.1) Pub-Med/MEDLINE, 2) American Psychological Association PsycINFO, 3) ScienceDirect, 4) Springer, 5) EBSCOhost, 6) Embase, 7) Cumulative Index to Nursing and Allied Health Literature (CINAHL), 8) Scopus, 9) Web of Science, and 10) SportDiscus.The following keywords and lists were used in the full-text search: Athletes, gymnasts, or elite, are on List A. On List B are eating disorders, eating habits, feeding disorders, eating symptoms, eating attitudes, or eating issues.The punctuation marks (*) ensure that the phrase's reverse word order is considered in the search method.For instance, "disordered eating" and "eating disorders" are included when searching for "eating disord*." The authors also manually searched the citation lists of included articles.The manual reference list checking helps safeguard against missing relevant literature, but incorporating as wide a range of terms as possible directly into the database searches aids the process.Meta-analyses that exclude grey literature have a higher propensity to overstate effect sizes and thus produce less accurate effect size estimations.As a result, we deliberately sought out organizational reports, unreleased studies, and research in lesser-known journals.
Three members of the team (LN, AH, and HT) independently evaluated the initial article selection.After screening titles, abstracts, and complete texts and eliminating duplicate studies, initial data were extracted, and their quality was assessed independently by two team members (LN and AH).Discussion among all team members was able to resolve any differences regarding a study's eligibility.

Criteria for exclusion, inclusion, and eligibility (selection process)
The complete text of English-language papers on SRDE among athletes of all sports types and all ages around the globe was obtained for meta-analysis.
Inclusion criteria: 1) English language 2) original research 3) all types of sports, all ages, anywhere in the world 4) the prevalence of DE needed to be included 5) evaluation was performed by a screening tool with welldefined cut-offs for clinical diagnosis of eating disorders.
Exclusion criteria: 1) studies of current or former players who did not meet athlete criteria at the time of the study 2) mental health studies in athletes unrelated to DE 3) data unavailable even after attempt to contact authors directly 4) book chapters and review articles (systematic and meta-analysis) 5) studies in which prevalence rates were reported but only in the form of group means.
The PRISMA2020 process flow for research selection is shown in Fig. 1.

Procedure (data collection process)
The 648 included papers were scanned and categorized using ASReview, an online free service that integrates digital tools (such as natural-language processing) with machine learning tools.By using a free, open-source, web-based program WebPlotDigitizer v4.5, data could be taken from plot images.
To ensure uniform data description, authors LN, AH, HT, and HG individually collected the following eight factors: author names, year of publication, country of data collection, representativeness of the sample, average BMI (kg/m 2 ), gender (% of female participants), and average age (in years), and method utilized to identify SRDE presence or absence.Using the regional categories of member states created by the United Nations, this metaanalysis study included data from forty countries, which were then categorized as Eastern or Western.Eastern countries were those with emerging markets or economies in development.Western countries are defined as advanced economies with high GDP per capita, such as those in Western Europe, the U.S., Australia [35].

Evaluation of study quality and bias risk (risk of bias assessment)
Participant selection (sample), comparability, and statistical analysis/results make up the first three items on the Newcastle-Ottawa Scale (NOS) checklist [36].Each item in the NOS is rated on a scale of one to three (or four) stars.The highest possible rating for each study, therefore, would either be nine (for cohort studies and crosssectional studies) or ten stars (case-control studies).Studies that score an eight are still judged to be of good quality with little chance of bias.A 5-7 score is judged to be of moderate quality with some chance of bias, and a 0-4 score is low quality with a significant chance of bias.
As noted earlier, the prevalence estimate for each study was based on a pre-defined cut-off score of a validated screening tool, that is, a continuous measure of DE risk such as the EAT-26 and Sick, Control, One stone, Fat, Food test (SCOFF) (Table 1).

Data visualization and analysis (synthesis method)
With the understanding that the real effects will change over time, a traditional meta-analysis utilizing the random-effects approach was applied.We estimated and corrected for the between-study variation in effects using the DerSimonian-Laird technique and the generic inverse contrast technique with logit transformed (PLO) ratios.The random effect model was adopted (e.g., EAT and SCOFF; Table 1) since various study sets assess distinct, conceptually connected consequences using various metrics.Ninety-five percent confidence intervals, and pooled prevalence are presented for each study.
A forest plot was used to present the data.Forest plots have the limitation of only offering confidence intervals at significant levels, i.e., p < 0.05.Drapery plots and p-curve analysis were also used since confidence intervals are needed to ascertain whether results are significant and replicable.The p-value function is represented in the   drapery plot by curves which stand for the forecast value from one study to the next, as well as for the combined meta-analytic values.We employed the I 2 statistic to estimate betweenstudy variability; a result between 75 and 100% indicates a greater standard of heterogeneity.Additionally, we assessed heterogeneity using tau 2 (τ 2 ), tau (τ), as well as Cochran's Q statistics.The square root for the following is the H statistic: Cochran's χ2 heterogeneity statistic divided by the level of freedom.We utilized a standard Galbraith radial plot to illustrate heterogeneity, where the opposite of standard errors (horizontal axis) is shown in relation to the actual effect magnitude or results normalized with their corresponding standard errors (vertical axis).An arc displays matching effect sizes or outcomes just on the right side of an entire Galbraith plot.Outlier inclusion could reduce the reliability and validity of meta-analysis.Studies are characterized as outliers whenever their confidence intervals do not match those of the pooled effects and sensitivity analysis can be used to address these studies.To avoid having an excessive amount of influence from any given study, we used a Jackknife sensitivity analysis to exclude studies one at a time.In this analysis, the primary meta-analysis is repeated one additional time for every study analyzed, with a different study being discarded every time.
Publication bias can occur if, for instance, studies with positive results are published more often than studies with negative results.As a first visual, funnel plots were used to examine publication bias.The Doi plot replaces the traditional scatter plot (funnel plot) of the ratio of accuracy to efficiency with a plot of the ratio of the normal quantile (z-score) to efficiency [37].If there is asymmetry, either the two ends of the graph are unequally offset from the midpoint, or one end has more trials than the other.If there is no asymmetry, a line drawn perpendicular to the x-axis from the top of the Doi graph is expected to divide the graph into two areas of equal area; the LFK index quantifies the difference between the two areas as the difference between their respective areas under the graph and the number of studies included at each edge; an LFK index closer to 0, the more symmetrical the Doi graph; LFK index values outside the range between -1 and + 1 are considered consistent with asymmetry (i.e., publication bias) [38].The trim and fill method was applied (if needed) to develop modified point estimates.Peters' correlations and Egger's regression also were employed as additional benchmarks for a more thorough analysis of publication bias.
Subgroup meta-analyses were utilized to look into varied results and respond to particular questions about various study characteristics or populations.Subgroup analyses were performed on categorical variables including participants' nationality, culture (Western versus Eastern), and frequently included clinical characteristics.Sports energy system and sports category subgroup meta-analysis was performed to examine differences between aerobic, anaerobic, and mixed sports.Aerobic sports are defined as sports that require prolonged aerobic metabolism.This includes long-distance running, cycling, swimming, rowing, and cross-country skiing [39].Anaerobic sports are defined as sports relying on anaerobic metabolism and involving short bursts of high-intensity activity such as sprinting, weightlifting, and gymnastics [39].Mixed sports are defined as sports that involve both anaerobic and aerobic energy systems, e.g., soccer, basketball, tennis, volleyball, lacrosse, and hockey [39].
Any subgroup of four papers or more was addressed by the subgroup meta-analyses, and all findings were represented by graphical representation utilizing forest plots.
In essence, meta-regressions are essentially regression models where the outcomes of one or more explanatory factors have been used to forecast the amount of variance.Meta-regressions were conducted to examine associations between sample characteristics and SRDE prevalence estimates [40].When an explanatory variable is raised by one unit in a meta-regression study, the regression coefficient shows how the output variable changes [40].R 2 was used to calculate the effect size in statistically significant meta-regression models, where 1-8%, 9-24%, and ≥ 25% of the variance corresponded to small, medium, and major effect size [40].
All data were analyzed using statistical computer software termed R. All classic meta-analyses were carried out using the packages "meta" and "metafor".Risk-of-bias plotlines were produced for use in evaluating quality, using the package "robvis".An overview plot (weighted) was developed for all studies to indicate the amount of data contained within every judgment for each topic.A traffic light plot shows the total risk as well as the bias risk for each domain.

Descriptive results
The literature search, ending 7th January 2024, yielded 4178 relevant studies.One hundred and thirty-two unique studies involving one hundred and seventy-seven studies (due to multiple screening tools or multiple data collection times) with a total sample size of 70,950 participants (K = 177 data points; N of participants = 70,950) met our inclusion and exclusion criteria.The details of those studies are shown in Table 1.The PRISMA2020 process flow for research selection is shown in Fig. 1.
Twenty-seven countries were represented in our sample (Table 1), of which 92.1% were categorized as Western culture (33.3% were from the USA).Thirty percent of the studies addressed aerobic sports systems although most studies (59.9%) reported on sports that were mixed between aerobic and anaerobic exercise energy systems.Only 10.7% were purely anaerobic.Moreover, while 54.2% of the studies were of mixed indoor/outdoor sports, 26.5% were of purely outdoor sports.The EAT-26 was used in 31.1% of the studies (Table 2).A clear majority 88.1% of the studies were cross-sectional, while, in 11.9%, the design was longitudinal.None of the studies meeting the inclusion and exclusion criteria were conducted during the lockdown period of the COVID-19 pandemic.Figure 2 shows that approximately 80% of the

Meta-analysis of the prevalence of SRDE
Meta-analysis results and the raw prevalence data of each study are presented in Fig. 3.According to the random-effects meta-analysis, the SRDE prevalence among athletes (K = 177, N = 70,957) was (95% CI) = 19.23%(17.04%; 21.62%),I 2 = (95% CI) = 97.4% (97.2%-97.6%),τ(95% CI) = 0.9481 (0.9374; 1.2182), τ2 = 0.8990 (0.8787-1.4840),H (95% CI) = 6.17 (5.95%-6.40%),p-value Cochran's Q = 0.00.Visual inspection of the funnel plot (Fig. 4), radial plot (Additional file 2), DOI plot (Additional file 3) as well as non-significant (at 0.05) Egger's regression and Peter's tests, indicated that the likelihood of publication bias was low.Additional file 4 represents the Drapery plot of DE in athletes; the drapery plot displays all primary study P-value functions in one graph, along with a P-value curve for each pooled estimate and a shaded prediction region.In contrast to the forest plot, the drapery plot displays results for any possible confidence level, not just one arbitrary confidence level.The results from our Drapery plot confirm that the results of the original/primary studies included are replicable.Additional file 5 presents possible outliers.The effect of outliers was minimal and required no accommodation.

Meta-regression and subgroup meta-analyses
Table 2 presents the analyses of subgroups, with the stipulation that there are at least three studies in a subgroup.

Country and culture
Additional file 9 and Additional file 10 show the weighted prevalence levels of SRDE as a function of the country and culture in which the data were collected.These varied substantially, and, as noted above, for many countries the number of studies meeting the inclusion and exclusion criteria was low.Australia (K = 11, N =

Sports energy system and sports category
Additional file 11 shows the weighted prevalence levels of SRDE according to the energy system (aerobic vs anaerobic) of the sport.There were large variations.Most studies reported a mix of energy systems (k = 106, N = 52,842) with 17% (14.6%-19.4%).However, aerobic energy system sports were nearly three times more often reported on as anaerobic ones, (k = 52, N = 16,287) versus (k = 19, N = 1,828) respectively.Anaerobic sports had the highest prevalence rate of SRDE at 37.9% (27%-50.2%).Aerobic sports came next at 19.6% (15.2%-25%), while mixed sports showed the lowest prevalence rate of SRDE at 17% (14.6%-19.4%).
Results according to the sport were very similar, as shown in Additional file 12, which illustrates the fact that mixed sports were most commonly studied and swimming was the sport least studied, (k = 96, N = 39,341) ) for outdoor sports.Gymnastics had the highest prevalence rate of SRDE at 41.5% (30.4%-53.6%).Martial arts had the next highest prevalence rate of SRDE at 26.3% (18.1%-36.4%),while outdoor sports had the lowest prevalence rate of SRDE at 15.6% (11.6%-20.2%).
Seven tools were used only once (k = 1); the Leisuretime Exercise and Anorexia Nervosa Questionnaire

Study design
Almost one-fifth of studies (k = 156) were cross-sectional involving (N = 67,538) participants.Only 21 studies were cohort studies, with 3,419 participants.The prevalence of SRDE in cohort studies was 22.2% (18.1%-27%), as shown in Additional file 14.

Discussion
The purpose of this meta-analysis was to estimate the prevalence of DE in athletes, based on published studies that used validated screening measures.The search and selection process resulted in 177 studies, conducted in 27 countries (total N of participants = 70,957).This meta-analysis indicates an estimate of CI 95% = 19.23%(17.0%-21.6%)as the global prevalence of SRDE in athletes.This is higher than in the general population [20].Given the physical and mental health consequences of DE and considering its status as a risk factor for clinically significant EDs, the prevalence rates are considered high [20].

Fig. 3 Meta-analysis of disordered eating in athletes
The most frequently associated with SRDE was competing in gymnastics and sports divided along weight lines [20,[41][42][43].With advance knowledge of risk, health care professionals are better equipped to provide early treatment when required [17,44].In general practice, this knowledge paired with an ongoing patient-provider trusting relationship, puts the physician in a unique position to diagnose and treat SRDE, thus preventing progression to an ED [23].Knowledge about risk factors is also important for health promotion and targeted preventive measures.Results of this research are recommended reading for athletes as well as trainers, coaches, physicians, Olympic committees and administrators of athletic organizations [28,[45][46][47].
The prevalence estimate of 19% was, significantly associates with age, the proportion of females in the sample, and sample's mean BMI level.The prevalence of SRDE was higher in Eastern countries.Of interest to Arab countries, the prevalence of SRDE in athletes was only available in two Arabic countries: 36.6% among Saudi athletes [48] and, in Jordan 33.7% [20].
Since self-screening leads to early risk identification and effective prevention, it will be important to survey self-identified DE and examine its history and symptom patterns in more detail [49].In a recent study of 249 Jordanian athletes aged over 18 years, Ghazzawi et al. [20] found an overall prevalence of SRDE of 33.7%, which is far higher than the mean estimate of 19.23% in our metaanalysis.In the Ghazzawi et al., 2022 study, outdoor sports (running, cycling, and walking), relevant to 8% of the participants, showed the lowest percentage of DE (3% of the 34%); the highest percentage (10% out of the 34%) was among gymnasts who accounted for 30% of all participants [20].
More studies, using established epidemiological methods (e.g., representative sampling), are needed to clarify the prevalence and correlates of SRDE in athletes.In this regard, the published research to date has been limited to samples from only 27 countries, that is about 10% of the world's approximately 250 independent territorial entities.Notably absent in the literature that met our criteria are studies from Latin American countries (other than one from Brazil), and there is either zero or one sole study from several countries that have contributed to the general literature on DE, such as Sweden, New Zealand, Portugal, Slovenia and the Middle East countries (other than Saudi Arabia and Jordan).
Researchers seeking to understand the prevalence of DE as a multifaceted construct have many screening tools from which to choose (Table 1).Based on the substantial variability (heterogeneity) of the prevalence estimates in our meta-analysis, and in order to facilitate comparisons across studies from different countries while avoiding estimates that are almost certainly far too high or too low, we recommend the standardized use of the EAT-26 [50] plus the TFEQ [51].
The EAT-26 has been extensively validated across ages, sex groups, and cultural contexts as a screening tool with Fig. 4 Funnel plot of disordered eating in athletes strong psychometric properties for identifying DE attitudes and behaviors [50].The TFEQ thoroughly assesses different problematic eating patterns such as restrained-, and emotional-eating that may represent risk factors for the development of EDs [29,52].Together, these two instruments capture a comprehensive range of symptoms and eating behavior anomalies indicative of risk.The self-report format makes them possible to use for large population screenings where clinician interviews are impractical.Furthermore, both questionnaires have been translated and validated in various languages, enabling consistent implementation globally [29,52].Standardizing to widely-used instruments like the EAT-26 and TFEQ with robust psychometric properties that comprehensively assesses DE will help generate prevalence estimates across studies that are more homogeneous, allowing for meaningful cross-national comparisons and meta-analytic examination [29,52].
The TFEQ, which was used in five data points [29,52] included in this meta-analysis, is a valid and widely used measure of ED behaviors, and, therefore, it can add behavioral information to the screening items included in the EAT-26.If the EAT-26 is impractical due to its length, then we recommend substituting the 5-item SCOFF, which, according to Table 1, has thus far been used in only seven studies of SRDE in athletes [15,[53][54][55][56].
TFEQ has been previously used in studies samples that included athletes, its psychometric properties and validity have not yet been investigated specifically in athletic populations, representing an important direction for further research [51].
Furthermore, the hormonal and metabolic dysregulation of DE can negatively impact athletic performance in various ways.Decreased estrogen and testosterone levels hinder the building and maintenance of muscle mass [11,57].Reduced IGF-1 and growth hormone lead to impaired bone development, and increased fracture risk [11,57,58].Electrolyte abnormalities from purging behaviors can disrupt cardiac function [57,58].Low energy availability alters substrate metabolism, making it difficult to meet energy demands during training and competition [11,57].Extreme weight loss and nutritional deficiencies can also impede recovery after exercise [11,57,58].

Study strengths and limitations
To the best of our knowledge, this is the first metaanalysis of the prevalence of SRDE among athletes.The substantial number of studies and participants included adds strength to this review.Limitations include the reliance of many studies on convenience samples rather than representative samples and our inability to review articles published in languages other than English.The limited nature of the information about participants also ruled out statistical examination of potentially important moderating variables such as ethnicity, immigration status, sexual orientation, and family history.
An additional limitation of our analysis stems from the small number of studies conducted in Eastern countries, as well as our ability to examine differences among countries solely along the Western vs. Eastern dichotomy.The small sample of studies from Eastern nations limits conclusions about the seemingly high prevalence rates found in these settings.Furthermore, grouping diverse countries into binary Western and Eastern categories obscures important heterogeneity within regions in terms of specific sociocultural factors influencing DE.Future research should include more studies across a wider range of Eastern countries and examine country-level differences using more nuanced categorization systems.Relatedly, our study was limited by its focus on the gender binary, when DE prevalence among transgender and nonbinary populations warrants dedicated investigation.
To maximize the breadth of the literature captured, we did manually search the reference lists of included studies to help identify any relevant articles that our search may have missed.However, we acknowledge that it is possible we may have inadvertently excluded some studies by not including a wider range of activity-specific terms in the database searches.For future updates to this meta-analysis, we will consider broadening the search to include additional keywords related to different sports, as well as other physical activities such as dance, mountain climbing, boating, and others.

Conclusion
The prevalence of SRDE in a meta-analysis of a very large sample of athletes from 27 countries was shown to be 19.23%, or one in every five athletes.DE is a health problem in itself, but it is also a strong risk factor for EDs.Our meta-analysis strongly suggests that, in the sports world, it affects females and males to an equal degree.While concentrating on sports injuries, public health has neglected the nutritional needs of athletes and the health consequences of dietary deficiencies in persons with high energy expenditure.

Table 1
Characteristics of the studies involved in the systematic review and meta-analysis about the prevalence of disordered eating in Athletes [81]De Borja et al., a 2021 USA Female sex = 100%; mean age 19.8 years; mean BMI 22.86 kg/m 2 Cross-sectional study M1 = BEDA-Q M2 = RD M3 = ESP [82] Dervish et al., b 2023 USA Female sex = 100%; mean age NR years; mean BMI NR Kg/m 2 Cross-sectional study FAST

Table 2
Random and common effects meta-analysis models of the prevalence of disordered eating in Athletes d Significant differences between samples in meta-analysis e Detects publication bias in meta-analysis f Represent the correlation between effect sizes and sample variation