The syndrome Relative Energy Deficiency in Sport (RED-S) describes impairments of multiple physiological functions including energy metabolism, reproductive function, bone health, immune function, protein synthesis, and cardiovascular health (Mountjoy et al., 2014, 2018). The etiology behind this syndrome is low energy availability (LEA), which can occur with or without disordered eating behavior or eating disorders (Mountjoy et al., 2014, 2018). Psychological factors can therefore precede RED-S, but LEA may also result in significant psychological distress (Mountjoy et al., 2014; Ackerman et al., 2019; Langbein et al., 2021; Rogers et al., 2021). LEA seems to affect both males and females, elite athletes and recreational exercisers (Torstveit and Sundgot-Borgen, 2005; Mountjoy et al., 2014, 2018), able-bodied athletes and para-athletes (Brook et al., 2019) and across all age groups (Sharps et al., 2021).

Although any athlete can suffer from RED-S, athletes participating in weight-sensitive sports (i.e., endurance sports, combat sports, and aesthetic sports) seem to be particularly at risk. Endurance athletes also have high exercise energy expenditure (Sjödin et al., 1994) as an additional risk factor for LEA (Burke et al., 2018). Furthermore, some studies report gender differences in the sensitivity of LEA as well as the endocrine and metabolic responses, indicating a vulnerability of LEA related to females compared to males (Koehler et al., 2016). Therefore, female endurance athletes seem to be a high-risk group for RED-S.

In female endurance athletes, the reported prevalence of LEA ranges between 8 and 56% when defined as energy availability <30 kcal/kg fat-free mass/day assessed with food and training records (Monyeki et al., 2014; Melin et al., 2015; Day et al., 2016; Muia et al., 2016; Heikura et al., 2017; Mccormack et al., 2019; Beermann et al., 2020). It has been suggested that assessing self-reported physiological symptoms of LEA using questionnaires, such as the Low Energy Availability in Females Questionnaire (LEAF-Q), provides a better assessment of the overall health status of an athlete than a snapshot of current energy availability using error-prone assessments of dietary energy intake and exercise energy expenditure (Heikura et al., 2017; Burke et al., 2018; Sim and Burns, 2021). When assessed as a LEAF-Q score ≥ 8, the LEA risk rate ranges from 31 to 80% in female endurance athletes (Melin et al., 2014; Folscher et al., 2015; Heikura et al., 2017; Carr et al., 2019; Ihalainen et al., 2021; Jesus et al., 2021).

Sustained research on factors associated with LEA is important for practitioners and other health professionals for early detection and aid in the nutrition interventions to prevent potential decrements in health and performance (Melin et al., 2019; Ackerman et al., 2020). The association between disordered eating behavior and health consequences as a result of LEA in female athletes is well-researched and suggest that particular athletes from weight-sensitive sports, including endurance sports, may be at higher risk (Gibbs et al., 2013). While the prevalence of disordered eating behavior, which may lead to LEA, is reported to be higher among athletes competing in weight-sensitive sports compared with controls (Torstveit et al., 2008; Wasserfurth et al., 2020), LEA without disordered eating behavior is also common among female endurance athletes (Melin et al., 2016).

Another potential reason for LEA might be the exclusion or avoidance of different foods due to food allergies or intolerances. Although self-reported incidences of food intolerances and adherence to special diets are commonly reported in athletes (Lis et al., 2015, 2019; Logue et al., 2019), associations with LEA and RED-S among female endurance athletes have not been examined yet.

High levels of exercise energy expenditure without a corresponding increase in energy intake can result in LEA (Burke et al., 2018). It has been proposed that athletes with propensity to exercise addiction, characterized as excessive exercise behavior with potential negative consequences such as injuries and mental health problems (Griffiths et al., 2005; Lichtenstein et al., 2017), could increase the vulnerability for RED-S (Turton et al., 2017). We have previously reported an association between symptoms of exercise addiction and markers of RED-S in male endurance athletes (Torstveit et al., 2019). However, research investigating the association between exercise addiction and symptoms of RED-S among female endurance athletes is scarce with only one study investigating this association in a group representing 63% female endurance athletes (Kuikman et al., 2021). This study reported an exacerbated risk of LEA when disordered eating is accompanied with exercise addiction compared to disordered eating in isolation (Kuikman et al., 2021).

Therefore, the aim of the present cross-sectional study was to identify the risk of LEA and associated risk factors in a multi-country cohort of competitive female endurance athletes. Specifically, we aimed to compare disordered eating behavior, exercise addiction, and food intolerances in athletes at risk of LEA vs. low risk of LEA. Finally, we aimed to assess explanatory variables for the risk of LEA in this cohort. We hypothesized that athletes with risk of LEA would report more disordered eating behavior, exercise addiction, and food intolerances, have lower BMI and higher training volume compared to athletes with low risk of LEA.

The present analysis is based on cross-sectional data collected during the screening and inclusion phase of an international multicenter intervention study aiming to induce health behavior change and improve nutritional status in female endurance athletes with symptoms of RED-S.

The main purpose of the study was to identify the risk of LEA and associated risk factors in this group of athletes, including exercise addiction and food intolerances where research has been lacking. Specifically, we aimed to compare disordered eating behavior, exercise addiction, and food intolerances in athletes at risk of LEA and low risk of LEA. A key novel finding of the study was that athletes at risk of LEA were more likely to exhibit symptoms of exercise addiction related to salience, conflicts, and withdrawal symptoms.

By using the LEAF-Q we found that 65.0% of the athletes were at risk of LEA, which is similar to the 62.2% clinically verified LEA in female endurance athletes by Melin et al. (2014). When compared to other studies which used the LEAF-Q in female endurance athletes, Carr et al. (2019), reported a lower number (31.0%) than the present study and Jesus et al. (2021) a higher number (79.5%.) In contrast to the present study, the study by Jesus et al. (2021) recruited only female athletes from cross-county running (n = 83) and when comparing their reported risk rate of LEA, it is similar to the risk rate among runners (85.2%) and orienteers (77.8%) in the present study. Carr et al. (2019) also had a lower number of participants (n = 13) only from cross-country skiing, which may partly explain the lower reported risk rate of LEA. Of note, the current cross-sectional study served as a screening phase for an intervention study aiming to improve dietary habits in athletes at risk of RED-S. Therefore, the recruitment may have attracted athletes who were interested in improving knowledge related to sports nutrition. Conversely, others may have not participated due to fears or anxiety coming from their LEA, disordered eating behavior and/or exercise addiction.

In this study we did not directly assess energy availability, which requires assessments of body composition, energy intake, and exercise energy expenditure. Although it is well-recognized that LEA is the etiological factor underpinning the syndrome of RED-S (Mountjoy et al., 2014), several barriers prohibit the direct measurement of energy availability from being a practical and reliable option (Heikura et al., 2017; Burke et al., 2018; Fahrenholtz et al., 2018; Mountjoy et al., 2018). Questionnaires can therefore be a convenient method for screening, with the LEAF-Q (Melin et al., 2014) and the EDE-Q (Fairburn and Beglin, 1994) being the most widely used and validated questionnaires in the research of RED-S (Sim and Burns, 2021).

We emphasize that although the LEAF-Q is useful for screening purposes it cannot be used as a diagnostic tool and additional individual evaluation is necessary in intervention contexts (Melin et al., 2014; Mountjoy et al., 2018). Despite acceptable sensitivity (78%) and specificity (90%) (Melin et al., 2014), there are risks of either false positive or false negative classifications. For instance, false positive cases may be due to injuries, gastrointestinal problems, and menstrual dysfunctions not related to LEA, including acute injuries, irritable bowel syndrome, or polycystic ovary syndrome, although features of polycystic ovary syndrome have been found to be prevalent in women with functional hypothalamic amenorrhea (Carmina et al., 2018). False negative cases may occur if participants use oral contraceptives, that may mask an underlying menstrual dysfunction. Other forms of hormonal contraceptives may, on the other hand, result in absence of bleeding, which accordingly can falsely be interpreted as a menstrual dysfunction.

Of note, 12.0% answered that they did not know whether they had a normal cycle and 18.7% of those who reported normal menstruation answered that they had irregular periods. This controversy indicates that a relatively high frequency of adult female endurance athletes lacks basic knowledge of female reproduction. In addition, only two athletes reported a diagnosed menstrual dysfunction, despite a high occurrence of amenorrhea and oligomenorrhea, which may relate to a perception that menstrual dysfunction often is accepted as a natural consequence of intense training programs (Beals and Meyer, 2007) or lack of awareness about the consequences of LEA among medical professionals (Curry et al., 2015). This is concerning since the menstrual cycle is an important health marker and calls for educational interventions on menstrual cycle function for female endurance athletes but also awareness among medical professionals.

In the present study, athletes at risk of LEA had lower BMI compared to athletes with a low risk of LEA and lower BMI was identified as a risk factor in the logistic regression analysis. Similar to our findings Christo et al. (2008) reported lower BMI in female endurance athletes with amenorrhea compared to their eumenorrheic counterparts. In contrast, Vanheest et al. (2014) reported higher BMI in amenorrhoeic swimmers compared to eumenorrheic swimmers, while Melin et al. (2014, 2015) found no difference in BMI comparing female endurance athletes at risk of LEA with those with a low risk of LEA. Thus, research seems conflicting when it comes to the relationship between BMI and LEA. In addition, it is important to notice, that the vast majority of athletes at risk of LEA in the current study had a BMI within the normal range, supporting that BMI alone should not be used for screening of RED-S due to potential metabolic compensatory mechanisms as a result of LEA (Mountjoy et al., 2014).

In this study, athletes at risk of LEA had higher EAI scores compared to athletes with low risk of LEA. Our results indicate that female endurance athletes with LEA are more likely to have symptoms of exercise addiction. However, in approximately half of the cases of athletes exceeding the cut-off point, exercise addiction was accompanied with disordered eating behavior. Therefore, both dietary patterns and exercise behavior should be addressed when treatment interventions for athletes with RED-S are developed. Although exercise addiction is often associated with excessive exercise (Lichtenstein et al., 2017) that may result in LEA, we found no association between training volume and EAI score, which is consistent with previous research in athletes (Lichtenstein et al., 2021b). Nor did we find a difference in the EAI item concerning increasing exercise amount when comparing athletes at high vs. low risk of LEA. One explanation for this can be that all participants in the current study had a high training volume and frequency, since this was a part of the inclusion criteria (being a competitive endurance athletes and training at least five times per week). For competitive endurance athletes it is a natural part of their training protocol to increase exercise amount to improve performance. However, we found an association between LEA and the EAI items concerning the importance exercise is attributed in life, conflicts with family and friends, and negative feelings if an exercise session is missed. The findings suggest that these exercise addiction related behaviors are more pronounced in athletes at risk of LEA, and therefore practitioners should pay attention to these personality characteristics when screening for LEA and RED-S.

Injuries are a potential consequence of both exercise addiction (Lichtenstein et al., 2017) and LEA (Mountjoy et al., 2014) and questions related to injuries are a part of the LEAF-Q due to its association to low bone mineral density (Melin et al., 2014). We found a higher injury score in athletes at risk of exercise addiction compared to athletes with low risk of exercise addiction, which may be a part of the explanation for the association between risk of LEA and exercise addiction in this group.

As reviewed by Di Lodovico et al. (2019) endurance sports are characterized with the highest risk of exercise addiction with a weighted average of 14.2% compared to 10.2% in mixed disciplines and 3.0% in the general population, although the reported risk rate spans from 0.5 to 43.0%. The risk rate of 23.3% reported in the present study is therefore considerably higher compared to previous research. The review did not distinguish between male and female responders, although female triathletes have been found to be at higher risk of exercise addiction compared to males (Griffiths et al., 2015). Hence, one potential reason for the differences in the results may be that our studied population is at high risk of disordered eating behavior. Exercise addiction is a frequent comorbid condition in female patients with eating disorders, often termed secondary exercise addiction, and therefore, eating disorder symptoms should be considered separately in the assessment of exercise addiction (Cook and Luke, 2017). Among athletes without disordered eating behavior (n = 159), 13.2% (n = 21) were at risk of exercise addiction, suggesting a risk rate of 10.4% for primary exercise addiction for the total sample. This is higher than the 4% among female athletes reported by Kuikman et al. (2021). Beyond recruiting athletes from different sports with no upper age limit, the study by Kuikman et al. (2021) differs from the present study by using the Exercise Dependence Scale, which generally identifies a lower proportion of individuals at risk of exercise addiction compared to the EAI (Di Lodovico et al., 2019). However, the occurrence of secondary exercise addiction in the present study (13%, n = 26) is similar to the 13% reported by Kuikman et al. (2021).

Since the EAI was originally validated in habitual exercisers, the interpretation of the questions by athletes may differ and researchers have suggested that exercise addiction is not the same phenomenon in competitive athletes as in non-competitive athletes (de la Vega et al., 2016). However, a recent study reported adequate psychometric properties of the EAI in a sample of elite athletes, suggesting that the EAI is useful for assessing symptoms of exercise addiction in competitive athletes (Lichtenstein et al., 2021b).

Disordered eating behavior and eating disorders are a well-known risk factors for LEA (Wasserfurth et al., 2020) supported by the findings in the present study. Nevertheless, the reasons for LEA are manifold (Wasserfurth et al., 2020) and in the present study, only 26.5% of the athletes at risk of LEA had disordered eating behavior, supporting the findings of Melin et al. (2014) where 28.6% of the female endurance athletes with a total LEAF-Q ≥8 had a clinically verified eating disorder or disordered eating behavior. Our results suggests that for the majority of the athletes, LEA is due to unintentional origins such as suppression of appetite post exercise (Larson-Meyer et al., 2012; Howe et al., 2016), low energy-dense diets (Melin et al., 2015), lack of knowledge regarding optimal sports nutrition (Benardot, 2013; Trakman et al., 2016; Heikkilä et al., 2018; Sim and Burns, 2021), lack of knowledge about the consequences of LEA (Folscher et al., 2015; Condo et al., 2019; Tosi et al., 2019; Logue et al., 2020) or a busy lifestyle with frequent traveling where lack of time and access to food become important barriers to adequately fueling (Benardot, 2013; Burke et al., 2018; Logue et al., 2021). Nevertheless, disordered eating behavior was common among this group of female endurance athletes with a risk rate of 21.3% using a EDE-Q global score 2.5 as cut-off compared to 24-25% earlier reported in elite female endurance athletes (Sundgot-Borgen and Torstveit, 2004; Melin et al., 2015) using the gold standard EDE Interview (Fairburn and Beglin, 1994). Collectively, these results suggest a concerning and persistently high prevalence of disordered eating among female endurance athletes.

Another potential origin to LEA could be food intolerances, since the risk of energy deficiency increases when food groups are removed from the diet, if proper replacements are not made (Lis et al., 2019). However, in contrast to our hypothesis, we did not find any differences of the frequency of reported food intolerances when comparing athletes at risk of LEA with athletes at low risk. This may indicate that athletes who report food intolerances are already familiar with finding dietary alternatives to compensate for potential deficiencies. We did, however, find a higher LEAF-Q total score among athletes reporting food intolerances, which was due to a higher gastrointestinal score. Since the study is based on self-reported data, we can only speculate whether the gastrointestinal problems are due to food intolerances or due to the gastrointestinal alterations as a result of LEA, where the symptoms may have been misinterpreted by some athletes.

Food intolerances were, however, more frequently reported by athletes with disordered eating behavior compared to athletes with low EDE-Q score, and also among athletes with risk of exercise addiction compared to athletes with low risk of exercise addiction. While there is evidence to suggest that a diagnosed food allergy increases the likelihood of developing a subsequent eating disorder (Jafri et al., 2021), it is also possible that the fear and anxiety regarding food and eating in a pre-existing eating disorder may lead to self-diagnosis to gain social acceptance for food exclusion. Likewise, the association in the present study between exercise addiction and food intolerances can potentially be explained by increased attention to the body and its reactions in athletes at risk of exercise addiction, since exercise addiction may develop as a way of dealing with difficult emotions and low self-esteem where one tries to get better mentally by finding simple physical explanations (Wågan et al., 2021). Therefore, some athletes, specifically those with secondary exercise addiction, may use food elimination as a part of their coping strategy.

This study confirms that female endurance athletes are at high risk of LEA (65%) and disordered eating behavior (21%), and that an association between the two exist. However, athletes with risk of LEA were also more likely to exhibit symptoms of exercise addiction and a high risk of both primary (10.4%) and secondary (13%) exercise addiction was found. In a multivariate analysis, however, symptoms of exercise addiction were not identified as predictor of LEA, nor were food intolerances, whereas lower BMI and symptoms of disordered eating behavior were. More studies are needed to investigate food intolerances and the role of exercise addiction with and without disordered eating behavior and associations with LEA among female endurance athletes.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The study was approved by the Regional Ethics Committee in Norway (31640), Sweden (2019-04809), and Norwegian Centre for Research Data (968634). Because data collection occurred remotely and included no medical procedures, the study was considered exempt from additional approval at the other study sites. Regardless, the study was conducted in full accordance with the Declaration of Helsinki at all sites. All data were stored and analyzed in Services for Sensitive Data. The patients/participants provided their written informed consent to participate in this study.

ILF, AKM, IG, and MKT were responsible for study design and conceptualization. AKM and MG were responsible for the Swedish cohort. ILF, MKT, and IG the Norwegian cohort. DL and SM for the Irish cohort. PW and KK for the German cohort. ILF was responsible for coordination and overall data collection, for data analyzation under the supervision of AS, and for constructing the article where everyone reviewed and gave feedback for improvements. All authors reviewed and approved the final version of the article.

This work was supported by Grants from the University of Agder and the Norwegian Olympic Sports Center.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.