Journal Logo

Clinical Sciences: Clinically Relevant

The Female Athlete Triad: Are Elite Athletes at Increased Risk?

TORSTVEIT, MONICA KLUNGLAND1; SUNDGOT-BORGEN, JORUNN1,2

Author Information
Medicine & Science in Sports & Exercise: February 2005 - Volume 37 - Issue 2 - p 184-193
doi: 10.1249/01.MSS.0000152677.60545.3A
  • Free

Abstract

A serious syndrome comprising three interrelated components—disordered eating, amenorrhea, and osteoporosis—has been termed the female athlete triad (the Triad) (21,33). It has been stated that all female athletes are potentially at risk of developing the Triad (19), but that athletes competing in sports in which leanness and/or a low body weight is considered important may be at increased risk of the Triad (19). However, a position statement (21) claims that this syndrome occurs not only in elite athletes but also in nonathletes and in physically active girls and women who are not training or competing in a specific sport. It is unclear, however, to what extent girls and women engaged in physical activity at different levels are at risk of the Triad.

The prevalence of disordered eating and eating disorders in young female athletes has been reported to be higher in athletes than in nonathletes, and particularly in athletes competing in sports that emphasize leanness or a low body weight (4,5,27,29). Furthermore, it has been reported that amenorrhea is more prevalent in the athletic population (3–66%) than in the general female population (2–5%) (20). Few prevalence studies related to premature osteoporosis in young athletes and nonathletes have been published. Two studies have reported a prevalence of osteoporosis of 10–13% in small groups of amenorrheic distance runners (22,23), whereas two other studies did not find any females with osteoporosis in their samples (15,34).

Each of the three disorders of the Triad alone may result in serious medical health consequences. The appearance of all three disorders of the female athlete triad increases the potential for considerable morbidity and even a higher rate of mortality. Because the Triad is frequently denied, not recognized, and underreported, proper screening for a number of symptoms and risk factors has been recommended (21). Because disordered eating (24) and most likely also menstrual dysfunction and osteoporosis are assumed to occur on a continuum, early detection and identification of females with at-risk behavior associated with these three components may actually prevent further development and worsening of Triad symptoms. Therefore, identification of at-risk factors is essential in the evaluation of the Triad. Furthermore, little is known about the prevalence of symptoms and risk factors of the Triad in elite athletes and nonathlete controls. Therefore, the purpose of this study was to examine the percentage of elite athletes and nonathlete controls at risk of the Triad.

METHODS

Participants

The total population of female elite athletes in Norway, 13–39 yr of age (N = 938) and nonathlete controls in the same age group (N = 900) were invited to participate in the study. Permission to undertake the study was provided by the Norwegian Olympic Committee and the Norwegian Confederation of Sports, the Data Inspectorate, and the Regional Committee for Medical Research Ethics. The secretary general of each sport federation and the head of the health care team for each of the national teams received detailed written information about the procedures and aims of the study. In addition, all secretaries general were asked to return a list containing the names, ages, and addresses of all eligible athletes in their federation competing for national teams. Lists were obtained from all secretaries general. All participants received an information letter and had to complete a written consent form. Parents of responders younger than 18 yr old had the opportunity to refuse participation on behalf of their child, whereas written parental consent was required for responders younger than 16 yr.

In this study, an elite athlete was defined as one who qualified for the national team at the junior or senior level or who was a member of a recruiting squad for that team. The athletes had to be 13–39 yr old. Exclusion criteria included pregnancy with a subsequent plan to discontinue the athletic career after delivery, severe injury that had prevented the athlete from training for more than 3 months, or participation in two different sport groups.

The athletes represented 66 different sports/events. For parts of the analysis, these sports/events were divided into different sport groups. The classification of sports/events has been developed and used by one of the authors (26). However, in this study, some additional sports/events were included, and it was therefore necessary to somewhat expand and revise the categories. Seven different sport groups were formed: technical (G1), endurance (G2), aesthetic (G3), weight-class (G4), ball game (G5), power (G6), and antigravitation (G7) sports. This revised classification system is used by the Norwegian Olympic Training Centre (Table 1). Furthermore, for parts of the analysis and in accordance with previous research (27), the seven different sport groups were divided into two groups: leanness sports and nonleanness sports. Athletes competing in sports in which leanness and/or a specific weight were considered important (G2, G3, G4, and G7) were included in the leanness group. Athletes competing in sports in which these factors were considered less important (G1, G5, and G6) were included in the nonleanness group.

TABLE 1
TABLE 1:
Classification of 66 different sports/events divided into sport groups (N = 669).

A randomly selected sample of controls (N = 900) representative of the female Norwegian population in the same age group as the athletes was also included in the study. A Norwegian bureau of statistics (Ergo group), which keeps records of all citizens of Norway, was responsible for the random control sample selection. Every county in Norway was represented, and the sample’s age distribution and geographical distribution approximated that of the total Norwegian women’s population 13–39 yr of age. The controls were not matched to the athletes because in this study we wanted to compare the total population of elite athletes with a sample of women from the general population in the same age range as the athletes. Exclusion criteria included severe illness, unfamiliarity with the Norwegian language leading to problems answering the questionnaire, and competition in sports for a junior or senior national team or its recruiting squad. Based on the latter criteria, our controls were classified as nonathlete controls.

Assessment Procedures

Questionnaire.

A questionnaire including a battery of assessment questions was sent to each of the 938 eligible athletes and 900 eligible controls. Thirty-one athletes did not receive the questionnaire because they were representing teams in other countries or were traveling. A total of 149 athletes were excluded from the study: 76 had ended their career, 35 did not compete at the national level, 15 were injured, 8 were pregnant and did not plan to continue their athletic careers after delivery, 5 were older than 39 yr of age, 9 did not complete the questionnaire satisfactorily, and 1 athlete competed in two different sport groups. In addition, 89 athletes of the remaining 758 (11.7%) did not respond for unknown reasons. The response rate of the total sample was 88.3%.

Twenty-three of the 900 eligible controls did not receive the questionnaire because of problems finding their addresses. In addition, 12 were excluded; nine did not understand the Norwegian language and three were severely ill and were unable to fill in the questionnaire. Therefore, 865 controls were available for participation in the study. A total of 258 of the initial sample (29.8%) did not respond for unknown reasons, leaving 597 available for analysis (70.2%).

In addition to questions regarding menstrual history, oral contraceptive use, and pregnancy, the Body Dissatisfaction (BD) and Drive for Thinness (DT) subscales of the Eating Disorder Inventory (EDI) (10) and questions regarding weight history, training history, training time and physical activity patterns, dietary history, nutritional habits, use of pathogenic weight-control methods, possible eating disorders, and injury history were included on the questionnaire.

Total training was defined as the total number of hours of training per week for the athletes (presented as a mean of the training and competition period during the previous year). Amount of physical activity among the controls was defined as the total number of hours of physical activity per week including physical education lessons, recreational sports, and active daily living like walking. This value was calculated based on questions about type of physical activity, frequency, and duration during the previous year. For parts of the analysis, the athletes and controls were divided into quartiles based on their physical activity levels.

In this study, primary amenorrhea was defined as the absence of menarche by the age of 16. Secondary amenorrhea was defined as the absence of three or more consecutive menstrual cycles after menarche and outside pregnancy, and oligomenorrhea was defined as menstrual cycles of 35 d or more. A short menstrual cycle may reflect a short luteal phase, and luteal inadequacy may be the first stage in the development of amenorrhea (8). In this study, short luteal phase was defined as a menstrual cycle of less than 22 d. Primary amenorrhea, secondary amenorrhea, oligomenorrhea, and short luteal phase were all defined as menstrual dysfunction (MD). If current primary amenorrhea, secondary amenorrhea, oligomenorrhea, or short luteal phase or a history of primary amenorrhea and secondary amenorrhea was reported, the subject was diagnosed with MD.

The purpose of the present study was to look at risk factors for the Triad components and not the prevalence of the end points of the Triad per se. Therefore, no blood samples, measurements of bone mineral density (BMD), or clinical interviews were conducted in the first part of this study.

Selection criteria.

The at-risk criteria used in this study (Table 2) were chosen based on the assumption that disordered eating (24) and the other components of the Triad occur on a continuum. Therefore, we found it important to include not only the end points of the Triad continuum—clinical eating disorders, amenorrhea, and osteoporosis—but to go beyond these three disorders and evaluate disordered eating, signs of menstrual dysfunction, and stress fractures. Furthermore, pressure placed on females to achieve or maintain an unrealistically low body weight is considered the main reason for developing the Triad, and it has therefore been claimed that women who have significant weight, eating, and body image concerns may be at risk of the Triad (21). High scores on two of the eight subscales included on the EDI, DT, and BD, as well as use of pathogenic weight-control methods, are also symptoms of disordered eating and/or eating disorders and predict the development of eating disorders (27,29). The EDI has been found suitable for use as a screening instrument for eating disorders in a nonclinical setting (10), and the two subscales EDI–DT and EDI–BD have been shown to be the only measures that predicted development of eating disorders (11). Furthermore, high EDI–DT and EDI–BD scores have also been used as selection criteria when investigating the prevalence of eating disorders in elite athletes (27). Therefore, participants in this study were classified as at risk of developing eating disorders, and thus the Triad, if their score on the EDI–DT or EDI–BD subscale was at or above the mean score for known anorectics (10). The EDI–BD and EDI–DT subscales have previously been evaluated as reliable using Cronbach’s alpha values (range 0.74–0.90) (28).

TABLE 2
TABLE 2:
Criteria for being classified as at risk of the female athlete triad: the subject had to meet to one or more of these criteria.

It has been claimed that a negative energy balance may be a key component in the development of the Triad (18). Unfortunately, measuring energy balance in both athletes and controls in this study was not possible. We chose to estimate energy balance indirectly by assuming that a very low body mass index (BMI) (32) may indicate energy deficiency. Therefore, a BMI of <18.5 was identified as an at-risk criterion for the Triad.

Irregular or absent menstruation is clearly associated with disordered eating and low bone mass (6,27). In fact, it has been reported that missing even single menstrual cycles at intervals of several months may lead to reduced bone mass (7). Therefore, the presence of any of the menstrual disorders (amenorrhea, oligomenorrhea, and short luteal phase) was included as an at-risk criterion for the development of the Triad. Furthermore, because stress fractures have been associated with disordered eating and amenorrhea (2), self-reported stress fracture was also included as an at-risk criterion for the Triad.

For parts of the analysis, the participants were divided into three different age groups: 13–19, 20–29, and 30–39 yr of age.

Statistical Analysis

All analyses were performed using SPSS software, version 11.0 (SPSS, Evanston, IL). Results are expressed as mean and SD. Comparisons between athletes and controls, between leanness and nonleanness sports, between the different age groups, and between the physical activity quartiles were carried out using a two-sample Student’s t-test for continuous data and a chi-square test for categorical data. Fisher’s exact test was carried out when the cells had expected counts of <5. Differences were considered statistically significant for P values ≤5%. Comparisons between the sport groups were carried out using nonparametric tests (Kruskal–Wallis and Mann–Whitney) for continuous data and chi-square test for categorical data. All tests were two tailed. To prevent Type I error, the Bonferroni method of adjustment was used when describing differences between the sport groups. For these multiple comparisons, the significance level was adjusted by dividing the conventional 0.05 level with the number of t-tests (N = 21) per variable. Therefore, the actual significance level for each of these multiple comparisons was ≤0.002. Only significant differences at or below this level are shown in the results. Binary logistic regression analysis was carried out to adjust for differences in age between athletes and controls.

RESULTS

Characteristics of the Participants

The athletes were younger than the controls and reported a lower BMI (P < 0.001). Athletes competing in aesthetic sports were younger with lower height and weight compared with all other sport groups (P < 0.002) (Table 3).

TABLE 3
TABLE 3:
Anthropometric data presented for each sport group and controls.

No age differences between athletes competing in leanness sports and nonleanness sports were found (21.1 ± 6.2 yr and 21.4 ± 4.8 yr, respectively). Athletes competing in leanness sports had a lower weight (57.2 ± 8.1 kg) and BMI (20.5 ± 2.4 kg·m−2) compared with athletes competing in nonleanness sports (64.3 ± 7.9 kg and 22.1 ± 2.2 kg·m−2) and controls (P < 0.001). Furthermore, athletes competing in nonleanness sports had lower weight and BMI compared with controls (P < 0.001).

A total of 18% of the athletes had been ranked among the three best; 12% had been ranked from places 4 to 10 and 12% from places 11 or lower in Olympic games, world championships, or world cups. The remaining athletes had represented the national team in other international and/or national competitions at the junior or senior level or as recruits. Eighty percent of all the athletes had represented the national team in international competitions for more than one season. The athletes trained an average of 13.2 (±5.2) h·wk−1. Athletes competing in aesthetic sports reported the highest training volume (16.7 ± 5.7 h), followed by athletes competing in power (14.7 ± 4.4 h), technical (14.5 ± 6.2 h), weight-class (14.0 ± 4.4 h), endurance (13.0 ± 4.4 h), antigravitation (12.9 ± 4.1 h), and ball game (11.8 ± 4.7 h) sports. The controls reported to be physically active 5.3 ± 5.3 h·wk−1.

At Risk of the Triad

Athletes versus controls.

A higher percentage of controls (69.2%) than athletes (60.4%) were classified as at risk of the Triad (P < 0.01). The difference was still significant after adjustment for age (P < 0.01). Age group comparisons within the athletic group and within the control group revealed no statistically significant differences in at-risk percentages. However, when comparing athletes with controls in the different age groups, a higher percentage of controls (73.9%) 20–29 yr of age was at risk of the Triad compared with athletes (59.3%) in the same age group (P < 0.001). There were no significant differences in the other age groups.

When dividing physical activity level into quartiles in the control group, no differences in percentage at risk of the Triad were seen between the four activity groups. Furthermore, when the control group was divided into two groups based on their physical activity level (highest quartile compared with the three lower quartiles), no significant differences in percentage at risk of the Triad were found. The athletes were also divided into quartiles based on total training volume. No significant differences in percentage at risk of the Triad were found between the quartiles or between the highest quartile and the three lower quartiles.

A higher percentage of controls reported use of pathogenic weight-control methods (36.7%) and a high EDI–BD score (27.6%) compared with athletes (20.2% and 14.7%, respectively) (P < 0.001), whereas more athletes reported low BMI (7.6%), menstrual dysfunction (31.4%), and stress fractures (17.2%) compared with controls (4.8%, 24.5%, and 12.2%, respectively) (P < 0.05). When adjusting these results for age, no difference in prevalence of being underweight between the athletes and controls was found. The other differences remained the same. No age-adjusted differences between athletes and controls were detected for high EDI–DT scores (4.7% and 6.5%, respectively), or self-reported eating disorders (18.4% and 21.1%, respectively).

A higher percentage of controls (70.8%) than athletes (51.3%) reported that they had tried to lose weight at least once (P < 0.001). Of these, 11.8% of the controls and 13.7% of the athletes tried continuously to lose weight (not significant), and a higher percentage of athletes (35.8%) than controls (28.4%) were currently dieting (P < 0.05).

Sport groups.

A higher percentage of athletes competing in leanness sports (70.1%) and controls (69.2%) were classified as at risk of the Triad compared with athletes competing in nonleanness sports (55.3%) (P < 0.001). A higher percentage of athletes competing in leanness sports were underweight (BMI <18.5) and reported menstrual dysfunction compared with both athletes competing in nonleanness sports (P < 0.001) and controls (P < 0.001). More controls than athletes competing in both leanness and nonleanness sports reported use of pathogenic weight-control methods (P < 0.001) and high EDI–BD scores (P < 0.001) (Table 4).

TABLE 4
TABLE 4:
Number and percentage of athletes divided into leanness and nonleanness sports and controls fulfilling one or more of the different at-risk criteria for the female athelete triad.

Significant differences in at-risk percentage were found when comparing the different sport groups (overall, P < 0.01). A higher percentage of athletes competing in aesthetic sports were classified as at risk of the Triad compared with athletes competing in ball game sports (Fig. 1).

FIGURE 1— Percentage with 95% confidence intervals of athletes at risk of the Triad (
FIGURE 1— Percentage with 95% confidence intervals of athletes at risk of the Triad (:
N= 404) divided into sport groups. *P< 0.001 compared with aesthetic sports.

A higher percentage of athletes competing in aesthetic sports were underweight compared with athletes competing in technical, ball game, and power sports (P < 0.001). The highest percentage of athletes using one or more pathogenic weight-control methods was found among those competing in weight-class sports (37.1%) and technical sports (30.0%). A higher percentage of athletes competing in technical sports than athletes competing in endurance sports reported a high score on the EDI–BD subtest (Table 5).

TABLE 5
TABLE 5:
Number and percentage of athletes (N = 669) in the different sport groups fulfilling one or more of the different at-risk criteria for the female athlete triad.

DISCUSSION

To our knowledge, this study is the first to publish data from the total population of elite athletes and age-representative controls from the general population of a country regarding the prevalence of at-risk subjects for the Triad. The main findings in this study are as follows: 1) more than 6 of 10 females were classified as at risk of the female athlete triad, with small differences detected between normal active females and elite athletes, and 2) higher percentages of both athletes competing in leanness sports and nonathlete controls were classified as at risk of the Triad compared with athletes competing in nonleanness sports.

Selection of At-Risk Criteria

It is surprising that as many as 64% of the women included in this study were classified as at risk of the Triad. However, it is important to remember that previous studies focusing on at-risk subjects (13,25) have mainly investigated undernutrition, weight loss, and risk of eating disorders, and not asked for menstrual dysfunction or stress fractures with the intention to investigate at-risk subjects for all three components of the Triad at the same time. Because disordered eating (24) and the other two components of the Triad presumably occur on a continuum, it was considered important to focus not only on the presence of symptoms of severe eating disorders, amenorrhea, and osteoporosis, but to go one step further and evaluate early warning signs, including disordered eating, signs of menstrual dysfunction, and stress fractures. Thus, a high percentage of females fulfilling the presented at-risk criteria in our study could be due to the high number and variety of criteria included in the at-risk selection process.

The subjects in the present study were considered at risk of the Triad if they met one or more of the criteria listed, including whether they answered “I don’t know” to a self-reported eating disorder and/or stress fracture. We did consider the possibility that these criteria were too lenient and could lead to a high number of false-positive subjects. Thus, we created two additional risk criteria models: one similar to the one presented in Table 2, but excluding those subjects answering “I don’t know,” and one excluding those subjects answering “I don’t know” in addition to requirements of a positive answer to at least two of the risk factors. As expected, the percentage of at-risk subjects declined using the stricter criteria; however, the statistically significant differences between the athletes and the controls remained the same as when using the criteria presented in Table 2 (data not shown). Because our main project also included a clinical part in which we were able to diagnose subjects with eating disorders and low BMD (however, not a part of this report), we were able to do a further analysis of those athletes and controls who had answered “I don’t know” to the question regarding an eating disorder or a stress fracture. By doing this, we were able to investigate whether those who only answered “I don’t know” actually had a subclinical or clinical eating disorder or low BMD (z-score <−2.0) (12,16). Of those five athletes who had answered “I don’t know” to the question regarding an eating disorder, four were diagnosed with a present subclinical or clinical eating disorder. A total of four controls had answered “I don’t know” to the question regarding an eating disorder; three of these were diagnosed with a present subclinical or clinical eating disorder. If we included past in addition to present subclinical or clinical eating disorders, all those athletes and controls answering “I don’t know” on the questionnaire were diagnosed. Based on this, we recommend including those who answered that they were uncertain as to whether they have or have had an ED when selecting at-risk subjects for the Triad. As a consequence, the use of “I don’t know” in addition to “yes” actually reduces the number of false-negative subjects, which is very important considering the serious consequences of the Triad and the importance of early detection. Of those 16 athletes and 13 controls who had answered “I don’t know” to the question regarding a stress fracture, two athletes and two controls were diagnosed with low BMD. Thus, including those who answered that they were uncertain as to whether they have or have had a stress fracture when selecting at-risk subjects for the Triad may lead to a high number of false-positive subjects, and should be considered based on this. In addition, by using the at-risk criteria presented in Table 2, the percentage classified as at risk is only slightly higher (3.2% and 3.3% for the athletes and controls, respectively) than when at-risk criteria excluding those answering “I don’t know” (data not shown) are used. In addition, according to the position stand on the Triad (21), physically active girls and women should be referred for medical evaluation at the first sign of any of the components of the Triad. As shown in the Methods section of this study, the presented at-risk criteria are all signs or symptoms of one or more of the components of the Triad. For instance, disordered eating encompasses a spectrum of abnormal eating patterns, including behaviors such as binging; purging; prolonged fasting; use of diet pills, diuretics, and laxatives; and thought patterns such as preoccupation with food, dissatisfaction with one’s body, fear of becoming fat, and a distorted body image (19). Considering the serious consequences of these at-risk Triad behaviors and the reported disorders, it is our opinion that it is important to evaluate athletes from a variety of perspectives. This would also reduce the risk of false-negative cases. Therefore, we think it is important that a number of relevant at-risk criteria, like those presented in this article, are included when identifying athletes with the Triad.

The high number of women classified as at risk of the Triad in this study may actually reflect the present situation in young females today (4,21). Western society’s focus on the female body—on thinness and low weight—may well lead to dissatisfaction with one’s own body, further leading to dieting and unhealthy eating behaviors and subsequent loss of menstruation and bone mass. This suggestion is confirmed by the findings in the present study showing that 51% of the athletes and 71% of the controls had tried to lose weight, 15% of the athletes and 28% of the controls had high scores on the EDI–BD scale, and as many as 20% of the athletes and 37% of the controls had used one or more pathogenic weight-control methods to reduce their weight. It is possible that an overreporting of symptoms of the Triad may have occurred as a consequence of the fact that guidance and information about nutrition, weight-control programs, and exercise (results from this part of the study are not included in this article) were offered for free. However, the question of whether the high percentage of both athletes and controls meeting the at-risk criteria in this study is due to the selection criteria or to the fact that such a high percentage of young women actually are at risk of or have developed one or more of the interrelated disorders of the Triad, can only be answered after a clinical evaluation of the subjects classified as at risk.

Athletes versus Controls

To date, the prevalence of the Triad in elite athletes has not been systematically investigated. One study was found that investigated the simultaneous occurrence of disordered eating, amenorrhea, and osteoporosis in military women (15). However, no control group was included in their study. Because our study investigated at-risk criteria for the Triad in elite athletes and nonathlete controls, a direct comparison between these two studies is difficult. Based on a number of studies concluding that eating disorders and menstrual dysfunction are more common in athletes than in controls (5,20,26,27,29), it would seem appropriate to assume that a higher percentage of athletes than controls are at risk of the Triad. In contrast to what was expected, the results of our study show that a significantly higher percentage of nonathlete controls than of athletes was classified as at risk of the Triad when including all the at-risk factors. It should be noted, however, that another pattern appears when viewing the at-risk criteria individually. A higher percentage of controls than athletes reported disordered eating including high scores on the EDI–BD subtest and use of pathogenic weight-control methods, whereas no differences were shown between these two groups with regard to EDI–DT score and self-reported eating disorders. Conversely, more athletes than controls reported menstrual dysfunction and stress fractures. It appears, then, that there is little difference between the nonathlete controls and the athletes with respect to disordered eating and eating disorders, whereas clear medical signs like a history of menstrual dysfunction and stress fractures is more common in the athlete population.

In a study investigating weight and diet concerns in 173 Finnish female athletes and 79 controls, the total EDI–DT and EDI–BD scores were higher in female controls than in endurance athletes (9). Furthermore, as in our study, Sundgot-Borgen and Larsen (25) found a higher percentage of controls compared with athletes with high scores on the EDI–BD subscale. This finding could partly be explained by the fact that in spite of the body weight pressure experienced in some sports, most of these athletes have bodies that are well formed and fit looking, which is the trend today. Furthermore, athletes often have more knowledge than controls about healthy nutrition; they can get guidance on how to keep their weight at healthy levels and may therefore use more healthy methods, such as exercise, to control weight instead of more risky methods like laxatives, diuretics, and vomiting. In contrast to previous findings (25), the present study seems to confirm that athletes are, in fact, using healthier weight-control strategies than controls, as shown by the higher percentage of controls than athletes reporting use of pathogenic weight-control methods.

Beals and Manore (1) recently published data on the prevalence of and relationship between the disorders of the Triad in 425 collegiate athletes. Their data indicated that even if few reported a clinical diagnosis of anorexia nervosa (3.3%) or bulimia nervosa (2.3%), a high number was at risk of an eating disorder using the EDI–BD subscale (32.4%). A lower percentage of athletes (14.3%) was classified as at risk of the Triad in our study using the EDI–BD subscale. However, it is difficult to compare the results of the two studies because the cutoff point in our study was higher (EDI–BD score of 14) than that in the study by Beals and Manore (EDI–BD score of 12). Johnson et al. (13) also found a high number of collegiate athletes at risk of anorexia nervosa (25%) and bulimia nervosa (38%), although they used somewhat different criteria. A total of 18% of the athletes in the present study reported present or a history of eating disorders. This is a much higher percentage than that found among the collegiate athletes in other studies (1,13,14), but is in accordance with the prevalence of clinically diagnosed eating disorders in elite athletes in a comparable Norwegian study a decade ago (27).

Even though a few other studies have investigated at-risk behavior for eating disorders among athletes (1,9,13–15,25,30), it is difficult to compare their results with ours because no control groups and/or athletes competing at different levels and in different sports have been included. One study based on data collected in 1997 is, however, comparable in terms of subjects investigated and methods used (29). The authors concluded that disordered eating behavior is more prevalent in athletes than in nonathlete controls (29), thus somewhat in contrast to what was found in the present study. However, in the 1997 study, the at-risk criteria were not exactly the same as those used in the present study investigating at-risk criteria for the Triad and not only risk of eating disorders. In the present study, the greatest difference between the athletes and the controls was found in the reported use of pathogenic weight-control methods, a criterion that was not included in the study from 1997 (29). Furthermore, disordered eating behavior and the prevalence of eating disorders may have changed in the past 5 yr, which may possibly help explain the discrepancy in the findings of these two studies.

In accordance with previous studies (20,26), a higher percentage of athletes than controls reported the occurrence of menstrual dysfunction in this study. However, when dividing the athletes into leanness and nonleanness groups, athletes competing in nonleanness sports reported the same amount of menstrual dysfunction as controls. However, Otis (20) claims that the estimated prevalence of secondary amenorrhea per se in athletes ranges from 3 to 66%, compared with 2–5% in the general population. The different prevalence reported in the different studies may be explained by a number of methodological factors, such as different definitions. It is therefore difficult to compare the prevalence of menstrual dysfunction in our study with that in other studies. The etiology of menstrual dysfunction is multifactorial, and low body weight, low energy availability, exercise, and psychological stress have been investigated as potential causes of disruption of the normal endocrine process (17,31). The athletes in our study had a lower mean body weight compared with the controls, which may reflect energy deficiency in athletes and could partly explain why menstrual dysfunction is more common among athletes than controls in this study. On the contrary, adjusted for age, we found no difference between these two groups in the percentage of underweight subjects (BMI <18.5). The low energy availability theory is, however, supported by the fact that only athletes competing in sports in which leanness and/or a specific weight are considered important experienced more menstrual dysfunction than controls. Furthermore, the amount of physiological stress (training time) and possible subsequent psychological stress may partly explain the distinction in prevalence of menstrual dysfunction.

There seems to be a lack of studies investigating the various forms of menstrual dysfunction as at-risk criteria for the Triad. However, Beals and Manore (1) found that athletes reporting menstrual dysfunction were more likely to score above the mean both on the Eating Attitude Test-26 (EAT-26) and EDI–BD compared with those reporting normal menses. Although Beals and Manore did not find a higher prevalence of musculoskeletal injuries among athletes reporting menstrual dysfunction, a recent study investigating female runners concluded that an elevated score on the EDI was associated with oligo/amenorrhea and that runners with oligo/amenorrhea had lower BMD than runners with eumenorrhea runners (6).

To our knowledge, no studies have investigated the prevalence or incidence of reported stress fractures among the total population of elite athletes and representative nonathlete controls from one country. However, a number of studies have assessed stress fracture incidence rates in track and field athletes and competitive runners, as well as in other athletes competing at different competition levels (3). The stress fracture incidence rates from studies investigating different sports have been between 3 and 7% (3), and thus lower than the prevalence obtained in this study. However, higher incidence rates (11–27%) have been found in prospective studies assessing competitive track and field athletes (3). Still, it is difficult to compare incidence rates of diagnosed stress fractures with the prevalence of self-reported stress fractures. In our study, 17% of the athletes and 12% of the controls reported a stress fracture, whereas 34% of college athletes reported experiencing bone injury in the study by Beals and Manore (1). Because a bone injury may include more than a stress fracture, the difference between the two studies is understandable.

It is possible that because some of the subjects may not know the difference between a stress fracture and a normal fracture, the indicated stress fracture prevalence may therefore encompass normal fractures that the subjects may have experienced in their life, in addition to stress fractures resulting from inadequate energy intake, loss of menstruation, or excessive exercise. This may be especially true for the control group because the athletes are more likely to be aware of the difference between a stress fracture and a normal fracture. The number of athletes and controls in our study who actually have sustained a stress fracture is not known, but the self-reported prevalence indicates that a high number is at risk of the Triad.

Sport Groups

The results from this study are in accordance with other studies reporting a high prevalence of disordered eating and menstrual dysfunction in sports focusing on leanness and/or a low body weight (4,26,27,29). When dividing the athletes into two groups in our study, we found that a higher percentage of athletes competing in leanness sports were classified as at risk of the Triad compared with athletes competing in nonleanness sports. Further, when we investigated the number of athletes at risk of the Triad in the seven different sport groups, we found an overall difference. No other studies have published data on at-risk criteria for the Triad in so many different sport groups, and it is therefore difficult to compare these results with those of other studies. However, Beals and Manore (1) recently published data on the prevalence of and the relationship between the disorders of the Triad in 425 collegiate athletes representing three different sport groups. No significant difference between aesthetic sports, endurance sports, and team/anaerobic sports with respect to self-reported eating disorders, EDI–BD score, irregular menstrual cycles, or few menstrual cycles during the year in question was found in their study (1). On the other hand, they reported that athletes in aesthetic sports scored higher on the EAT-26 and reported using very low calorie diets, fasting, vomiting, and laxatives for weight loss compared with athletes in the other two sport groups. In addition to athletes competing in aesthetic sports, we found that endurance and weight-class athletes were also considered to be particularly at risk of the Triad. Most sports included in these sport groups focus on a low weight and leanness. Fogelholm and Hiilloskorpi (9) also reported a high percentage of individuals classified as at risk of eating disorders among weight-class athletes (13%) followed by controls (11%) and aesthetic athletes (7%). It is important to remember, however, that different at-risk criteria have been used by each study.

If we look into the different at-risk criteria examined in the present study, there are disparities between the sport groups. The prevalence of underweight is fairly high in athletes competing in aesthetic, endurance, and antigravitation sports, in agreement with the study by Sundgot-Borgen and Larsen (26). Almost 3 of 10 of the athletes in the aesthetic sport group were underweight, a high percentage compared with most other sport groups. However, it should be noted that these athletes were also younger than athletes in the other sport groups and 16 of the 18 underweight athletes were younger than 16 yr of age. A high percentage of athletes competing in weight-class sports (37%) and technical sports (30%) report use of pathogenic weight-control methods. Our findings on weight-class athletes are in agreement with those other studies (9,25); however, few studies have been carried out focusing on athletes competing in technical sports. These athletes often compete in sports in which the energy expenditure is low (i.e., bowling, curling, shooting) and may therefore choose other, more pathogenic methods to lose weight. In addition, athletes competing in technical sports in this study had a high BMI compared with athletes competing in endurance sports, aesthetic sports, and antigravitation sports, and a high percentage also reported high EDI–BD scores compared with the other sport groups. These results suggest that body dissatisfaction may lead to dieting. The only comparable study that we found was a study carried out by Sundgot-Borgen (27) in 1993. She found that the highest use of pathogenic weight-control methods was among those competing in aesthetic sports (34%), weight-dependent sports (32%), and endurance sports (20%) and not in technical sports, as shown in the present study. The differences between the two studies may represent differences in the studies themselves, or may actually indicate a change in behavior patterns in the sport groups even in the past decade. In particular, focus on prevention of the Triad in recent years has been primarily geared toward the at-risk sports (i.e., aesthetic sports and endurance sports) and not technical sports, which were not considered at-risk sports based on investigations a decade ago (26). Such prevention efforts may possibly have affected the choice of weight-control method for athletes in at-risk sports. However, the differences between the two studies may be due to the fact that somewhat different sports were included in the technical sport group in the present study. The athletes in the present study also had higher BMI and possibly less knowledge about how to properly lose weight, resulting in a higher use of pathogenic weight-control methods.

A high percentage of athletes competing in endurance, aesthetic, and weigh-class sports reported menstrual dysfunction. This is in agreement with other studies (9,26), and might reflect the consequences of competing in sports in which leanness and low weight are considered important (26).

Data on the prevalence of stress fractures among athletes divided into several different sport groups are lacking. In this study, athletes competing in power sports reported a particularly high prevalence of stress fractures (35%). The reason for this is uncertain, but it may reflect the extent of physical training or extrinsic, intrinsic, physiological, hormonal, or nutritional factors involved (2). It should be noted that most track and field athletes are included in the power sports group and that these athletes tend to have high incidence rates of stress fractures (3).

Based on the high prevalence of exercising girls and women at risk of the Triad, it is necessary to continue the ongoing education of the athletes themselves, coaches, leaders, and parents, as well as physically active girls and women in general about the life-threatening health consequences of inadequate energy intake and excessive exercise habits. In turn, it may be possible to prevent more females from developing one or more of the components of the Triad. Given the relatively high percentage of subjects in this study classified as at risk of the Triad and the potentially severe physical and psychological consequences associated with the Triad, the questionnaire-based results of this study should be further examined by clinical interview and clinical examination.

The results of this study should be considered generalizable to other athletes competing at the same performance level as the athletes included in our study, competitors at Olympic and/or world cup levels or participants in international or national competitions at a junior or senior national team level or as recruits for such a team. Furthermore, in the present study, all European sports are included, and most of these are commonly known also in other parts of the world. We therefore assume that our results concerning the athletic population are generalizable to national level athletes and that the results concerning the nonathlete control group are generalizable to nonelite females in general in many other cultures and countries.

REFERENCES

1. Beals, K. A., and M. M. Manore. Disorders of the female athlete triad among collegiate athletes. Int. J. Sport Nutr. Exerc. Metab. 12:281–293, 2002.
2. Bennell, K., G. Matheson, W. Meeuwisse, and P. Brukner. Risk factors for stress fractures. Sports Med. 28:91–122, 1999.
3. Bennell, K. L., and P. D. Brukner. Epidemiology and site specificity of stress fractures. Clin. Sports Med. 16:179–196, 1997.
4. Byrne, S., and N. McLean. Eating disorders in athletes: a review of the literature. J. Sci. Med. Sport. 4:145–159, 2001.
5. Byrne, S., and N. McLean. Elite athletes: effects of the pressure to be thin. J. Sci. Med. Sport. 5:80–94, 2002.
6. Cobb, K. L., L. K. Bachrach, and G. Greendale, et al. Disordered eating, menstrual irregularity, and bone mineral density in female runners. Med. Sci. Sports Exerc. 35:711–719, 2003.
7. Cumming, D. C., and C. E. Cumming. Estrogen replacement therapy and female athletes: current issues. Sports Med. 31:1025–1031, 2001.
8. De Souza M. J., B. E. Miller, A. B. Loucks, et al. High frequency of luteal phase deficiency and anovulation in recreational women runners: blunted elevation in follicle-stimulating hormone observed during luteal-follicular transition. J. Clin. Endocrinol. Metab. 83:4220–4232, 1998.
9. Fogelholm, M., and H. Hiilloskorpi. Weight and diet concerns in Finnish female and male athletes. Med. Sci. Sports Exerc. 31:229–235, 1999.
10. Garner, D. M., M. P. Olmsted, and J. Polivy. Manual of Eating Disorder Inventory. Odessa, FL: Psychological Assessment Resources, 1984, pp. 1–19.
11. Garner, D. M., P. E. Garfinkel, W. Rockert, and M. P. Olmsted. A prospective study of eating disturbances in the ballet. Psychother. Psychosom. 48:170–175, 1987.
12. International Society for Clinical Densiometry Position Development Conference. The diagnosis of osteoporosis in men, premenopausal women, and children. J. Clin. Densitom. 7:17–26, 2004.
13. Johnson, C., P. S. Powers, and R. Dick. Athletes and eating disorders: the National Collegiate Athletic Association study. Int. J. Eat. Disord. 26:179–188, 1999.
14. Kurtzman, F. D., J. Yager, J. Landsverk, E. Wiesmeier, and D. C. Bodurka. Eating disorders among selected female student populations at UCLA. J. Am. Diet. Assoc. 89:45–53, 1989.
15. Lauder, T. D., M. V. Williams, C. S. Campbell, G. Davis, R. Sherman, and E. Pulos. The female athlete triad: prevalence in military women. Mil. Med. 164:630–635, 1999.
16. Leib, E. S., E. M Lewiecki, N. Binkley, and R. C. Hamdy. Official Positions of the International Society for Clinical Densitometry. J. Clin. Densitom. 7:1–6, 2004.
17. Loucks, A. B., M. Verdun, and E. M. Heath. Low energy availability, not stress of exercise, alters LH pulsatility in exercising women. J. Appl. Physiol. 84:37–46, 1998.
18. Manore, M. M. Dietary rec ommendations and athletic menstrual dysfunction. Sports Med. 32:887–901, 2002.
19. Nattiv, A., R. Agostini, B. Drinkwater, and K. K. Yeager. The female athlete triad. The inter-relatedness of disordered eating, amenorrhea, and osteoporosis. Clin. Sports Med. 13:405–418, 1994.
20. Otis, C. L. Exercise-associated amenorrhea. Clin. Sports Med. 11:351–362, 1992.
21. Otis, C. L., B. Drinkwater, M. Johnson, A. Loucks, and J. Wilmore. American College of Sports Medicine position stand. The Female Athlete Triad. Med. Sci. Sports Exerc. 29:i–ix, 1997.
22. Pettersson, U., B. Stalnacke, G. Ahlenius, K. Henriksson-Larsen, and R. Lorentzon. Low bone mass density at multiple skeletal sites, including the appendicular skeleton in amenorrheic runners. Calcif. Tissue Int. 64:117–125, 1999.
23. Rutherford, O. M. Spine and total body bone mineral density in amenorrheic endurance athletes. J. Appl. Physiol. 74:2904–2908, 1993.
24. Shisslak, C. M., M. Crago, and L. S. Estes. The spectrum of eating disturbances. Int. J. Eat. Disord. 18:209–219, 1995.
25. Sundgot-Borgen, J., and S. Larsen. Pathogenic weight-control methods and self-reported eating disorders in female elite athletes and controls. Scand. J. Med. Sci. Sports. 3:150–155, 1993.
26. Sundgot-Borgen, J., and S. Larsen. Preoccupation with weight and menstrual function in female elite athletes. Scand. J. Med. Sci. Sports. 3:156–163, 1993.
27. Sundgot-Borgen, J. Prevalence of eating disorders in elite female athletes. Int. J. Sport Nutr. 3:29–40, 1993.
28. Sundgot-Borgen, J., J. H. Rosenvinge, R. Bahr, and L. S. Schneider. The effect of exercise, cognitive therapy, and nutritional counselling in treating bulimia nervosa. Med. Sci. Sports Exerc. 34:190–195, 2002.
29. Sundgot-Borgen, J., and M. K. Torstveit. Prevalence of eating disorders in elite athletes is higher than in the general population. Clin. J. Sport Med. 14:25–32, 2004.
30. Warren, B. J., A. L. Stanton, and D. L. Blessing. Disordered Eating Patterns in Competitive Female Athletes. Int. J. Eat. Disord. 9:565–569, 1990.
31. Warren, M. P., and N. E. Perlroth. The effects of intense exercise on the female reproductive system. J. Endocrinol. 170:3–11, 2001.
32. World Health Organization. Obesity. Preventing and Managing the Global Epidemic. Report of a WHO Consultation on Obesity. Geneva, 3–5 June. Geneva: World Health Organization, 1998.
33. Yeager, K. K., R. Agostini, A. Nattiv, and B. Drinkwater. The female athlete triad: disordered eating, amenorrhea, osteoporosis. Med. Sci. Sports Exerc. 25:775–777, 1993.
34. Young, N., C. Formica, G. Szmukler, and E. Seeman. Bone density at weight-bearing and nonweight-bearing sites in ballet dancers: the effects of exercise, hypogonadism, and body weight. J. Clin. Endocrinol. Metab. 78:449–454, 1994.
Keywords:

EATING DISORDERS; DISORDERED EATING; AMENORRHEA; MENSTRUAL DYSFUNCTION; LOW BONE MASS; OSTEOPOROSIS

©2005The American College of Sports Medicine