The Female Athlete Triad

Aurelia Nattiv, M.D., FACSM; Anne B. Loucks, Ph.D., FACSM; Melinda M. Manore, Ph.D., R.D., FACSM; Charlotte F. Sanborn, Ph.D., FACSM; Jorunn Sundgot-Borgen, Ph.D.; Michelle P. Warren, M.D.


March 01, 2010

In This Article



In athletes, the prevalence of disordered eating, menstrual disorders, low BMD and stress fractures varies widely.[31,151,161] The prevalence of inadvertent low energy availability without disordered eating or eating disorders is unknown.

Many studies of the prevalence of disordered eating and eating disorders in athletes have yielded unreliable results due to nonstandard diagnostic procedures, small sample sizes, absent or inappropriate control groups, inadequate statistics, and heterogeneity in the type and level of the athletes studied.[30] Only two large, well-controlled studies have diagnosed clinical eating disorders according to the Diagnostic and Statistical Manual of Mental Disorders[8] to obtain unbiased and reliable estimates of the prevalence of eating disorders in elite female athletes in different types of sports.[31,186] One found eating disorders in 31% of elite female athletes in "thin-build" sports compared to 5.5% of the control population.[31] The other found that 25% of female elite athletes in endurance sports, aesthetic sports, and weight-class sports had clinical eating disorders compared to 9% of the general population.[186] A small study of collegiate gymnasts (N = 42) found a prevalence of disordered eating behaviors as high as 62%.[170] A larger study of collegiate gymnasts (N = 218) showed specific weight control behaviors included binge eating (33% ≥ once a week), exercise for the purpose of burning calories (57% ≥ 2 h·d−1), and fasting or strict dieting (28% ≥ 4 times in the past year). However, induced vomiting and the use of diuretics and/or laxatives were rare events.[158]

The prevalence of secondary amenorrhea, long known to vary widely with sport, age, training volume and body weight,[161] has been reported in small studies to be as high as 69% in dancers[1] and 65% in long-distance runners[51] compared to 2-5% in large studies of the general population.[11,159,175] Within distance runners, prevalence of amenorrhea increased from 3% to 60% as training mileage increased from < 13 to >113 km·wk−1 while their body weights decreased from >60 to < 50 kg.[172] Prevalence of secondary amenorrhea is higher (67%) in female runners less than 15 yr of gynecological age compared to older women (9%).[13] The prevalence of primary amenorrhea is less than 1% in the general population[33] and more than 22% in cheerleading, diving and gymnastics.[18] Subclinical menstrual disorders typify both highly trained[121] and recreational[42,53] eumenorrheic athletes: luteal deficiency or anovulation was found in 78% of eumenorrheic recreational runners in at least one menstrual cycle out of three.[42]

Low BMD has been associated with disordered eating even in eumenorrheic athletes.[34] BMD is lower in amenorrheic athletes than in eumenorrheic athletes.[46,127,143,151,162] A systematic review of studies that employed WHO T-scores for diagnosis, found prevalence of osteopenia (T-score between −1.0 and −2.5) ranging from 22% to 50% and prevalence of osteoporosis (T-score ≤ −2.5) ranging from 0% to 13% in female athletes[104] compared to 12% and 2.3% expected in a normal population distribution.

Only three studies of female athletes have investigated the simultaneous occurrence of disordered eating, menstrual disorders and low BMD according to ISCD criteria.[17,152,191] Only one diagnosed eating disorders.[191] The prevalence of the entire Triad in elite athletes from 66 diverse sports (4.3%; 8/186) was similar to controls (3.4%; 5/145),[191] but the athletes' BMD Z-scores were calculated relative to the controls rather than instrument norms. Half the athletes and none of the controls had clinical eating disorders and BMD Z-scores < −2.0. All controls with some, but not all Triad components were overweight rather than underweight and had a history of pathogenic weight loss behavior. The other two studies referenced BMD Z-scores to instrument norms. One found the entire Triad in 2.7% (3/112) of collegiate athletes from seven diverse sports.[17] The other found the entire Triad in 1.2% (2/170) of high school athletes.[152] All these studies defined the Triad more narrowly than this Position Stand: none estimated energy availability, diagnosed subclinical menstrual disorders or the cause of amenorrhea, or assessed changes in BMD.

Future epidemiologic studies should include cases of low energy availability without disordered eating or an eating disorder, luteal deficiency and anovulation, and declining as well as low BMD Z-scores based upon the best available standardized normative database. Where these databases do not yet exist, they should be developed.

Evidence Statement. Disordered eating, eating disorders and amenorrhea occur more frequently in sports that emphasize leanness. Evidence category A.

Risk Factors

Athletes at greatest risk for low energy availability are those who restrict dietary energy intake, who exercise for prolonged periods, who are vegetarian, and who limit the types of food they will eat.[34,125,126] Many factors appear to contribute to disordered eating behaviors and clinical eating disorders.[19,167] Dieting is a common entry point[167] and interest has focused on the contribution of environmental and social factors, psychological predisposition,[34,167] low self-esteem,[160,167] family dysfunction,[137] abuse,[168] biological factors,[98] and genetics.[28,179] Additional factors for athletes include early start of sport-specific training and dieting, injury, and a sudden increase in training volume.[184] Surveys show more negative eating attitude scores in athletic disciplines favoring leanness.[18,27] Disordered eating behaviors are risk factors for eating disorders:[167] Sundgot-Borgen found that 18% of elite female athletes and 5% of controls with disordered eating behaviors were diagnosed with clinical eating disorders.[183]

Surveys of menstrual history have identified potential risk factors associated with amenorrhea, but not hormone levels. Most potential risk factors have been shown not to be causal factors in the disruption of reproductive function in athletes. For example, body weight and fatness are often low in amenorrheic athletes, but eumenorrheic and amenorrheic athletes span a common range of body weight and body fatness leaner than the general population.[161,171] In addition, exercise training has no suppressive effect on luteinizing hormone (LH) pulsatility when energy intake is increased to compensate for exercise energy expenditure.[123]

Risk factors for stress fracture include low BMD, menstrual disturbances, late menarche, dietary insufficiency, genetic predisposition, biomechanical abnormalities, training errors, and bone geometry (e.g., narrower tibia width, shorter tibia length).[14,23,60,144,201]