Many athletes, and especially those in aesthetic (gymnastics, etc.), endurance (running, etc.), and weight-class (wrestling, etc.) events, assume that success is associated with low body weight or fat content. This widespread belief may lead to concerns about body weight and body weight reduction attempts by gradual (diet, exercise) or rapid (sweating, vomiting, etc.) techniques (9). In addition to poor recovery and impaired sports performance, repeated or severe weight loss attempts may have negative health consequences (16,30). The recent concern has been that an increased pressure for body weight loss may lead to eating disturbances or even clinical eating disorders (30,36). Weight reduction and distorted eating have also been connected with hormonal imbalance and menstrual disturbances, with a concomitant reduction of bone density in female athletes (27,35).
Data on the prevalence of eating disturbances in athletes are contradictory. Aesthetic (8,11,17,31), weight-class (31,36) and other female athletes (1,17) have scored higher (i.e., more at risk) than untrained female controls in the Eating Attitude Test (14) or the Eating Disorders Inventory (12) questionnaire. Moreover, a high frequency (25-62%) of aesthetic and weight-class female athletes were found to practice rapid (pathogenic) weight control methods (22,23,33). In contrast, other studies have found aesthetic (1,3,21) and other female athletes, including runners (3,24,37,38), to be at similar or lower risk for eating disorders, compared with female controls.
Very little is known about eating problems in male athletes. Elevated Eating Attitude Test (7) or Eating Disorders Inventory (34) scores have been reported in weight-class athletes. In contrast, much lower results were found in figure skaters (25), swimmers, and runners (7,24).
The motivation for the present study was the arousing worry on negative health consequences of weight preoccupation and weight reduction attempts in athletes (30,35). There was a need for a large study including different kinds of athletic groups. Moreover, data from male athletes were incomplete. Consequently, the aim of the present study was to compare weight reduction techniques, weight and diet concerns, and other factors related to a risk for eating disorders in five groups of female and male athletes and in untrained controls. In addition, factors associated with menstrual status were studied.
Subjects. This study was approved by an independent ethical committee. The athletes were recruited through national sports federations. All athletes in the national training teams, or training teams for young adults (18-20 yr), were invited. In addition, female and male professional dancers and advanced scholars from the Finnish National Ballet were also asked to participate. Female and male, untrained, controls were recruited from a high-school and a technical college in the city of Tampere (180,000 inhabitants).
A total of 363 athletes (173 females and 190 males) participated. The subjects were top or good national standard, but only a few could be considered international-level elite athletes. The athletes were grouped for data analysis as aesthetic, speed, endurance, weight-class, and ballgame athletes (Table 1). Seventy-nine females and 61 males volunteered as untrained controls.
Data collection. All data were collected under the investigators' supervision in training camps (athletes), in the National Ballet, and in classrooms (controls). The subjects were weighed in an unfasted state, in light clothes, and without shoes. The self-reported height was used to calculate body mass index (BMI = weight (kg) divided by height (m) squared).
The subjects filled in a 121-item questionnaire, which included all the questions of the Eating Disorders Inventory (EDI) (12). In addition to the EDI, the following questions were used in the present study: the amount (h·wk−1) of sports training during the previous 4 wk, actual and preferred weight (kg), occurrence and techniques of weight-reduction attempts during the previous 12 months, and whether the subject had received any instructions for weight reduction. In addition, the female participants were queried about the age of menarche, use of contraceptive hormones, reason for the use of contraceptive methods (i.e., prevention of pregnancy or regulation of menstruation), and present menstrual status (number of yearly cycles, timing of the previous period). The subjects put their name on the questionnaire, but the strict confidentiality of the answers was assured both in written form and orally.
Data analysis. The main analyses were done separately for female and male subjects and by using the six-group classification (5 athlete groups and 1 control group) as described above. The occurrence of any weight-reduction attempt, rapid weight-reduction techniques (forced sweating and restricted fluid intake or use of diuretics, laxatives, or vomiting), and of unsupervised weight-reduction attempts was analyzed using a dichotomous (yes/no) classification.
The drive for thinness and body dissatisfaction subscales scores were used as the main outcome variables of the EDI questionnaire. The sum score was used as an index of weight and diet concerns (WDC) (4). An individual was regarded to be at risk for eating disorder, if the body-dissatisfaction score was above 17, or if the drive for thinness score was above 15 (12). Moreover, the total sum score of the 64-item EDI-questionnaire (EDI-64) was calculated.
Those female subjects, who did not use contraceptive methods, were also classified by their menstrual status (5): primary amenorrhea (no menstruation before age 16), secondary amenorrhea (age ≥ 16 yr, ≤ 2 cycles/yr, and ≥ 90 d since previous period), oligomenorrhea (age ≥ 16 yr, ≤ 9 cycles/yr, and no amenorrhea), and eumenorrhea (10-15 cycles/yr). The subjects who used contraceptive agents were divided in two groups: those who used pills to regulate their menstruation and those who used pills solely for prevention of pregnancy.
Because of the skewed distribution of particularly the EDI scores, the group differences for the continuous variables were tested by the Kruskal-Wallis nonparametric test, and by post-hoc Mann-Whitney comparisons. The results are reported as median values, and the 25th (Q1) and 75th (Q3) percentiles. The classified variables were tested by the Pearson chi-squared test, and post-hoc contrasts (20). Factors associated with menstrual dysfunction were tested by a fixed-model logistic regression, using age, weekly training hours, occurrence of weight-reduction attempts (yes/no), WDC score, and age of menarche as independent variables. P < 0.05 was accepted as the lowest level of significance. All statistical analyses were done by BMDP statistical soft-ware, 1990 version (University of California, Berkeley, CA).
The age range of all subjects was from 14 to 40 yr. The female aesthetic and weight-class athletes tended to be younger than the other groups, but the difference was not statistically significant (Table 2). The aesthetic athletes, including ballet dancers, were the leanest (P < 0.05) by both weight and BMI, and they also spent much more time in training than any other athletes. In male subjects, the speed and endurance athletes were older (P < 0.05) than aesthetic and ballgame athletes. The male aesthetic athletes were the leanest by weight (P < 0.05) but not by BMI (P = 0.23). The time spent in training was again clearly highest in male aesthetic athletes and ballet dancers, but the median training time for the ballgame and endurance athletes was also more than 15 h·wk−1.
Weight and diet concerns. The median WDC (weight and diet concerns = sum of subscales body dissatisfaction and drive for thinness) was 2.0 (Q1,3: 0.0, 9.0) and 2.5 (Q1,3: 0.0, 6.8) in female and male subjects, respectively (P = 0.62). The corresponding mean values were 6.4 (SD 9.0) and 5.0 (SD 6.8) in females and males, respectively. The female aesthetic athletes and controls had a rather similar distribution for WDC scores (Fig. 1A). All other female groups had seemingly lower scores. However, only the result of endurance athletes was significantly (P < 0.05) different from controls (Kruskal-Wallis, P = 0.003). The prevalence of individuals classified as at risk for eating disorders was highest in female weight-class athletes (13%), followed by controls (11%) and aesthetic athletes (7%). However, the group differences for the prevalences were not significant (P = 0.28).
Most male groups had zero as the median WDC score (Fig. 1B). Group differences were not significant (P = 0.08). The prevalences of at-risk subjects was also low (0-5%; P = 0.47).
The WDC and EDI-64 scores correlated significantly (r = 0.87, P < 0.001). Hence, the EDI-64 results resembled those of WDC, but the group differences were even smaller than for WDC. The median EDI-64 scores for females were between 10.0 (endurance) and 13.5 (aesthetic) (P = 0.81), and for males between 9.0 (speed) and 13.5 (aesthetic) (P = 0.48). The mean EDI-64 scores were 15.5 (SD 12.5) and 15.1 (SD 12.5) in females and males, respectively.
The preferred weight change (preferred weight minus actual body weight) was −2.0 (−4.0, −1.0) and 0.0 (−2.2, 0.0) in females and males, respectively (P < 0.001) (Fig. 2, A and B). The preferred weight change in the female control group (−4.0; Q1,3: −6.2, −2.0) was different (P < 0.05) from the result in the aesthetic (−2.0; Q1,3: −3.0, −1.0), speed (−2.0; Q1,3: −2.7, −0.2), endurance (−2.0; Q1,3: −3.0, −1.1), and weight-class (−1.0; Q1,3: −2.0, 0.0) athletes (Kruskal-Wallis P < 0.001). The preferred weight change was negatively associated with body mass index (r = −0.47 to −0.73, P < 0.03) in all athletic groups, except in weight-class athletes (r = 0.01, P = 0.95). The negative correlation between preferred weight change and BMI was significant also among control females (r = −0.86, P < 0.001).
Both male athletes and controls were, on average, satisfied with their present weight (Fig. 2B). However, the distribution was large with the preferred weight change ranging from −15 (an obese control) to +26 kg (a heavyweight wrestler). Group differences were not significant (P = 0.15). Another contrast to the female groups was that only ballgame athletes (r = −0.85, P < 0.001) and controls (r = −0.49, P = 0.002) showed a significant, negative correlation between BMI and the preferred weight change.
Weight reductions. Of the female participants, the weight-class athletes reported the highest prevalence (85%) of weight-reduction attempts (Table 3). This prevalence was significantly (P < 0.05) higher compared with endurance and ballgame athletes and the control group (29-58%). However, the frequency of weight reduction in the control females did not differ from any other athletic group, except from the weight-class athletes. The prevalence of weight reduction with rapid techniques, among the female participants, was clearly highest in the weight-class athletes (78%). The occurrence of unsupervised weight reduction was similar in all female groups (19-30%, P = 0.67).
The weight-reduction patterns in male subjects resembled those of the female participants, that is, both the overall frequency (93%) and the frequency of rapid weight reduction (79%) was highest in the weight-class athletes (P < 0.001). The occurrence of unsupervised weight reduction appeared to be higher in speed athletes versus weight-class and ballgame athletes. However, despite a significant chisquare test (P = 0.04), the post-hoc tests did not reach the 0.05 level.
Menstrual disorders. The age of menarche varied between 10 and 19 yr. The median menarcheal age of aesthetic, speed, and ballgame athletes (14 yr; Q1,3: 13, 16) was higher (P < 0.05) than in the control group (13 yr; Q1,3: 12, 14) (Kruskal-Wallis P < 0.001).
A total of 102 (40%) subjects used oral contraception. Secondary amenorrhea was found in only two aesthetic athletes (Fig. 3). However, the total frequency of menstrual dysfunction (primary and secondary amenorrhea and oligomenorrhea) among the nonpill users was 32-37% in aesthetic, endurance, and weight-class athletes but only 5% in controls (P = 0.06). When the subjects who used pills to regulate their menstruation were added to "menstrual disturbances," and the prevalence was calculated in all subjects, the frequency of menstrual dysfunction was still apparently highest (41%) in the endurance athletes (P = 0.08).
In the logistic regression analysis, menarcheal age (95% confidence interval for the exponent: 1.2-2.9) was the only variable that was significantly (P < 0.05) associated with the occurrence of menstrual dysfunction in nonpill users.
Weight and diet concerns. This is hitherto the largest study on eating problems among male athletes in several different sports, and after the landmark study of Sundgot-Borgen and Larsen (31), one of the largest reports on female athletes. The main outcome variable, sum score of body dissatisfaction and drive for thinness subscales (WDC) confirmed the earlier finding that the risk for eating disorders is dependent on the kind of athletic event (3,17,31). However, the claim that some female athlete groups are at greater risk than untrained controls (17,31,35) did not receive evidence in the present study. In contrast, female endurance athletes had lower WDC than untrained controls.
The finding of similar or lower EDI scores in athletes than in controls could theoretically have been caused by high EDI scores in the untrained control groups. However, compared with the mean EDI-64 scores (20.6 to 39.3) of untrained controls in other studies (3,17,27,36-38), and to the prevalence (22%) of subjects at risk for eating disorders (31), both the present athletes and controls scored low in WDC and EDI-64.
The wide range of control results in previous studies (3,17,27,36-38) underscore the difficulty of between-study comparisons. In addition to cultural factors, the age of the subjects, conditions of filling the questionnaire, and the instructions given to the subjects may affect the outcome (21). Like Sundgot-Borgen and Larsen (31) (J. Sundgot-Borgen, personal communication), we asked our subjects to fill in their name. This might have resulted in lower EDI-scores. However, also the perfectionism-subscale scores were low (mean 2.9 (SD 2.5) and median 2.0 for all subjects). Because faking is characterized by high perfectionism scores (21), we presume that the present results were not biased by deliberate cheating but that the subjects were more generally cautious in giving extreme answers.
Because comparison of EDI scores in different studies is apparently difficult, a representative control group is central in all studies on eating problems in athletes. The best and largest control group (522 age- and home community-matched subjects) appeared in the study of Sundgot-Borgen and Larsen (31). Our controls' age range was also close to the athletes' age range, which is important in studies on eating disorders (16). The present controls' social background was variable (i.e., not purely university level or working class), which also should allow an unbiased comparison with the athletes (16).
Female subjects, in comparison with men, were less satisfied with their present body weight, which agrees with other studies (40). In harmony with the WDC scores, the female controls were the least satisfied with their weight. Because a large discrepancy between actual and self-defined ideal weight seem to be associated with eating disorders (29), our finding was against the view of higher risk for eating disorders in athletes.
Weight reduction. The prevalence of athletes reporting weight reduction (51% and 55% among female and male athletes, respectively) was much higher than the 31% reported in Norwegian female athletes (31). Even without the weight-class athletes, the present prevalence was 44% and 33% in female and male athletes, respectively. However, also almost half of the female controls reported one or more weight loss attempts during the previous 12 months. An interesting finding was that the prevalence of weight reduction in males and females was similar among aesthetic, speed, endurance, and weight-class athletes but higher in female subjects among ballgame athletes and controls.
In agreement with others (28,33), most of the weight-class athletes used rapid weight reduction techniques. The prevalence of rapid weight reduction in other female and male groups (0-20%) was lower than in some studies from the U.S. (32-62%) (22,23) but similar to the 6-17% found among Norwegian female athletes and controls (31). The varying definitions of rapid or pathogenic weight reduction techniques may partly be responsible for different prevalences.
The data of Sundgot-Borgen (29) suggested that the risk of eating disorders was increased if weight reduction attempts were unsupervised. In the present study, the prevalence of unsupervised weight reduction was not higher in athletes compared with the controls. Weight-class athletes, in particular, seemed to have received guidance (usually from their coach) for weight reduction. Hence, the entire data on weight reduction attempts and techniques was in concert with the WDC data, that is, athletes did not appear to have an increased risk for eating disorders.
Menstrual disorders. The prevalence of menstrual abnormalities among athletes varies between 6 and 79%, depending on the definitions of amenorrhea/oligomenorrhea and on the population surveyed (2). Among the aesthetic, endurance, and weight-class athletes, the present prevalence of amenorrhea (8-15%) and any menstrual dysfunction (27-37%) was lower than the respective results in Norway (amenorrhea 26-36% and oligomenorrhea 50-62%) (32) but similar to results found among Dutch ballet dancers (10).
In agreement with Sundgot-Borgen and Larsen (32), the prevalence of menstrual disorders was highest in athlete groups characterized by a lean body (aesthetic), heavy training by volume (aesthetic) or intensity (endurance), or frequent weight reduction attempts (weight-class athletes). However, the above variables did not explain menstrual dysfunction in the logistic regression analysis. This apparent discrepancy is difficult to interpret, but it is probably related to the complex etiology of menstrual disorders (15) and difficulties in finding a suitable mathematical model. The importance of the age of menarche has been shown earlier (10,26).
General discussion. A critical evaluation of the present and earlier studies shows that the association between sports and eating disorders is not clear-cut. A speculative, but interesting, suggestion is that the prevalence of eating disorders in a given population is affected by the balance between promoting and protecting factors (21). Promoting factors might include (unsupervised) dieting (13,29), strenuous training (6,29), and sociocultural pressure for competition success or lean appearance (6,13). On the other hand, the risk for eating disorders might be decreased, if an increased energy expenditure (EE) in athletes helps in maintaining body weight. Moreover, appropriate nutrition education given at training camps or by the coach might also protect against eating problems. Finally, compared with an average schoolchild, athletes may cope psychologically better with the demands of adolescence and early adulthood (19).
The methodology and sometimes unexplicit definition of an eating disorder further complicates interpretation of results on eating problems in athletes. Compared with the prevalence of clinical eating disorders, questionnaires, such as the EDI, yield false positive results in a reference population (18,31). The opinions concerning athletic populations vary from a suspected overestimation (18) or underestimation (39) to correct identification (29) of clinical and sub-clinical eating disorders. Another question, which needs clarification, is the importance of some related behaviors, such as frequency of weight reduction, rapid weight reduction techniques, or unsupervised weight reduction (16,30).
1. Benson, J. E., Y. Allemann, G. E. Theintz, and H. Howald. Eating problems and calorie intake levels in Swiss adolescent athletes. Int. J. Sports Med.
2. Constantini, N. W. Clinical consequences of athletic amenorrhea
. Sports Med.
3. Davis, C., and M. Cowles. A comparison of weight and diet concerns and personality factors among female athletes and non-athletes. J. Psychosom. Res.
4. Davis, C., J. Fox, M. Cowles, P. Hastings, and K. Schwass. The functional role of exercise in the development of weight and diet concerns in women. J. Psychosom. Res.
5. Deuster, P. A., S. B. Kyle, P. B. Moser, R. A. Vigersky, A. Singh, and E. B. Schoomaker. Nutritional intakes and status of highly trained amenorrheic and eumenorrheic women runners. Fertil. Steril.
6. Eisler, I., and D. le Grange. Excessive exercise and anorexia nervosa. Int. J. Eating Disord.
7. Enns, M. P., A. Drewnowski, and J. A. Grinker. Body composition, body size estimation, and attitudes towards eating in male college athletes. Psychosom. Med.
8. Evers, C. L. Dietary intake and symptoms of anorexia nervosa in female university dancers. J. Am. Diet. Assoc.
9. Fogelholm, M. Effects of bodyweight reduction on sports performance. Sports Med.
10. Fogelholm, M., W. van Marken Lichtenbelt, R. Ottenheijm, and K. Westerterp. Amenorrhea
in ballet dancers in the Netherlands. Med. Sci. Sports. Exerc.
11. Frusztajer, N. T, S. Dhuper, M. P. Warren, J. Brooks-Gunn, and R. P. Fox. Nutrition and the incidence of stress fractures in ballet dancers. Am. J. Clin. Nutr.
12. Garner, D. M. Eating Disorder Inventory-2. Professional Manual. Odessa, FL: Psychological Assessment Resources, Inc., 1991.
13. Garner, D. M. Pathogenesis of anorexia nervosa. Lancet
14. Garner, D. M. and P. E. Garfinkel. The eating attitudes test: an index of the symptoms of anorexia nervosa. Psychol. Med.
15. Highet, R. Athletic amenorrhea
: an update on aetiology, complications and management. Sports Med.
16. Hsu, L. K. G. Eating Disorders.
New York: The Guilford Press, 1990.
17. 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.
18. Leon, G. R. Eating disorders
in female athletes. Sports Med.
19. Mallick, M. J., T. W. Whipple, and E. Huerta. Behavioral and psychological traits of weight-conscious teenagers: a comparison of eating-disordered patients and high- and low-risk groups. Adolescence
20. Marascuilo, L. A., and M. McSweeney. Nonparametric and distribution-free methods for the social sciences. Monterey, CA: Brooks/Cole Publishing Company, 1977.
21. O'Connor, P. J., R. D. Lewis, and E. M. Kirchner. Eating disorder symptoms in female college gymnasts. Med. Sci. Sports. Exerc.
22. Rosen, L. W., and D. O. Hough. Pathogenic weight-control behaviors of female college gymnasts. Physician Sportsmed.
23. Rosen, L. W., D. B. McKeag, D. O. Hough, and V. Curley. Pathogenic weight-control behaviour in female athletes. Physician Sportsmed.
24. Rosenvinge, J. H., and C. Vig. Eating disorders
and associated symptoms among adolescent swimmers. Scand. J. Med. Sci. Sports
25. Rucinski, A. Relationship of body image and dietary intake of competitive ice skaters. J. Am. Diet. Assoc.
26. Sandborn, C. F., B. H. Albrecht, and W. W. Wagner Jr. Athletic amenorrhea
: lack of association with body fat. Med. Sci. Sports Exerc.
27. Snead, D. B., C. C. Stubbs, J. Y. Weltman, et al. Dietary patterns, eating behaviors, and bone mineral density in women runners. Am. J. Clin. Nutr.
28. Steen, S. N., and K. D. Brownell. Patterns of weight loss and regain in wrestlers: has the tradition changed? Med. Sci. Sports Exerc.
29. Sundgot-Borgen, J. Risk and trigger factors for the development of eating disorders
in female athletes. Med. Sci. Sports Exerc.
30. Sundgot-Borgen, J. Eating disorders
in female athletes. Sports Med.
31. 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
32. Sundgot-Borgen, J., and S. Larsen. Preoccupation with weight and menstrual function in female elite athletes. Scand. J. Med. Sci. Sports
33. Sykora, C., C. M. Grilo, D. E. Wilfrey, and K. D. Brownell. Eating, weight, and diet disturbances in male and female light-weight and heavyweight rowers. Int. J. Eating Disord.
34. Thiel, A., H. Gottfried, and F. W. Hesse. Subclinical eating disorders
in male athletes. Acta Psychiatr. Scand.
35. Tofler, I. R., B. K. Stryer, L. J. Micheli, and L. R. Herman. Physical and emotional problems of elite female gymnasts. N. Engl. J. Med.
36. Walberg, J. L., and C. S. Johnston. Menstrual function and eating behavior in female recreational weight lifters and competitive body builders. Med. Sci. Sports Exerc.
37. Warren, B. J., A. L. Stanton, and D. L. Blessing. Disordered eating patterns in competitive female athletes. Int. J. Eating Disord.
38. Weight, L. M., and T. D. Noakes. Is running an analogue of anorexia?: a survey of the incidence of eating disorders
in female distance runners. Med. Sci. Sports Exerc.
39. Wilmore, J, H. Eating and weight disorders in the female athlete. Int. J. Sports Nutr.
40. Ziebland, S., M. Thorogood, A., Fuller, and J. Muir. Desire of the body normal: body image and discrepancies between self-reported and measured height and weight in a British population. J. Epidemiol. Community Health