The osteogenic impact of exercise on bone health in exercising women is well documented (12,13), and weight-bearing exercise, in particular, may act as a preventive strategy against low bone mineral density (BMD) and osteoporosis (21). However, in the presence of inadequate energy intake relative to exercise energy expenditure, also known as low energy availability (EA), certain exercising women may be susceptible to low BMD (11) and impaired bone microarchitecture (1). Low EA and/or an energy deficiency may negatively impact longitudinal bone growth and maturation in adolescent female athletes (7) and may promote bone loss in premenopausal exercising women with or without oligo/amenorrhea (20). All of these clinical outcomes are associated with an interrelated syndrome known as the female athlete triad (Triad) (26).
The Triad is a well-documented syndrome in exercising women and is characterized across a continuum of healthy to subclinical and clinical conditions of low EA (with or without disordered eating), menstrual disturbances, and low BMD (26). The Triad is most often observed in exercising women participating in leanness focused sport/activity (34) characterized by stringent weight control, such as long-distance running and gymnastics. Low EA and/or an energy deficiency are frequently associated with elevated dietary restraint and menstrual disturbances that often result in hypoestrogenism, a mechanism for bone loss in exercising women.
The interrelationships among Triad disorders in exercising women are well established (19,27,33). Low EA acts to promote low BMD via an energy deficiency (10), and it is presumably the disruptions in metabolic hormones, including reductions in insulin-like growth factor-1 (IGF-1) and leptin (16,38), that suppress bone formation. In addition, peak bone mineral accrual in girls is strongly associated with menarche (24). Thus, as adolescence is such a critical period for bone acquisition, positive benefits of exercise on bone may be negated in the presence of late menarche and menstrual dysfunction (11).
There is emerging interest in the evaluation of risk factors associated with the Triad, which may identify low BMD in exercising women (6,19,30). To date, no investigator has examined whether certain combinations of or multiple Triad risk factors are associated with a higher cumulative risk for low BMD in exercising women. Thus, a comprehensive examination of the association between Triad risk factors and low BMD in exercising women is warranted. Such data may also provide evidence in support of establishing screening and risk stratification strategies for low BMD associated with the Triad in exercising women.
The purpose of this study is twofold: 1) to evaluate the association between Triad risk factors (individually and in combination) and low BMD in exercising women and 2) to determine whether an increase in the percentage of exercising women with low BMD is associated with an increase in number of Triad risk factors. First, we hypothesized that exercising women with menstrual dysfunction (current oligo/amenorrhea or late menarche) and underweight status (low body mass index (BMI) or low body weight [BW]) in combination will demonstrate the highest risk for low BMD (z-score < −1 and ≤ −2). Second, we hypothesized that the percentage of exercising women with low BMD (z-score < −1 and ≤ −2) will increase as women meet the criteria for an increasing number of Triad risk factors, such that the risk for low BMD will be higher in association with a greater number of Triad risk factors. Third, we predicted that current oligo/amenorrhea, late menarche, low BMI, elevated dietary restraint, and lean sport/activity participation will represent significant predictors of risk for low BMD (z-score < −1 and ≤ −2) in exercising women.
MATERIALS AND METHODS
This retrospective study was designed to examine the percentage and the risk for low BMD associated with Triad factors among a large sample (n = 437) of adolescent and adult exercising women. The study consisted of cross-sectional data obtained at baseline from four prospective cohort studies completed in the United States and Canada (Pennsylvania State University [PSU], University of Toronto [UT], University of California at Los Angeles [UCLA], and San Diego State University [SDSU]).
Data collection from UCLA, collected between the fall of 1996 and the spring of 2001, consisted of female Division I collegiate track and field and cross-country athletes (n = 77). Participants were followed, prospectively, for 5 years. Data collection for the SDSU study occurred during the 2003–2004 academic year. Female interscholastic high school athletes were between ages 13 and 18 yr (n = 325). Before the 3- to 4-month sport season, participants were evaluated for risk factors for low BMD. Data collection from PSU and the UT, between 2005 and 2011, consisted of data from a cross-sectional study designed to assess cardiovascular status in exercising women (n = 54) and data from the baseline period of a prospective study designed to assess the effects of a 12-month intervention of increased energy intake on indices of bone health and menstrual status in exercising women with severe menstrual disturbances versus exercising women with regularly ovulatory menstrual cycles (n = 202). Participants in the PSU and UT studies were young, healthy adult women primarily from the general population of college-age women (18–35 yr). These participants included competitive and recreational-level exercising women.
For each study, one or more questionnaires were completed to obtain demographic and general background information including eating attitudes and behaviors, menstrual function, sports/activity participation, and medication use. Height and weight were measured. BMD and body composition were measured for each participant using dual-energy x-ray absorptiometry (DXA). Each of the studies excluded individuals taking medications known to affect BMD. Each study was approved by their respective university Institutional Review Board.
Study databases were merged in a careful and rigorous manner. Definitions of terms and methodology for the collection of variables from each study were reviewed. Data for variables that were measured and defined similarly among databases were merged. For variables that represented a similar data construct but were not defined and/or measured identically, a new variable, with a definition that combined the representations of the data construct, was created. For example, among the databases, amenorrhea was defined as the absence of three or more consecutive menstrual cycles in the past year and reporting a frequency of less than four menstrual cycles in the past year. In these situations, a new variable combining both definitions was used. Categorical variables that were collected and defined similarly but coded differently among databases were recoded to ensure consistency. Only participants with complete data collection for all examined study variables were included in this manuscript (n = 437). As such, participants with missing data for any one of the examined study variables were excluded from analyses (n = 221).
This study sample included 437 adolescent and adult women (ages 13–35 yr), with 17 from the UCLA site, 272 from the SDSU site, 34 from the UT site, and 114 from the PSU site. Participants were either recreational- or competitive-level exercising women who were involved in a variety of sports and exercise training. Recreational exercising women (n = 117) participated in ≥2 h·wk−1 of “purposeful exercise” (mean ± SD = 7.5 ± 5.3 h·wk−1) (2). Competitive exercising participants (n = 320) were current members of a high-school or collegiate athletic team. The inclusion criteria for this study were as follows: 1) age 13–35 yr, 2) no history of any chronic illness, 3) not taking any hormonal therapy in the past 12 months, and 4) not taking any medications that affect BMD. Written consent from participants (if participant was 18 yr or older) or participants and parents (SDSU study) was obtained before participation.
Height and weight, without shoes, was measured in each laboratory to the nearest 1.0 cm and 0.5 kg, respectively. BMI was calculated as a ratio of weight to height (kg·m−2).
Bone density and body composition
BMD (areal BMD, bone mineral content, and bone area) and body composition (percent body fat, fat mass, fat-free mass, and lean body mass [LBM]) were analyzed by a certified technician using DXA. Participants were scanned on either a GE Lunar (GE Lunar Corporation, Madison, WI) DPX-NT (n = 272), Prodigy (n = 34), iDXA (n = 114), or a Hologic QDR 4500A (n = 17, Hologic Inc., Bedford, MA) DXA scanner. For the SDSU study, the coefficients of variation (CV) for the BMD measurements were 0.6% for the total hip, 1.2% for the lumbar spine (L1–L4), and 0.99% for the total body. For the UCLA study, the CVs were 1% for total hip, 1% for the lumbar spine (L1-L4), and 1% for the total body. For the UT study, the CVs were 0.47% for the total hip, 1.0% for the lumbar spine (L1–L4), 1.03% at the femoral neck, and 0.65% at the total body. For the PSU study, the CVs were 0.61% at the total hip, 0.93% at the lumbar spine (L1–L4), 0.97% at the femoral neck, and 0.66% at the total body. Because no cross-calibration study was completed, data analyses for primary objectives used standardized scores only.
Menstrual function was determined in all participants using a questionnaire. Our criteria for menstrual status considered the number of self-reported menstrual cycles in the past 12 months or menstrual cycle length before study enrollment. Categories of menstrual status included the following: 1) amenorrhea defined as the absence of menses for at least three consecutive months in the past year or less than four cycles in the past year, 2) oligomenorrhea defined as a menstrual cycle length >35 d or between four and nine cycles in the past year, or 3) eumenorrhea defined as a menstrual cycle length between 24 and 35 d or ≥10 cycles in the past year. Age of menarche (yr) was reported, and late menarche was defined as the onset of menses at age 15 yr or older (28).
Eating attitudes and behaviors
Dietary restraint was obtained using the Three-Factor Eating Questionnaire (TFEQ) (32) (PSU and UT), the Eating Disorder Examination Questionnaire (EDE-Q) (25) (SDSU), or a preparticipation examination questionnaire corroborated by clinical interview with a physician (UCLA). The TFEQ is a 51-item questionnaire that measures three dimensions of human eating behavior: 1) dietary restraint, 2) disinhibition, and 3) hunger (32). A cutoff of ≥9 was used to identify an elevated value for the dietary restraint subscale. A TFEQ dietary restraint score of ≥9 was chosen as prior investigators have reported this cutoff as the median score for premenopausal women (37). The EDE-Q measures four subscales: 1) weight concern, 2) shape concern, 3) eating concern, and 4) dietary restraint. Scores ranging from 0 to 6 on a Likert scale correspond to the number of days over the past 4 wk the respondent experienced a particular attitude, feeling, or behavior. An EDE-Q dietary restraint score of ≥3 was chosen to identify an elevated value (5,27). Because all participants in the current study were assessed for dietary restraint using the TFEQ, the EDE-Q, or an eating attitudes/behavior survey, elevated dietary restraint was defined/identified as either a score of ≥9 on the TFEQ, a score of ≥3 on the EDE-Q, or by a preparticipation examination questionnaire corroborated in a clinical interview with a physician. History of pathogenic weight control behavior was defined as reporting at least one prior episode of self-induced vomiting or use of laxatives, diuretics, and/or diet pills (34).
Sport/activity participation was obtained using a questionnaire wherein the participants reported the type, frequency (d·wk−1), and duration per session (min·d−1) in the past 6 months (PSU and UT) or number of years and/or months per year of the sport in which they participated from age 10 yr to current year (UCLA and SDSU). Participants were categorized based on primary mode of sport/activity (lean vs nonlean), using a classification system modified from Torstveit and Sundgot-Borgen (36). Lean sports/activities included endurance (cycling, middle- and long-distance running [>800 m], rowing, swimming, triathlon, and cross-country skiing), aesthetic (figure skating, gymnastics, rhythmic gymnastics, dance, diving, cheerleading, and acrobatics), weight class (boxing, weight lifting, karate, judo, tae kwon do, wrestling, kickboxing, jujitsu, and horseback riding–racing), antigravitation sports (indoor and outdoor rock climbing, high jump, long jump, pole vaulting, and triple jump), and aerobic activities (aerobics and gym-related cardiorespiratory exercise such as elliptical and stair climber machines). Nonlean sports/activities included technical (bowling, curling, fencing, golf, horseback riding–dressage, sailing, snowboarding, and water skiing), ball (badminton, basketball, ice hockey, field hockey, soccer, lacrosse, softball, baseball, table tennis, squash, team handball, tennis, rugby, and volleyball), and power sports (alpine skiing, discus, hammer, hurdle, javelin, shot put, speed skating sprint, and sprint (≤800 m) (36).
Definitions of low BMD
Participants were categorized as having low BMD if their BMD values at the lumbar spine or total body in the adolescent population (age 15 to 17 yr) (n = 268) or the lumbar spine or total hip in the premenopausal female population (age 18 to 35 yr) (n = 169) were more than 1 or more than or equal to 2 SD below the age-matched, sex-specific reference data from the GE/Lunar and Hologic databases (z-score < −1 or ≤ −2, respectively).
The International Society of Clinical Densitometry (ISCD) defines low BMD for chronological age as an age, sex, and body size-adjusted BMD z-score ≤ −2, whereas the American College of Sports Medicine (ACSM) 2007 Position Stand on the Triad (26) defines low BMD for an exercising woman as a BMD z-score < −1, particularly if accompanied by secondary clinical risk factors for fracture, such as amenorrhea. We chose to examine both z-score cutoffs (ACSM and ISCD criteria) (30) because exercising women, particularly those engaged in weight-bearing sport/activity, typically should demonstrate BMD values approximately 10% higher than sedentary controls (26). As such, a BMD z-score between −1 and −2 requires attention.
Triad risk factors for low BMD
Before the initiation of data analysis, we identified potential risk factors of low BMD associated with the Triad (i.e., factors associated with low EA/chronic energy deficiency and menstrual dysfunction). Definitions and/or cutoff points used for each risk factor were decided based on their recognition as the criteria for an established clinical condition, or evidence linking the factor to the Triad and related clinical sequelae. The risk factors assessed included the following: low BMI (<18.5 kg·m−2), low BW (<90% ideal BW [IBW]), elevated dietary restraint, pathological weight control behavior, lean sport/activity participation, late menarche, and current oligo/amenorrhea. The Centers for Disease Control cites a BMI <18.5 kg·m−2 as underweight. Torstveit and Sundgot-Borgen (34) also defined BMI <18.5 kg·m−2 as an at-risk criterion of the Triad in a large population of competitive female athletes by assuming that a very low BMI may indicate chronic energy deficiency. In addition, a BW less than 10% of ideal, using the IBW equation (18), has been used to define a moderate underweight status (29). Amenorrhea and oligomenorrhea are considered the most severe clinical menstrual disturbances in the 2007 ACSM Position Stand on the Triad (26). Primary or secondary amenorrhea is one of the required criteria defined by the World Health Organization and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition for anorexia nervosa and also represents a well-documented factor associated with compromised bone mineral accretion and/or bone loss in exercising women (11). Pathological weight control behaviors are unhealthy dieting strategies that may be used by female athletes to achieve a low BW considered advantageous for performance or appearance-related reasons (34). Late menarche indicates a history of primary amenorrhea and is associated with a negative effect on bone mineral accrual during adolescence (24). Investigators have previously reported associations among the following factors associated with the Triad and low BMD in exercising women: low BMI/BW, elevated dietary restraint (as measured by the EDE-Q, TFEQ, or clinical interview), lean sport/activity participation, and menstrual dysfunction (5,6,30,33,37).
All statistical analyses were conducted using the Statistical Package for the Social Sciences (version 19.0; SPSS Inc., Chicago, IL). All hypothesis tests were two sided, and P < 0.05 was considered significant. Data screening was conducted before analysis, involving outlier detection, evaluation of assumptions of normality, and regression diagnostics. Clinical characteristics (i.e., age, ethnicity, height, body mass, BMI, body composition, age of menarche, and gynecological age) and percentage of Triad risk factors were determined to describe the study sample. Descriptive statistics were reported to include means and standard deviations for continuous data and frequency and percent for categorical data between participants with BMD z-score < −1 and those participants with BMD z-score ≥ −1 based on the ACSM criteria (26). Psychometric survey data were recoded and scored to determine subscale or total scores. Group comparisons for descriptive purposes were performed using independent Student t-tests for continuous data and chi-square tests for categorical data. When variables were not normally distributed, the Mann–Whitney test was used to determine group differences.
Chi-square tests were performed to determine the following: 1) the association between Triad risk factors (individually and in combination) and low BMD (z-score < −1 and ≤ −2) in exercising women and 2) the association between the cumulative effect of Triad risk factors in exercising women (those who met the criteria for multiple [any 2, 3, 4, or 5] Triad risk factors) and low BMD (z-score < −1 and ≤ −2). Crude odds ratios (OR) and 95% confidence intervals (95% CI) were determined to compare the risk for low BMD (z-score < −1 and ≤ −2) in participants with risk factors associated with the Triad (individually and in combination) compared with those participants that did not meet the criteria for the evaluated individual or combined Triad risk factors. Adjusted OR with 95% CI were generated using multivariable logistic regression analyses to determine the strongest predictors of low BMD in exercising women (31). Items in the multivariable logistic regression model analyses included the Triad risk factors (current oligo/amenorrhea, late menarche, low BMI, elevated dietary restraint, and lean sport/activity participation) and relevant covariates known to affect BMD (age [yr] and LBM [kg]). On the basis of the tight correlation between BMI and BW, low BW was excluded from the multivariable regression analyses. Adjusted models were conducted for low BMD z-score < −1 and low BMD z-score ≤ −2. A sample size calculation based on data from previous publications (5,8,19,33) indicated that 375 women would provide adequate power (1 − β = 0.80) to detect a significant 1) association between Triad risk factors (individually and in combination) and low BMD (z-score < −1 and ≤ −2) in exercising women and 2) association between the cumulative effect of Triad risk factors in exercising women (those who met the criteria for multiple [any 2, 3, 4, or 5] Triad risk factors) and low BMD (z-score < −1 and ≤ −2).
The descriptive characteristics of the total sample of 437 exercising women are presented in Table 1. Participants in this study were 18.0 ± 3.5 yr, weighed 57.5 ± 7.1 kg with 24.5% ± 6.1% body fat. The sample included a racial/ethnic distribution of 70.5% White/Caucasian, 9.6% Hispanic, 7.8% Asian, 7.1% African American, and 5.0% chose “other.” In the total sample, 16.9% of the participants reported using oral contraceptives in their lifetime. The percentage of Triad risk factors for the total sample is reported in Table 1. Of the total sample of participants, 42.8% presented with current oligo/amenorrhea, 71.6% participated in lean sport/activity, 13.3% demonstrated late menarche, 30.2% exhibited an elevated dietary restraint, 8.0% had engaged in pathological weight control behavior, and 13.7% and 11.4% had low BW and low BMI, respectively.
Descriptive characteristics and percentage of Triad risk factors in exercising women grouped by BMD status
Descriptive characteristics and percentage of Triad risk factors in exercising women grouped by BMD status are presented in Table 1. Participants with BMD z-score < −1 were lighter, leaner, less gynecologically mature, and, as expected, exhibited a lower BMD at the total body, lumbar spine, total hip, and femoral neck than participants with a BMD z-score ≥ −1. A greater percentage of the participants with BMD z-score < −1 compared with participants with BMD z-score ≥ −1 participated in a lean sport/activity, demonstrated current oligo/amenorrhea and late menarche, and had low BW/BMI.
Risk for low BMD associated with individual Triad risk factors
The percentage of low BMD (z-score < −1 and ≤ −2) in exercising women with individual Triad risk factors is shown in Figure 1. Current oligo/amenorrhea (χ2 = 5.6, P = 0.018), late menarche (χ2 = 22.6, P < 0.001), low BMI (χ2 = 17.5, P < 0.001), low BW (χ2 = 8.9, P = 0.003), and lean sport/activity participation (χ2 = 17.3, P < 0.001) were associated with low BMD (z-score < −1). Current oligo/amenorrhea (χ2 = 4.1, P = 0.043), late menarche (χ2 = 10.7, P = 0.001), and low BMI (χ2 = 14.2, P < 0.001) were associated with low BMD (z-score ≤ −2). Of the individual risk factors, participants with late menarche, low BMI, and low BW demonstrated the highest percentage of low BMD (z-score < −1), 55.2%, 54.0%, and 45.0%, respectively. Participants with low BMI, late menarche, and low BW also demonstrated the highest percentage of low BMD (z-score ≤ −2), 16.0%, 13.8%, and 10.0%, respectively. Crude OR with 95% CI for low BMD (z-score < −1 and ≤ −2) in exercising women with individual Triad risk factors are presented in Table 2. Participants that reported late menarche (n = 58) were four times more likely to present with low BMD (z-score < −1 and ≤ −2) compared with those with normal age at the onset of menarche (n = 379). Participants that had low BW/BMI were more likely to present with low BMD (z-score < −1 and ≤ −2). Participants that reported low BMI (n = 50) were three and five times more likely to present with low BMD (z-score < −1 and ≤ −2) compared with those with normal BMI (n = 387). Those participants that reported low BW (n = 60) were two times more likely to present with low BMD (z-score < −1) compared with those with normal BW (n = 377). Participants that reported current oligo/amenorrhea (n = 187) were two and three times more likely to present with low BMD (z-score < −1 and ≤ −2) compared with those with current eumenorrhea (n = 250). Lean sport/activity participants (n = 313) were three times more likely to demonstrate low BMD (z-score < −1) compared with nonlean sport/activity participants (n = 124).
Risk for low BMD associated with combined Triad risk factors
Of the combined risk factors, participants with late menarche + low BMI, late menarche + low BW, and late menarche + elevated dietary restraint demonstrated the highest percentage of low BMD (z-score < −1), 73.3%, 70.6%, and 64.0%, respectively. Participants with late menarche + low BMI, late menarche + low BW, and current oligo/amenorrhea + late menarche demonstrated the highest percentage of low BMD (z-score ≤ −2), 20.0%, 17.6%, and 17.4%, respectively. Crude OR with 95% CI for low BMD (z-score < −1 and ≤ −2) in exercising women with combined Triad risk factors are presented in Table 3. Varying combinations of risk factors indicative of current oligo/amenorrhea, late menarche, low BMI/BW, and lean sport/activity participation were strongly associated (P < 0.05) with low BMD (z-score < −1 and ≤ −2). Elevated dietary restraint + lean sport/activity participation and elevated dietary restraint + late menarche also represented combinations of Triad risk factors strongly associated (P < 0.05) with low BMD (z-score < −1) (Table 3).
Risk for low BMD based on number of Triad risk factors
The percentage of low BMD (z-score < −1 and ≤−2) in exercising women who met the criteria for 0 (n = 59), 1 (n = 159), 2 (n = 120), 3 (n = 58), or 4 (n = 39) Triad risk factors (low BMI, late menarche, elevated dietary restraint, lean sport/activity participation, and current oligo/amenorrhea) are presented in Figure 2. The percentage of low BMD (z-score < −1 and ≤ −2) in participants increased from 10.2% to 61.5% and from 1.7% to 17.9%, respectively, as participants met the criteria for 0, 1, 2, 3, or 4 Triad risk factors. Two participants met the criteria for all five risk factors and BMD z-score < −1, and zero participants met the criteria for all 5 risk factors and BMD z-score ≤ −2.
Triad risk factors as predictors for low BMD
Adjusted OR with 95% CI for low BMD (z-score < −1 and ≤ −2) in exercising women associated with Triad risk factors are presented in Table 4. All risk factors (low BMI, late menarche, elevated dietary restraint, lean sport/activity participation, and current oligo/amenorrhea) were entered into the model in addition to relevant covariates (age [yr] and LBM [kg]). Multivariable logistic regression analyses revealed that late menarche, lean sport/activity participation, and LBM were the strongest predictors of low BMD (z-score < −1) when adjusting for BMI status, menstrual status, dietary restraint status, and age. Late menarche and LBM were the strongest predictors of low BMD (z-score ≤ −2) when adjusting for sport/activity participation type, BMI status, menstrual status, dietary restraint status, and age.
In this study, we examined the cumulative effect of Triad risk factors associated with low BMD in adolescent and adult exercising women. The Triad is well recognized as an interrelated syndrome wherein several clinical sequelae (i.e., menstrual dysfunction, disordered eating, low EA), ranging in severity, may predominate and have a negative effect on BMD (5,6,11,27). To our knowledge, this is the first investigation that has reported a “dose response” or cumulative effect of Triad risk factors on BMD. We demonstrated that exercising women presenting with multiple Triad risk factors were at a higher risk for low BMD than exercising women with no or individual Triad risk factors. In addition, we reported associations between combined Triad risk factors, particularly those related to menstrual dysfunction and low BMI/BW, and higher risk for low BMD. Furthermore, our findings suggest that as the number of Triad risk factors increase from one to four factors in exercising women, there is an increase in the percentage of participants with low BMD (z-score < −1 and z-score ≤ −2), ranging from 21% to 62% and from 3% to 18%, respectively. Therefore, in our study, the exercising women with greater exposure to Triad risk factors were more likely to demonstrate low BMD. Findings from the current study also reinforce prior associations between Triad risk factors and low BMD such as late menarche (6), current oligo/amenorrhea (11), low BMI/BW (14,39,40), participation in lean sport/activity (8,35), and elevated dietary restraint (5,37). As such, the presentation of Triad-related disorders (i.e., hypoestrogenism, low EA, disordered eating behavior) likely translates to higher risk of low BMD (23), particularly in those exercising women with multiple persistent Triad conditions. Overall, our findings underscore the importance of early intervention in the presence of any one of the evaluated Triad risk factors in order to prevent an accumulation of risk for low BMD in exercising women.
Current oligo/amenorrhea and late menarche are risk factors indicative of chronic menstrual dysfunction with clinical implications for bone health. In our study, late menarche was a significant predictor of low BMD in the presence of other Triad risk factors, and notably, 55% and 14% of participants with late menarche reported low BMD (z-score < −1 and z-score ≤ −2, respectively). Adolescence is a critical time for bone mineral accretion, wherein approximately 35% of total body and lumbar spine bone mineral and 27% of femoral neck bone mineral are deposited (3). As such, normal age of menarche promotes skeletal benefits in exercising girls, whereas late menarche may result in decreased estrogen exposure, which has the potential to exert a profound effect on the axial skeleton and in trabecular bone (4,11,24). Estrogen also plays a key role in maintaining BMD in adult women, and importantly, oligo/amenorrhea associated with hypoestrogenism may represent a potential indicator of risk for bone loss (11). In this study, low BMD (z-score < −1 and z-score ≤ −2) was present in a significant proportion of exercising women with current oligo/amenorrhea, 35% and 8%, respectively, and the risk for low BMD in these participants was higher than their regularly menstruating counterparts (10,11). In addition, the combination of these markers of menstrual dysfunction (current oligo/amenorrhea and late menarche) indicated a significantly higher risk for low BMD compared with a regularly menstruating woman with a normal age of menarche. Notably, self-report menstrual status can only identify menstrual disturbances clearly apparent to women as an absence of menses for greater than 3 months (amenorrhea), an irregularity in menstrual cycle length (oligomenorrhea), or late menarche (≥15 yr). Self-reported menstrual history alone does not detect subclinical menstrual disturbances (i.e., luteal phase defects and anovulation) (9). As such, objective measures of hormone analyses improve the accuracy of measuring the presence or absence of menstrual disturbances (subclinical and clinical). However, the implications for BMD in women with clinical menstrual disturbances (oligo/amenorrhea) are much more severe based on lower estrogen exposure compared with those with subclinical menstrual disturbances (11). Furthermore, objective measures of menstrual status are often not feasible in a clinical or field setting. Taken together, self-report measures of menstrual function may provide an easy-to-obtain indication of risk for low BMD in exercising women.
Several investigators have reported associations among energy deficiency, menstrual disturbances, and low BMD in exercising women (10). Previous evidence supports the combined effect of estrogen and energy deficiency on bone turnover markers and/or BMD (10,40). In a study by De Souza et al. (10), researchers reported that estrogen deficiency in the presence of an energy deficiency was associated with low BMD in exercising women. In the current study, combinations of menstrual dysfunction (late menarche or current oligo/amenorrhea) and low BMI/BW were associated with a higher risk for low BMD in exercising women compared with those with normal menstrual function and normal BMI/BW. Specifically, the combination of low BMI and late menarche represented the most robust indicator of risk for low BMD such that participants with late menarche and low BMI were up to seven times more likely to demonstrate low BMD compared with those with normal age of menarche and normal BMI. Accordingly, the percentage of low BMD (z-score < −1) in exercising women with different combinations of low BMI, low BW, late menarche, or current oligo/amenorrhea was notably high, ranging from 49% to 73%, indicative of clusters of Triad risk factors associated with the highest risk for low BMD.
Low BMI and low BW represented individual Triad risk factors of low BMD. We demonstrated that participants with low BMI/BW were between two to five times more likely to demonstrate low BMD. Low BMI and low BW were also significantly associated with low BMD in combination with other Triad risk factors, suggesting that there is a cumulative effect of low BMI/BW on BMD in exercising women along with late menarche, lean sport/activity participation, elevated dietary restraint, and/or current oligo/amenorrhea. In addition, findings from case study reports in female athletes support the importance of nutritional and weight recovery to coincide with BMD recovery (14,39). From a methodological standpoint, the measurement of bone turnover markers, metabolic hormones, and DXA, or other imaging analyses of BMD, are often not feasible in clinical and field settings. Therefore, the assessment of BMI or BW may represent a noninvasive first-pass indicator of risk for low BMD, which can serve to assist health care providers in determining the need for further assessment and workup in athletes, especially when low BMI/BW is exhibited in the presence of additional Triad-related risk factors for low BMD.
The Triad is most often observed in exercising women participating in lean sport/activity due to the emphasis on low BW and possibly elevated dietary restraint, which may lead to low EA or chronic energy deficiency (33). Beals and Hill (8) reported low BMD (defined as a z-score < −2.0) in 3.1% and 0% of lean and nonlean sport athletes, respectively. The prevalence of low BMD increases to 15.4% in lean sport athletes using a less conservative criterion of z-score < −1 for low BMD compared with 0% in nonlean sport athletes (8). In addition, lean sport athletes exhibited 7% lower lumbar spine (L1–L4) BMD than the nonlean sport athletes (8). Similarly, we demonstrated that the lean sport/activity participants were more likely to present with low BMD compared with their nonlean sport/activity counterparts. The lower BMD in the lean sport/activity group observed in our study may be explained by a greater frequency of current oligo/amenorrhea in the lean sport/activity group versus the nonlean sport/activity group (48% vs 30%, respectively). In addition, there may be a genetic predisposition of those exercising women with higher bone strength and greater LBM to favor participation in nonlean sport/activity, whereas those women with low BW and a lean figure (specifically, lower LBM) may self-select participation in lean sport/activity (35). Our findings showed that both lean sport/activity participation and LBM were identified as significant predictors of low BMD when adjusting for the other Triad risk factors and age. Accordingly, participants with low BMD (z-score < −1) demonstrated significantly lower LBM in comparison with women with normal BMD (z-score ≥ −1). Therefore, we suggest that lean sport/activity may have negative consequences on BMD independent of and in combination with other Triad risk factors. Furthermore, we demonstrated the robust effects of LBM on BMD in exercising women.
Associations between low BMD and disordered eating behavior (i.e., dietary restraint and history of pathological weight control behavior) have been established in previous reports in adolescent and adult exercising women (5,8,30). Findings from Barrack et al. (5) in adolescent female competitive cross-country runners demonstrate that elevated dietary restraint has a negative effect on BMD and bone mineral content, particularly at the lumbar spine. Vescovi et al. (37) provided evidence that elevated dietary restraint scores were associated with reduced lumbar spine and total body BMD in exercising women concomitant with a greater frequency of current oligo/amenorrhea. In contrast to these previous reports, we found that elevated dietary restraint was not associated with low BMD as an individual risk factor. However, when combined with other Triad risk factors, such as lean sport/activity participation and late menarche, elevated dietary restraint was associated with a higher percentage of participants with low BMD. As such, elevated dietary restraint may contribute to the accumulation of risk for low BMD in exercising women. Notably, dietary restraint and history of pathological weight control behavior represented the only indicators of disordered eating behavior examined in this study. However, other disordered eating behaviors such as a high drive for thinness or current or history of an eating disorder may also be linked to low BMD by way of a chronic energy deficiency and/or hypoestrogenism (15).
Because the Triad is a complex syndrome to assess, the study of this interrelated medical condition is limited by methodological and experimental challenges. The objective of merging data obtained from several studies was to examine the cumulative effect of Triad risk factors on BMD in a large sample of adolescent and adult exercising women, which is notably one of the strengths of this study. The use of self-report or field measures in the present study may represent a useful approach in identifying low BMD in exercising girls and women in a clinical or field setting, wherein “gold-standard” methods are not feasible. The advancement of a practical approach using easy-to-obtain and noninvasive measures to assess risk for low BMD associated with the Triad is necessary, particularly for physicians, athletic trainers, and health practitioners. Furthermore, the assessment of these risk factors may be considered for validation and incorporation into preparticipation screening in athletes at risk for low BMD. Importantly, these Triad risk factors may be effective at identifying exercising women that may require further assessment of BMD using DXA or other imaging analyses.
We acknowledge that our retrospective analysis involved certain limitations. Particularly, our sample includes exercising women of a wide range in age and growth/development similar to other large-scale studies of Triad risk factors in athletic and exercising women (33,34). It is important to mention that there are distinct physiological and developmental differences between adolescent and adult exercising women. However, based on prior investigations in adolescent and high-school athletes (5,19,27), it is clear that the Triad is a clinically relevant issue in this younger population of exercising women similar to their adult counterparts. Also, our analysis of BMD includes data from multiple DXA scanners, and no cross-calibration study was completed. Thus, data were compared using standardized scores only. Other risk factors that were considered but not included in this study due to inconsistencies or lack of measurement in any one of the included studies were current or prior history of a clinical eating disorder, menstrual disturbances before the past year, cumulative menstrual history, subclinical menstrual disturbances, high training volume, high drive for thinness, and low EA. Despite our inability to assess these potential risk factors, we encourage future investigation of other Triad risk factors contributing to this concept of cumulative risk for low BMD associated with the Triad in exercising girls and women. Lastly, it is difficult to interpret our findings with respect to the influence of sport/activity participation (lean vs nonlean) on BMD because both groups included low-, medium-, and high-impact loading sport/activity. Therefore, classifying exercising women by sport/activity type defined as lean versus nonlean may not appropriately capture the influence of mechanical loading properties on BMD (35). Future research should determine the degree of mechanical loading or the peak strain score in these exercising women similar to prior investigations (17,22,35).
In conclusion, our retrospective study provides evidence of a cumulative effect of Triad risk factors on BMD in adolescent and adult exercising women. In addition, we comprehensively evaluated the percentage of exercising women with low BMD associated with Triad risk factors (individually and in combination). Future investigation on the utility of these Triad risk factors in a preparticipation screening protocol or a user-friendly algorithm to identify risk for low BMD and risk for bone stress injury and fracture in exercising women is necessary. Optimally, these risk factors may be used as indicators of risk for low BMD in exercising women in a clinical or field setting and may provide useful information on BMD status. The development of preventive and treatment strategies should focus on validating first-pass indicators of risk for low BMD for use by physicians, athletic trainers, and health professionals to flag at-risk exercising women for which additional assessment using objective measures, such as DXA or other imaging analyses, would be recommended. Novel strategies for early detection of possible pathological bone loss, failure to reach peak bone mass, or bone stress injury may be paramount in determining withdrawal from or clearance for participation in sports/exercise and preventing detrimental bone loss across the life span. Therefore, this study presents valuable information on the cumulative effect of Triad risk factors on BMD in exercising women and emphasizes the importance of healthy, replete energy status combined with normal menstrual function and eating behavior to optimize osteogenic loading forces on BMD achieved during sport and exercise.
This study was funded by the U.S. Department of Defense, Army Medical Research and Material Command (grant no. PR054531); the Arthur Thornton Cardiopulmonary Fund, New Britain General Hospital, Connecticut; the UCLA (GCRC no. M01-RR00865); the U.S. Olympic Committee Sports Science and Technology Grant; the University of California Center for Health and Nutrition Research; the NATA Research Foundation (grant nos. 206GGP008 and 903GGT005); and the ACSM NASA Space Physiology Research. There are no further disclosures and conflicts of interest to report for any of the authors.
The authors extend their gratitude to Dr. David A. Wagstaff for his statistical consultation and all of the participants for their time and efforts in these research studies.
The findings of the present study do not constitute endorsement by the ACSM.
1. Ackerman KE, Nazem T, Chapko D, et al. Bone microarchitecture is impaired in adolescent amenorrheic athletes compared with eumenorrheic athletes and nonathletic controls. J Clin Endocrinol Metab. 2011; 96 (10): 3123–33.
2. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. 7th ed. Philadelphia (PA): Lippincott, Williams, and Wilkins; 2006.
3. Bailey DA, McKay HA, Mirwald RL, Crocker PR, Faulkner RA. A six-year longitudinal study of the relationship of physical activity
to bone mineral accrual in growing children: the university of Saskatchewan bone mineral accrual study. J Bone Miner Res. 1999; 14 (10): 1672–9.
4. Barkai HS, Nichols JF, Rauh MJ, Barrack MT, Lawson MJ, Levy SS. Influence of sports participation and menarche on bone mineral density of female high school athletes. J Sci Med Sport. 2007; 10 (3): 170–9.
5. Barrack MT, Rauh MJ, Barkai HS, Nichols JF. Dietary restraint and low bone mass in female adolescent endurance runners. Am J Clin Nutr. 2008; 87 (1): 36–43.
6. Barrack MT, Rauh MJ, Nichols JF. Prevalence of and traits associated with low BMD among female adolescent runners. Med Sci Sports Exerc. 2008; 40 (12): 2015–21.
7. Baxter-Jones AD. Growth and development of young athletes. Should competition levels be age related? Sports Med. 1995; 20 (2): 59–64.
8. Beals KA, Hill AK. The prevalence of disordered eating, menstrual dysfunction, and low bone mineral density among US collegiate athletes. Int J Sport Nutr Exerc Metab. 2006; 16 (1): 1–23.
9. De Souza MJ, Toombs RJ, Scheid JL, O’Donnell E, West SL, Williams NI. High prevalence of subtle and severe menstrual disturbances in exercising women: confirmation using daily hormone measures. Hum Reprod. 2010; 25 (2): 491–503.
10. De Souza MJ, West SL, Jamal SA, Hawker GA, Gundberg CM, Williams NI. The presence of both an energy deficiency and estrogen deficiency exacerbate alterations of bone metabolism in exercising women. Bone. 2008; 43 (1): 140–8.
11. Drinkwater BL, Bruemner B, Chesnut CH 3rd. Menstrual history as a determinant of current bone density in young athletes. JAMA. 1990; 263 (4): 545–8.
12. Duncan CS, Blimkie CJ, Cowell CT, Burke ST, Briody JN, Howman-Giles R. Bone mineral density in adolescent female athletes: relationship to exercise type and muscle strength. Med Sci Sports Exerc. 2002; 34 (2): 286–94.
13. Fehling PC, Alekel L, Clasey J, Rector A, Stillman RJ. A comparison of bone mineral densities among female athletes in impact loading and active loading sports. Bone. 1995; 17 (3): 205–10.
14. Fredericson M, Kent K. Normalization of bone density in a previously amenorrheic runner with osteoporosis. Med Sci Sports Exerc. 2005; 37 (9): 1481–6.
15. Gibbs JC, Williams NI, Scheid JL, Toombs RJ, De Souza MJ. The association of a high drive for thinness with energy deficiency and severe menstrual disturbances: confirmation in a large population of exercising women. Int J Sport Nutr Exerc Metab. 2011; 21 (4): 280–90.
16. Grinspoon S, Baum H, Lee K, Anderson E, Herzog D, Klibanski A. Effects of short-term recombinant human insulin-like growth factor I administration on bone turnover in osteopenic women with anorexia nervosa. J Clin Endocrinol Metab. 1996; 81 (11): 3864–70.
17. Groothausen J, Siemer H, Kemper HCG, Twisk JWR, Welten DC. Influence of peak strain on lumbar bone mineral density: an analysis of 15-year physical activity
in young males and females. Ped Exerc Sci. 1997; 9 (2): 159–73.
18. Hamwi GJ. Therapy: changing dietary concepts. In: Danowski TS, editor. Diabetes Mellitus: Diagnosis and Treatment. New York (NY): American Diabetes Asspiation; 1964. pp. 73–8.
19. Hoch AZ, Pajewski NM, Moraski L, et al. Prevalence of the female athlete triad in high school athletes and sedentary students. Clin J Sport Med. 2009; 19 (5): 421–8.
20. Ihle R, Loucks AB. Dose–response relationships between energy availability and bone turnover in young exercising women. J Bone Miner Res. 2004; 19 (8): 1231–40.
21. Kemmler W, Engelke K, Weineck J, Hensen J, Kalender WA. The Erlangen Fitness Osteoporosis Prevention Study: a controlled exercise trial in early postmenopausal women with low bone density-first-year results. Arch Phys Med Rehabil. 2003; 84 (5): 673–82.
22. Kemper HC, Bakker I, Twisk JW, van Mechelen W. Validation of a physical activity
questionnaire to measure the effect of mechanical strain on bone mass. Bone. 2002; 30 (5): 799–804.
23. Khan AA, Hanley DA, Bilezikian JP, et al. Standards for performing DXA in individuals with secondary causes of osteoporosis. J Clin Densitom. 2006; 9 (1): 47–57.
24. McKay HA, Bailey DA, Mirwald RL, Davison KS, Faulkner RA. Peak bone mineral accrual and age at menarche in adolescent girls: a 6-year longitudinal study. J Pediatr. 1998; 133 (5): 682–7.
25. Mond JM, Hay PJ, Rodgers B, Owen C. Eating Disorder Examination Questionnaire (EDE-Q): norms for young adult women. Behav Res Ther. 2006; 44 (1): 53–62.
26. Nattiv A, Loucks AB, Manore MM, Sanborn CF, Sundgot-Borgen J, Warren MP. American College of Sports Medicine Position Stand. The female athlete triad. Med Sci Sports Exerc. 2007; 39 (10): 1867–82.
27. Nichols JF, Rauh MJ, Lawson MJ, Ji M, Barkai HS. Prevalence of the female athlete triad syndrome among high school athletes. Arch Pediatr Adolesc Med. 2006; 160 (2): 137–42.
28. Practice Committee of American Society for Reproductive Medicine. Current evalution of amenorrhea. Fertil Steril. 2008; 90 (5 Suppl): S219–25.
29. Quillman SM. Nutrition and Diet Therapy. 2nd ed. Springhouse (PA): Springhouse; 1994.
30. Rauh MJ, Nichols JF, Barrack MT. Relationships among injury and disordered eating, menstrual dysfunction, and low bone mineral density in high school athletes: a prospective study. J Athl Train. 2010; 45 (3): 243–52.
31. Rothman KJ, Greenland S. Modern Epidemiology. Philadelphia (PA): Lippincott Wiliams & Wilkins; 1998. p. 737.
32. Stunkard AJ, Messick S. The Three-Factor Eating Questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res. 1985; 29 (1): 71–83.
33. Torstveit MK, Sundgot-Borgen J. The female athlete triad exists in both elite athletes and controls. Med Sci Sports Exerc. 2005; 37 (9): 1449–59.
34. Torstveit MK, Sundgot-Borgen J. The female athlete triad: are elite athletes at increased risk? Med Sci Sports Exerc. 2005; 37 (2): 184–93.
35. Torstveit MK, Sundgot-Borgen J. Low bone mineral density is two to three times more prevalent in non-athletic premenopausal women than in elite athletes: a comprehensive controlled study. Br J Sports Med. 2005; 39 (5): 282–7; discussion-7.
36. Torstveit MK, Sundgot-Borgen J. Participation in leanness sports but not training volume is associated with menstrual dysfunction: a national survey of 1276 elite athletes and controls. Br J Sports Med. 2005; 39 (3): 141–7.
37. Vescovi JD, Scheid JL, Hontscharuk R, De Souza MJ. Cognitive dietary restraint: impact on bone, menstrual and metabolic status in young women. Physiol Behav. 2008; 95 (1–2): 48–55.
38. Welt CK, Chan JL, Bullen J, et al. Recombinant human leptin in women with hypothalamic amenorrhea. N Engl J Med. 2004; 351 (10): 987–97.
39. Zanker CL, Cooke CB, Truscott JG, Oldroyd B, Jacobs HS. Annual changes of bone density over 12 years in an amenorrheic athlete. Med Sci Sports Exerc. 2004; 36 (1): 137–42.
40. Zanker CL, Swaine IL. Relation between bone turnover, oestradiol, and energy balance in women distance runners. Br J Sports Med. 1998; 32 (2): 167–71.