INTRODUCTION
The Male and Female Athlete Triad (Triad) and Relative Energy Deficiency in Sport (REDs) are disorders defined by low energy availability, which is insufficient energy intake compared with energy expenditure.1–3 Low energy availability impacts hormone levels, bone mineral density (BMD), and bone stress injury (BSI) risk.4–6 Few studies have examined Triad risk in ultramarathon runners.7,8
We previously reported the prevalence of Triad and REDs risk factors in male and female ultramarathon runners using data collected at a 100-mile race in 2018 and 2019. We also reported exploratory analyses examining correlations between BMD and hormones. The goal of this analysis was to update our results to include data from the 2021 race to allow for a more robust analysis.
METHODS
Subjects
The study population consisted of male and female ultramarathon runners competing in the 2018, 2019, and 2021 Western States Endurance Run. Registered runners received a recruitment email 2 months before race day. Each participant provided informed written consent, and the study was approved by institutional review boards at University of California, Davis, in 2018 and Stanford University in 2019 and 2021.
Study Procedures
All study participants completed an online Qualtrics questionnaire that assessed training history, nutritional status, menstrual history (women only), and BSI history. The survey included questions from the Female and Male Athlete Triad Screening Questionnaire and the Eating Disorder Examination Questionnaire.
Participants completed dual energy x-ray absorptiometry (DXA) scans to measure BMD in the 2 days before the race using BodySpec (Culver City, CA), a mobile DXA company. We measured total body, lumbar spine (L1–L4), total hip, femoral neck, and distal one-third of the radius BMD; measurements were reported as Z-scores standardized by age, sex, and ethnicity.
To assess Triad risk, we used the Male and Female Athlete Triad Cumulative Risk Assessments (CRAs),1,3 which assign scores from 0 to 2 for each Triad component. The total CRA score is used to classify athletes into overall low-risk, moderate-risk, and high-risk categories.
The 2019 and 2021 cohorts also completed fasting blood draws the day before the race to measure 25-hydroxy vitamin D (25(OH)D), ferritin, total testosterone, free testosterone, and estradiol through InsideTracker (Cambridge, MA), a biotechnology company. Free testosterone was measured differently in 2019 and 2021 (direct vs calculated measurement) and was not updated in this report.
Statistical Analysis
We calculated descriptive statistics including means, SD, or Ns and percentages. Our exploratory analyses calculated Spearman correlation coefficients between biomarkers and BMD Z-scores in 28 statistical tests. Data were analyzed using SAS Software Version 9.4 (Cary, North Carolina). An anonymized subset of the data and the statistical code are available as supplements to this manuscript (see Supplemental Digital Content 1, https://links.lww.com/JSM/A402, Supplemental Digital Content 2, https://links.lww.com/JSM/A403).
RESULTS
Study Population
Our sample included 179 participants (123 men and 56 women); 161 completed DXA scans, and 102 completed laboratory draws. Table 1 presents participant characteristics.
Descriptive Characteristics of the Study Population at Baseline, Mean ± SD, or N (%)
| Characteristic | Women (n = 56)* | Men (n = 123)* |
| Age (yr) | 41.0 (8.1) | 46.6 (9.9) |
| Race/ethnicity (%) | ||
| White | 52 (92.9) | 108 (87.8) |
| Hispanic/Latino | 0 (0.0) | 6 (4.9) |
| Other | 4 (7.2) | 8 (6.5) |
| Missing | 0 (0.0) | 1 (0.8) |
| Country (%) | ||
| The United States | 52 (92.9) | 104 (84.6) |
| International | 3 (5.4) | 19 (15.4) |
| Missing | 1 (1.8) | 0 (0.0) |
| Running level (%) | ||
| Professional | 10 (17.9) | 2 (1.6) |
| Competitive age group | 22 (39.3) | 60 (48.8) |
| Middle of the pack | 19 (33.9) | 55 (44.7) |
| Finisher | 5 (8.9) | 6 (4.9) |
| Mileage (miles/week) | 59.6 (14.8) | 56.6 (18.9) |
| Triad cumulative risk score (%) | n = 2 missing | n = 13 missing |
| Low risk | 21 (38.9) | 79 (71.8) |
| Moderate risk | 29 (53.7) | 27 (24.5) |
| High risk | 4 (7.4) | 4 (3.6) |
| Energy availability risk score (%) | n = 0 missing | n = 3 missing |
| Low risk | 25 (44.6) | 74 (61.7) |
| Moderate risk | 21 (37.5) | 38 (31.7) |
| High risk | 10 (17.9) | 8 (6.7) |
| BMI risk score (%) | n = 4 missing | n = 12 missing |
| Low risk | 46 (88.5) | 111 (100.0) |
| Moderate risk | 5 (9.6) | 0 (0.0) |
| High risk | 1 (1.9) | 0 (0.0) |
| BSI risk score (%) | n = 0 missing | n = 1 missing |
| Low risk | 36 (64.3) | 95 (77.9) |
| Moderate risk | 9 (16.1) | 13 (10.7) |
| High risk | 11 (19.6) | 14 (11.5) |
| BMD risk score (%) | n = 4 missing | n = 11 missing |
| Low risk | 42 (80.8) | 79 (70.5) |
| Moderate risk | 10 (19.2) | 27 (24.1) |
| High risk | 0 (0.0) | 6 (5.4) |
| Oligomenorrhea risk score† (%) | n = 31 missing† | — |
| Low risk | 20 (80.0) | — |
| Moderate risk | 2 (8.0) | — |
| High risk | 3 (12.0) | — |
| Menarche risk score (%) | n = 0 missing | — |
| Low risk | 40 (71.4) | — |
| Moderate risk | 8 (14.3) | — |
| High risk | 8 (14.3) | — |
| 25 (OH)D, ng/mL‡ | 35.9 (15.5) | 35.0 (13.3) |
| Total testosterone, ng/dL§ | 29.1 (10.7) | 615.4 (167.3) |
| Estradiol, pg/mL‡,¶ | 89.4 (104.2) | 33.3 (9.3) |
*4 women and 1 man participated in the study in 2018 and 2019. We only counted and included data from 2019 for repeat participants.
†Excludes women who are postmenopausal (n = 6) or on hormonal contraceptives (n = 25).
‡Data not collected in 2018. n = 31 women, n = 71 men.
§Data not collected in 2018. n = 30 women (n = 1 missing due to value representative of pathology), n = 70 men (n = 1 missing due to value representative of pathology).
¶Includes women who are postmenopausal (n = 6) or on hormonal contraceptives (n = 25) due to the importance of these values for calculating BMD correlations.
Updated Triad Prevalence
Table 1 presents the updated Triad risk prevalences. Energy availability risk was high: 55.4% of women and 38.4% of men were at moderate risk or high risk for low energy availability. Whereas BMD risk was lower: 19.2% of women and 29.5% of men were at moderate risk or high risk for low BMD. Overall, more than half of women (61.1%) and one-quarter of men (28.1%) were at moderate or high cumulative risk for the Triad (compared with 66.7% of women and 34.8% of men in our prior publication).
Exploratory Analyses
In women, we found significant correlations between hip BMD and total testosterone (Spearman correlation coefficient, r = 0.58, P = 0.001) as well as femoral neck BMD and total testosterone (r = 0.45, P = 0.01) (Table 2 and Figure 1). No other BMD-hormone associations achieved statistical significance.
Spearman Correlation Coefficients Between BMD Z-Scores and Endocrine Biomarkers for 2019 and 2021 Cohorts
| Biomarker By Gender | Spine Z-Score | Total Hip Z-Score | Femoral Neck Z-Score | Forearm Z-Score | ||||
| r |
P |
r |
P |
r |
P |
r |
P |
|
| Men (n = 70)*,† | ||||||||
| Estradiol | 0.23 | 0.06 | 0.24 | 0.04 | 0.27 | 0.03 | 0.30 | 0.01 |
| Testosterone | 0.05 | 0.72 | −0.06 | 0.61 | −0.01 | 0.95 | 0.03 | 0.82 |
| Vitamin D | 0.07 | 0.56 | 0.18 | 0.13 | 0.27 | 0.03 | 0.24 | 0.048 |
| Women (n = 31)* | ||||||||
| Estradiol | 0.26 | 0.16 | 0.18 | 0.34 | 0.08 | 0.65 | 0.09 | 0.64 |
| Testosterone | 0.11 | 0.57 | 0.58 | 0.001 | 0.45 | 0.01 | 0.19 | 0.31 |
| Vitamin D | 0.00 | 0.99 | 0.04 | 0.82 | 0.04 | 0.83 | 0.01 | 0.97 |
Correlation coefficients are not adjusted for age given that Z-score accounts for age.
*Excludes 1 man and 1 woman who had total testosterone levels indicative of pathology via reference range by sex.
†Note that men have missing BMD values for spine (n = 2), femoral hip (n = 1), and forearm (n = 1).
Spearman correlation coefficients between BMD Z-scores and endocrine biomarkers for 2019 and 2021 cohorts in women and men. Significant correlations (P A. Total hip and testosterone in women (n = 31). B. Femoral neck and testosterone in women (n = 31). C. Femoral neck and estradiol in men (n = 70). D. Forearm and estradiol in men (n = 70).
In men, we found significant correlations of small-to-moderate size (r = 0.24-0.30, P = 0.01–0.04) between estradiol and BMD at almost all anatomic sites, as well as between vitamin D and forearm and femoral neck BMD (Table 2 and Figure 1). No other BMD-hormone associations achieved statistical significance.
DISCUSSION
We found that 61% of women and 28% of men were at moderate Triad risk or high Triad risk; though these estimates are lower than our previous estimates of 67% and 35%, some shift is expected due to the wide margin of error on the original estimates. With the addition of about 50% more study participants in 2021, our updated numbers reflect a more precise prevalence measurement (95% CI for women: 0.47–0.74 and for men: 0.19–0.36). This article also provides more precise prevalence estimates for individual components of the Triad.
As expected for exploratory analyses, which are not expected to be robust, our results differed considerably from our earlier publication. In the first publication, in women, we found significant correlations between total hip BMD and total testosterone (r = 0.71), and between spine BMD and estradiol (r = 0.52). In our updated analyses, the significant correlation between hip BMD and total testosterone persisted but was attenuated (r = 0.58), and the other associations did not persist. In men, we found no significant associations in the earlier publication, but here, we found weak-to-moderate significant correlations between estradiol and BMD at almost all anatomic sites. Prior studies show relationships between estradiol and BMD in women9 and testosterone and BMD in men,10 whereas our study has attenuated associations in testosterone and BMD in women and weak-to-moderate associations in estradiol and BMD in men. Our results highlight limitations of exploratory analyses.
In our earlier publication, we clearly labeled the exploratory results as “exploratory” to signal to readers that our results may not be robust. We also provided data visualizations of the significant relationships; these visuals help readers to visually assess the robustness (or lack of robustness) of relationships. We recommend that authors publishing exploratory results follow these steps to prevent readers from overinterpreting results. Additional ways to improve the transparency of exploratory analyses include stating the number of statistical tests run or using multiple testing corrections.
CONCLUSION
Our results expand on our 2022 CJSM publication.8 As before, we observed a high prevalence of Triad risk factors, which highlights the need for further study of Male and Female Athlete Triad in ultramarathon runners, a group that has been understudied relative to high school and college runners. We found some significant associations between hormones and BMD in both men and women, but these should be interpreted cautiously given the unstable nature of exploratory analyses. Finally, we highlight the importance of transparency when reporting the results of exploratory analyses.
References