Brooke R Brisbine1,2, Julie R Steele1, Elissa J Phillips3, Deirdre E McGhee1. 1. the Biomechanics Research Laboratory, School of Medicine, Faculty of Science, Medicine and Health, University of Wollongong, Australia. 2. Applied Technology & Innovation, Australian Institute of Sport, Canberra, Australia. 3. Performance Networks & Partnerships, Australian Institute of Sport, Canberra, Australia.
Abstract
OBJECTIVE: To evaluate whether a simple 4-factor model using self-reported data could be used to predict exercise-induced breast pain in elite female athletes. DESIGN: Survey study. SETTING: Online or hard-copy surveys completed at sporting competitions and training facilities around Australia. PARTICIPANTS: Four hundred ninety female athletes competing nationally or internationally across 49 sports. INDEPENDENT VARIABLES: A binomial logistic regression analysis was used to evaluate the strength of a predictive model that included 2 continuous independent variables (age and body mass index) and 2 binary independent variables (breast size and sports bra use). Odds ratios were also calculated to determine the likelihood of an athlete reporting exercise-induced breast pain in association with each of the 4 variables. MAIN OUTCOME MEASURES: Exercise-induced breast pain was the binary dependent variable. RESULTS: The model incorporating athlete age, breast size, body mass index, and sports bra use was found to be statistically significant, but weak, in its ability to predict exercise-induced breast pain in elite female athletes (correctly identified 66% of athletes). For every 1-year increase in age, a significant 2.6% increase in the likelihood of experiencing exercise-induced breast pain was observed. Athletes with medium-to-hypertrophic sized breasts were 5.5 times more likely to experience exercise-induced breast pain than athletes with small breasts. CONCLUSIONS: Although the current model was not sensitive enough for use by clinicians and coaches, age and breast size were both identified as critical variables in the prediction of exercise-induced breast pain. Future research is encouraged to investigate whether incorporating additional variables such body fat percentage, bra fit, and other relevant factors can add strength to the model.
OBJECTIVE: To evaluate whether a simple 4-factor model using self-reported data could be used to predict exercise-induced breast pain in elite female athletes. DESIGN: Survey study. SETTING: Online or hard-copy surveys completed at sporting competitions and training facilities around Australia. PARTICIPANTS: Four hundred ninety female athletes competing nationally or internationally across 49 sports. INDEPENDENT VARIABLES: A binomial logistic regression analysis was used to evaluate the strength of a predictive model that included 2 continuous independent variables (age and body mass index) and 2 binary independent variables (breast size and sports bra use). Odds ratios were also calculated to determine the likelihood of an athlete reporting exercise-induced breast pain in association with each of the 4 variables. MAIN OUTCOME MEASURES: Exercise-induced breast pain was the binary dependent variable. RESULTS: The model incorporating athlete age, breast size, body mass index, and sports bra use was found to be statistically significant, but weak, in its ability to predict exercise-induced breast pain in elite female athletes (correctly identified 66% of athletes). For every 1-year increase in age, a significant 2.6% increase in the likelihood of experiencing exercise-induced breast pain was observed. Athletes with medium-to-hypertrophic sized breasts were 5.5 times more likely to experience exercise-induced breast pain than athletes with small breasts. CONCLUSIONS: Although the current model was not sensitive enough for use by clinicians and coaches, age and breast size were both identified as critical variables in the prediction of exercise-induced breast pain. Future research is encouraged to investigate whether incorporating additional variables such body fat percentage, bra fit, and other relevant factors can add strength to the model.