Bra knowledge – the primary outcome – was measured using a custom-designed, 50-item, self-administered questionnaire. Details of the questions

which covered bra design, bra component parts, bra sizing, as well as correct and incorrect bra fit and bra wearing habits, can be found in Appendix 1 (see eAddenda for Appendix 1). Responses included multiple choice options, true/false, and short answers; an ‘I do not know’ response was offered for every question. Face validity was verified through focus groups. Bra fit was measured using the Bra Fit Assessment test (Choice Magazine 2005) as pass/fail. To be ranked a pass, the front band had to be in contact with the sternum; the posterior and side band had to have no flesh bulging above its superior edge (too small) and was not EGFR inhibitor to move upward if the arms were raised above the head three times (too big); the cup had to have no aspect of the breast bulging above its superior

or medial edge (too small) and no wrinkles in the cup material (too big); the straps were not to be digging into (too small) or slipping off (too big) the shoulders; and the cup underwire had to be resting on the ribs and sternum, not on any breast tissue. If one or more of these six components were ranked a ‘fail’ grade in fit, and the straps or the band could not be adjusted by the assessor to achieve correct fit, an overall ‘fail’ grade was awarded in BMS-354825 in vitro the Bra Fit Assessment test. Level of breast support was measured using the Level of Breast Support test as pass/fail. To be ranked a pass for design, the bra had to be a sports bra, or any two bra combination for any bra size, or a crop top only for cup sizes A or B. Lifespan was ranked

a fail (too old) if the material/elastic or underwire of any bra, of any design, had deteriorated. Both bra design and lifespan had to pass for an overall ranking of pass in the Level of Breast Support test. Discomfort during exercise was measured using a 10-cm visual analogue isothipendyl scale where participants were asked to rate their breast discomfort when wearing this bra during sport. Bra knowledge was calculated as the mean (SD) percentage of correct answers, while lack of bra knowledge was calculated as the mean (SD) percentage of ‘I do not know’ answers. Number of participants passing the Bra Fit Assessment and Level of Breast Support tests was reported. Analysis was by intention-to-treat, whereby all participants were analysed in the groups that they were randomised to and all available data were included in the analysis. Statistical significance was set at p < 0.05, so mean difference (95% CI) or risk difference (95% CI) between groups are presented. Four sporting academies agreed to participate. Three academies declined due to time constraints of their teams and coaches.