Materials and methods
Participants
Forty-five adult male competitive cyclists (equivalent to British Cycling [BC] category 2 or above) who had participated in data collection early in the road race season3 returned at the end of the racing season during September and October.
Allocation of cyclists to intervention/no intervention groups
In the early race season, cyclists were matched, in pairs, based on Z-scores for lumbar spine BMD. One member of each pair was randomly allocated to receive educational interventions from the lead clinician. Technicians performing the dual-energy X-ray absorptiometry (DXA) scans were blinded to the allocation.
Educational nutrition and skeletal loading intervention
The nutritional advice (see online supplementary file 3) aimed to maintain adequate energy availability, with recommendations for general nutrition and fuelling around training sessions. These recommendations had been approved by registered clinical sports dietitians. The skeletal loading exercises (see online supplementary file 2) were designed to be practical, based on current recommendations to improve bone health,8 with input from qualified Pilates teachers and physiotherapists working with cyclists. For cyclists in the intervention group, both the nutrition and exercise recommendations were fully explained and written sheets provided. Online resources included the sheets and a video demonstrating the exercises. Initially weekly, then monthly, these cyclists were contacted to check on progress. All cyclists were asked to record any sustained changes in off-bike exercise and nutrition, both baseline and fuelling around training.
Sport-specific Energy Availability Questionnaire and Interview
The clinical assessment of energy availability by SEAQ-I made in our published baseline early race season study3 was significantly associated with lumbar spine BMD Z-score in this group of cyclists. We used a similar approach to devise a follow-up SEAQ-I with input from the same clinical sports endocrinologist, sports research scientist, registered clinical sports dietitians, cyclists and coaches for validation of content (see online supplementary file 1).
The follow-up SEAQ-I was conducted at the end of the race season. After the cyclists had completed the questionnaire, the sports clinician interviewed them individually to verify answers and gather more details on the responses provided, including the following:
Cycling training, race results and quantification of any changes in skeletal loading exercises performed, as per recommendations, or otherwise.
Nutritional information: any changes in baseline nutrition, fuelling around training, as per recommendations, or otherwise; intention to change body weight/composition; addition of any supplements.
Medical history during study period: number, nature and site of injuries; number of days off training due to illness.
From the follow-up SEAQ-I, cyclists were assessed in each of the areas of nutrition and skeletal loading as implementing either positive change, no change or negative change. Qualitative information was gathered on reasons for cyclists being unable to adhere to the recommendations for their allocated group.
Sport-specific performance measures
Cyclists recorded the number of BC race points won during the study period, together with any race highlights. The number of points won was verified from the BC website. For those competing outside of BC races (eg, Commonwealth Games, European Championships and time trial events), an equivalent number of BC points was estimated, to reflect race performance over the season. Cyclists also reported their 60 minute functional threshold power (FTP) in watts.
Bone health measures
Body weight was measured to the nearest 0.1 kg using calibrated electronic scales (Seca Alpha, Birmingham UK), and standing height was measured to the nearest 0.1 cm using a stadiometer (Seca Alpha) with head in Frankfurt plane. Bone health and body composition were evaluated using DXA (GE Lunar iDXA, GE Healthcare, UK) according to best practice recommendations for densitometry in athletes, and the same trained, International Society for Clinical Densitometry (ISCD)-certified densitometrist interpreted all examinations to ensure consistency of region of interest placement between baseline and follow-up measures.9 BMD was evaluated at the anterior-posterior lumbar spine (L1–L4) and femoral neck. Age-matched BMD Z-scores were derived for each cyclist, at each skeletal site by the DXA software, using UK reference population data (GE Lunar Encore V.15.0, GE Healthcare, Madison, Wisconsin). Precision estimates (coefficient of variation) are 0.4% for lumbar spine BMD and 0.9% for femoral neck BMD.10 Body composition was derived from a total body scan with precision estimates being 0.5%–0.9%.11
Endocrine health measures
Endocrine and metabolic markers were assessed from capillary blood samples taken in the morning after waking to minimise diurnal variation. Samples were analysed to determine concentrations of total testosterone, vitamin D (25-hydroxy), free triiodothyronine, albumin, calcium, corrected calcium and alkaline phosphatase at Surrey University-accredited laboratories using cobas 8000 analyser with interassay coefficient of variation from <2% to 7% for the markers above. Absolute mean values with SD were determined and results were also expressed as Z-scores, using population mean and SD derived from the definition of the reference range, as covering 95% of a normal distribution.
Statistical analysis
Analyses were performed using open source packages Orange12 (Bioinformatics Lab at the University of Ljubljana, Slovenia) and SciPy (Enthought, Austin, Texas, USA). The data set included categorical and continuous observations, taken from the follow-up SEAQ-I, blood markers and DXA results. The means and SD of continuous variables were evaluated and, where relevant, compared against appropriate population reference ranges.
Explanatory analyses identified attributes associated with target variables relating to changes in bone health and to cycling performance over the racing season. Since experimental interventions were designed to benefit lumbar spine, the change in lumbar BMD was the target variable for bone health. The target variable for cycling performance was the number of BC points attained.
In accordance with the ISCD recommendations,13 the precision error of DXA measurements was taken into account in assessing changes in BMD. A change in BMD of an individual was considered to be meaningful if exceeding the least significant change (LSC), defined as 2.77 times precision error.9 Thus the LSC for lumbar spine BMD was 1.1%.
Changes were considered only for those of the original 50 participants who attended the second round of DXA scans and provided a second set of blood results. The resulting set of paired samples was analysed using the paired sample t-test, after checking for normality using the D’Agostino and Pearson test. The equivalence of means of multiple subgroups was tested by analysis of variance. The significance of the regression coefficients between continuous variables was based on the t-statistic.
Patient and public involvement
The research, clinical and support pathways for male athletes at risk of RED-S are lacking compared with provision for female athletes. During a pilot study of cyclists, these issues were discussed as SEAQ-I was trialled and refined. Male cyclists were instrumental in the initiation of this current study through involvement of coaches, and recruitment of team-mates and riders from other teams. A male cyclist in this study diagnosed with osteoporosis due to RED-S wrote a patient voice piece for British Journal of Sports Medicine. Other cyclists from the study experiencing consequences of RED-S contributed to articles in cycling magazines and to the TrainBrave campaign to raise awareness. Cyclists and coaches are supportive of and providing input for an educational website on RED-S (www.Health4Performance.co.uk) backed by the British Association of Sport and Exercise Medicine, which has been developed by the authors of this study.