Statistical analysis

All data was analysed using R studio version 4.0.2 and conducted on an intention-to-treat basis, which included all randomly assigned participants. Linear mixed models were used to determine the effect of exercise intensity (group) over time and to determine the interaction between timepoint and group (between-group differences). To examine the within group changes, we calculated the estimated marginal means from the linear mixed models. The unique participant codes were used as a random effect to account for repeated measures. All models were adjusted for age. Simple linear regressions were used to determine the association between VO_2peak, insulin sensitivity or body composition, and mental health or HRQoL outcomes using the pooled data from all participants. Data are presented as mean ± SD and median and percentiles for boxplots. P values were deemed statistically significant when < 0.05. The following packages were used in our analysis; lme4⁴⁷, lmerTest⁴⁸ and tidyverse⁴⁹. The estimation of sample size was based on the primary outcome of VO_2peak, based on this, we aimed to recruit 30 participants³⁷. As this was a secondary analysis, power calculations were not done for the outcome measures reported in this study.