Implementation of multiple statistical methods to estimate variability and individual response to training

Multiple statistical methods have been proposed to estimate individual responses to exercise training; yet, the evaluation of these methods is lacking. We compared five of these methods including the following: the use of a control group, a control period, repeated testing during an intervention, a reliability trial and a repeated intervention. Apparently healthy males from the Gene SMART study completed a 4-week control period, 4 weeks of High-Intensity Interval Training (HIIT), >1 year of washout, and then subsequently repeated the same 4 weeks of HIIT, followed by an additional 8 weeks of HIIT. Aerobic fitness measurements were measured in duplicates at each time point. We found that the control group and control period were not intended to measure the degree to which individuals responded to training, but rather estimated whether individual responses to training can be detected with the current exercise protocol. After a repeated intervention, individual responses to 4 weeks of HIIT were not consistent, whereas repeated testing during the 12-week-long intervention was able to capture individual responses to HIIT. The reliability trial should not be used to study individual responses, rather should be used to classify participants as responders with a certain level of confidence. 12 weeks of HIIT with repeated testing during the intervention is sufficient and cost-effective to measure individual responses to exercise training since it allows for a confident estimate of an individual's true response. Our study has significant implications for how to improve the design of exercise studies to accurately estimate individual responses to exercise training interventions.HighlightsWhat are the findings? We implemented five statistical methods in a single study to estimate the magnitude of within-subject variability and quantify responses to exercise training at the individual level.The various proposed methods used to estimate individual responses to training provide different types of information and rely on different assumptions that are difficult to test.Within-subject variability is often large in magnitude, and as such, should be systematically evaluated and carefully considered in future studies to successfully estimate individual responses to training.How might it impact on clinical practice in the future?Within-subject variability in response to exercise training is a key factor that must be considered in order to obtain a reproducible measurement of individual responses to exercise training. This is akin to ensuring data are reproducible for each subject.Our findings provide guidelines for future exercise training studies to ensure results are reproducible within participants and to minimise wasting precious research resources.By implementing five suggested methods to estimate individual responses to training, we highlight their feasibility, strengths, weaknesses and costs, for researchers to make the best decision on how to accurately measure individual responses to exercise training.