A comparison of different methods to analyse data collected during time-to-exhaustion tests

Purpose: Despite their widespread use in exercise physiology, time-to-exhaustion (TTE) tests present an often-overlooked challenge to researchers, which is how to computationally deal with between- and within-subject differences in exercise duration. We aimed to verify the best analysis method to overcome this problem. Methods: Eleven cyclists performed an incremental test and three TTE tests differing in workload as preliminary tests. The TTEs were used to derive the individual power–duration relationship needed to set the workload (corresponding to an estimated TTE of 1200 s) for four identical experimental TTE tests. Within individuals, the four tests were subsequently rank ordered by performance. Physiological and psychological variables expected to change with performance were analysed using different methods, with the main aim being to compare the traditional “group isotime” method and a less-used “individual isotime” method. Results: The four tests, ranked from the best to the worst, had a TTE of 1526 ± 332, 1425 ± 313, 1295 ± 325, and 1026 ± 265 s. Ratings of perceived exertion, minute ventilation, respiratory frequency, and affective valence were sensitive to changes in performance when their responses were analysed with the “individual isotime” method (P < 0.022, ηp 2 > 0.144) but not when using the “group isotime” method, because the latter resulted in partial data loss. Conclusions: The use of the “individual isotime” method is strongly encouraged to avoid the misinterpretation of the phenomenon under study. Important implications are not limited to constant-workload exercise, but extend to incremental exercise, which is another commonly used test of exercise tolerance. © 2019, The Author(s).