Construct Validity, Test-Retest Reliability, and Repeatability of Performance Variables Using a Flywheel Resistance Training Device

Power production is highly associated with physical performance; however, the ability to quantitatively measure power output during resistance exercise is lacking. The purpose of this study was to determine the validity and test-retest reliability of flywheel-based performance testing. Twelve young, resistance trained subjects completed 2 bouts of resistance exercise using a flywheel resistance training device (Exxentric kbox 4 Pro). Each session consisted of 3 sets of 3 exercise (bent-over row, Romanian deadlift, and biceps curl) with varying moments of inertia (0.050, 0.075, and 0.100 kg·m², respectively) in random order. Each set consisted of 5 maximal effort repetitions with 3-minute recovery between sets. Average power, peak concentric and eccentric power, average force, average speed, and total work for each set were recorded. Regression analysis revealed a near-perfect relationship between measured and predicted power, force, and work at given workloads. Pearson's r between trials 1 and 2 revealed good (≥0.70) to excellent (≥0.90) test-retest reliability for all outcomes with the exception of peak eccentric power for biceps curls (r = 0.69), which narrowly missed the cutoff for acceptable reliability. Bland-Altman plots revealed small (approximately 5-15%), but statistically significant bias between the 2 trials for some measures. Coefficient of repeatability for all outcomes was relatively high, indicating poor repeatability. Flywheel-based performance testing provides valid data. However, reliability varies between individual lifts and specific outcomes. Given the poor repeatability between trials, it is likely that subjects who are unaccustomed to this modality may require multiple testing sessions or a thorough familiarization period to ensure accurate measures of power, force, speed, and work during flywheel-based performance testing.