Experimental verification of regression to the mean in redox biology: differential responses to exercise

Abstract

An important methodological threat when selecting individuals based on initial values for a given trait is the “regression to the mean” artifact. This artifact appears when a group with an extreme mean value during a first measurement tends to obtain a less extreme value (i.e. tends toward the mean) on a subsequent measurement. The main aim was to experimentally confirm the presence of this artifact in the responses of the reference oxidative stress biomarker (F2-isoprostanes) after exercise. Urine samples were collected before and immediately following acute exercise in order to determine the level of exercise-induced oxidative stress. Afterwards, participants were arranged into three groups based on their levels of exercise-induced oxidative stress (low, moderate and high oxidative stress groups; n = 12 per group). In order to verify the existence of the regression to the mean artifact, the three groups were subjected to a second exercise trial one week after the first trial. This study confirmed the regression to the mean artifact in a redox biology context and showed that this artifact can be minimized by performing a duplicate pretreatment measurement after completing a nonrandom sorting based on the first assessment. This study also indicated that different individuals experience high oxidative stress or reductive stress (or no stress) to the same exercise stimulus even after adjusting for regression to the mean. This finding substantiates the methodological choice to divide individuals based on their degree of exercise-induced oxidative stress in future experiments to investigate the role of reactive species in exercise adaptations.