Using historical controls to evaluate resistance training-induced strength adaptations: Does history repeat itself?

To determine whether a historical control group could appropriately quantify resistance training-induced changes in one-repetition maximum (1RM) strength and the magnitude of variability in training-induced 1RM changes, we reanalyzed data from three different studies. Statistically significant 1RM strength increases were observed for three high-load training groups compared to a time-matched, non-exercise control and historical control group (all p < 0.001). However, effect sizes derived from historical control comparisons were biased [e.g. mean difference (95%CI): 2.3 (1.6, 3.0) and 1.8 (1.1, 2.5) kg versus time-matched, non-exercise and historical controls, respectively]. To illustrate how future research might implement historical controls, we reanalyzed a study using time-matched, non-exercise control groups from two separate studies as ‘historical’ controls. Comparisons following high-load training were statistically consistent, whereas low-load training showed mixed results (p < 0.001 versus p = 0.244). Likewise, when reevaluating inter-individual variability in strength gains, Levene’s tests lacked consistency (p = 0.144 versus p = 0.022), and estimates of true inter-individual variation differed based on the control group [e.g. 0.49 (−0.71, 0.71) versus 1.04 (0.73, 1.27) kg]. Our results suggest that historical controls may be appropriate for evaluating large, group-level effects (i.e. task-specific strength changes following high-load training), but lack consistency for low-load training interventions and assessing inter-individual variability in 1RM strength responses.