The influence of baseline capability on intervention effects in strength and conditioning

A review of concepts and methods with meta-analysis

Abstract

Background: In strength and conditioning (S&C) it is commonly believed that baseline capability influences response to an intervention, such that in general, those with the higher baseline values experience reduced change. Such differences are referred to as intervention differential effects (IDE) and are important in the study of tailoring training programs in S&C. There are, however, several conceptual and technical issues that present a challenge when investigating whether baseline capability causes IDE. The present review provides an overview of these conceptual and technical issues, highlighting important differences between changes within and between populations, and the role of measurement error and subsequent regression to the mean when performing standard analyses.

Methods: The present review also includes a meta-analysis to explore more generally, whether those with higher baseline values experience reduced change. Baseline and post-intervention standard deviations were extracted from 421 S&C training studies including the 1RM squat (121 studies; 329 outcomes), 1RM bench press (103 studies; 307 outcomes), vertical jump (312 studies; 896 outcomes), 10 m sprint time (95 studies; 194 outcomes), 20 m sprint time (97 studies; 193 outcomes), and 30 m sprint time (58 studies; 118 outcomes). For each outcome, a Bayesian three-level hierarchical meta-analysis model was conducted to estimate the pooled mean difference of the standard deviations. Where results indicated that the post-intervention standard deviation was equal to, or less than the baseline standard deviation, this was interpreted as evidence of a negative relationship between baseline and change values. Where results indicated a greater post-intervention standard deviation, this was judged as indeterminate due to the potential for random variation in intervention effects to increase the post-intervention standard deviation.

Results: Moderate evidence was obtained for a reduction in standard deviation post-intervention for the vertical jump (Difference_0.5 = -0.07 [95%CrI: -0.16 to 0.02 cm]; p(Difference <0)= 0.933); and strong evidence for the same with sprint time across all three distances (10 m: Difference_0.5 = -0.007 [95%CrI: -0.012 to -0.003 s]; p(Difference <0)>0.999; 20 m: Difference_0.5 = -0.020 [95%CrI: -0.034 to -0.009 s]; p(Difference <0)>0.999; 30 m: Difference_0.5 = -0.011 [95%CrI: -0.020 to -0.002 s]; p(Difference <0)=0.992). In contrast, strong evidence was obtained for an increase in post-intervention standard deviation for the 1RM squat (Difference_0.5 = 0.93 [95%CrI: 0.52 to 1.34 kg]; p(Difference <0)<0.001) and bench press (Difference_0.5 = 0.69 [95%CrI: 0.40 to 0.98 kg]; p(Difference <0)<0.001).

Conclusion: Collectively, the results present evidence for a negative IDE of baseline capability for sprint and vertical jump performance, but not maximum strength. Further research that is cognisant of the conceptual and analytical challenges in determining if baseline capability causes IDE is required, including the contexts and populations in S&C which may alter the interactions.