Dose-Response Modelling of Resistance Exercise Across Outcome Domains in Strength and Conditioning: A Meta-analysis

Abstract

Background: Resistance exercise is the most common training modality included within strength and conditioning (S&C) practice. Understanding dose-response relationships between resistance training and a range of outcomes relevant to physical and sporting performance is of primary importance for quality S&C prescription.

Objectives: The aim of this meta-analysis was to use contemporary modelling techniques to investigate resistance-only and resistance-dominant training interventions, and explore relationships between training variables (frequency, volume, intensity), participant characteristics (training status, sex), and improvements across a range of outcome domains including maximum strength, power, vertical jump, change of direction, and sprinting performance.

Methods: Data were obtained from a database of training studies conducted between 1962 and 2018, which comprised healthy trained or untrained adults engaged in resistance-only or resistance-dominant interventions. Studies were not required to include a control group. Standardized mean difference effect sizes were calculated and interventions categorized according to a range of training variables describing frequency (number of sessions per week), volume (number of sets and repetitions performed), overall intensity (intensity of effort and load, categorised as low, medium or high), and intensity of load (represented as % of one-repetition maximum [1RM] prescribed). Contemporary modelling techniques including Bayesian mixed-effects meta-analytic models were fitted to investigate linear and non-linear dose-responses with models compared based on predictive accuracy.

Results: Data from a total of 295 studies comprising 535 groups and 6,710 participants were included with analyses conducted on time points ≤ 26 weeks. The best performing model included: duration from baseline, average number of sets, and the main and interaction effects between outcome domain and intensity of load (% 1RM) expressed non-linearly. Model performance was not improved by the inclusion of participant training status or sex.

Conclusions: The current meta-analysis represents the most comprehensive investigation of dose-response relationships across a range of outcome domains commonly targeted within strength and conditioning to date. Results demonstrate the magnitude of improvements is predominantly influenced by training intensity of load and the outcome measured. When considering the effects of intensity as a % 1RM, profiles differ across outcome domains with maximum strength likely to be maximised with the heaviest loads, vertical jump performance likely to be maximised with relatively light loads (~ 30% 1RM), and power likely to be maximised with low to moderate loads (40-70% 1RM).

Fig. 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram for included studies. S&C strength and conditioning

Fig. 2

(Top): Posterior distributions of pooled mean effect sizes across outcome domains. (Bottom): Values represent shrunken values after fitting meta-analytic model also accounting for time of measurement. Black lines represent 75% and 95% credible intervals

Fig. 3

(Top): Marginal effects of smooth terms illustrating interactions between intensity of load expressed as percentage of maximum and outcome domain. (Bottom): Density plots at the top of each figure illustrate the distribution of the load intensity variable for the given outcome domain. Solid lines represent the best estimate of the smooth relationship and shaded regions represent intervals of uncertainty (75% and 95%)