Author(s): MACPHERSON, T., GRAHAM, M., MCLAREN, S., ATKINSON, G., WESTON, M., Institution: NORTHUMBRIA UNIVERSITY, Country: UNITED KINGDOM, Abstract-ID: 2452

INTRODUCTION:
Despite the potential of dRPE to provide a more precise appraisal of exercise exertion, questions exist of its usefulness and worthwhileness due to lack of differences between the two constructs. We aimed to quantify the magnitude of the difference between RPE-P and RPE-C (P–Cdifference) and to explore the moderating influence of methodological factors that could impact P–Cdifference.
METHODS:
In accordance with PRISMA guidelines. we searched five academic databases for original research articles published up to October 2023 and available in English. A Boolean search phrase was created to include search terms relevant to internal load / ratings of perceived exertion (19 keywords), dRPE (20 keywords) and exercise/ training sessions (60 keywords). We obtained the mean difference between peripheral and central RPE, as well as the standard deviation of the difference, by extracting individual data and calculating the mean difference in ratings manually (i.e. peripheral exertion score minus central exertion score), allowing dominance between the two perceptual ratings to be assessed. The final data sample included 94 datasets, from 70 individual studies and included 2167 participants from a mixture of activity levels. Random effects meta-analyses were conducted on each dataset to determine the extent of the P–Cdifference. To explore moderator effects of the P–Cdifference, we utilised subgroup comparison analysis when two of more studies were available for analysis. The categorical moderators that we used were study population, exercise modality and whether familiarisation was reported in the study. Magnitude-based decisions provided interpretation of the P–Cdifference in relation to a threshold of 10% of the mean dRPE scores (0.6 arbitrary units [au]). A P–Cdifference was only declared clear if the probability for the difference was ≥ 95% (i.e., very likely).
RESULTS:
Random effects meta-analysis revealed no clear rating dominance between RPE-P and RPE-C (0.6 Category–Ratio 10 arbitrary units [au]; 95%CI: 0.4 to 0.7). Moderator analysis showed study population explained 27% of the variance in the P–Cdifference (τ = 0.76, r² = 0.27), with RPE-P clearly dominant in disabled (1.3 au; 95%CI 0.7 to 1.8 au) and normal/ healthy (0.9 au; 95%CI 0.5 to 1.3 au) participants when compared to RPE-C. Exercise mode explained 13% of the heterogeneity of the P–Cdifference (τ = 0.76, r² = 0.13), with RPE-P clearly dominating during arm crank (2.1 au; 95%CI 1.4 to 2.8 au) and cycling (0.8 au; 95%CI 0.5 to 1.1 au) exercise when compared to RPE-C. Familiarisation failed to explain the heterogeneity of the P–Cdifference (τ = 0.76, r² = 0.00), although when studies reported habituation to RPE procedures, RPE-P was dominant (0.8 au; 95%CI 0.6 to 1.0 au).
CONCLUSION:
We found the pooled effect to show no clear dominance for either RPE-P or RPE-C but our moderator analysis helped to explain the variance in P–Cdifference and thus supporting dRPE to be sensitive in detecting methodological changes.