Title screen

This study aimed to evaluate whether similar associations between indexes derived from heart rate variability (HRV) analyses and training load (TL) could be obtained by using the commonly used Pearson correlation technique and the repeated measures correlation (rmcorr). Fourteen well-trained swimmers (18.5 ± 1.6 years) participated. The training period lasted 4 weeks with a gradual increase in TL. Daily external TL (exTL) and internal TL (inTL) were summed to obtain a weekly TL, and HRV analyses were performed every Saturday morning. During the 4-week period, exTL and inTL increased (p < 0.05) together with a decrease (p < 0.05) in heart rate and an increase (p < 0.05) of cardiac parasympathetic indexes. No significant correlation was found using Pearson correlation while significant associations were found using rmcorr; considering exTL, positive (mean R-R interval [MeanRR], root mean square of differences between successive RR interval [RMSSD], low frequency [LF], high frequency [HF], instantaneous beat-to-beat variability [SD1], continuous beat-to-beat variability [SD2], SD1/SD2; r from 0.59 to 0.46, p value from <0.001 to 0.002) and negative (mean heart rate [meanHR]; r = −0.55, p < 0.001) associations were found. Considering inTL, positive (MeanRR, RMSSD, LF, HF, HFnu, SD1, SD2, SD1/SD2; r from 0.56 to 0.34, p-value from <0.001 to 0.025) and negative (meanHR, LFnu, LF/HF; r from −0.49 to −0.34, p value from 0.001 to 0.025) associations were found. The rmcorr statistical method was able to show associations between parasympathetic indexes and TL contrary to Pearson correlation analysis. Because rmcorr is specifically designed to investigate within-individual association for paired measures assessed on 2 or more occasions for multiple individuals, it should constitute a tool for future training monitoring researches based on a repeated-measures protocol.