To assess measurement sensitivity and diagnostic characteristics of athlete-monitoring tools to identify performance change.
Fourteen nationally competitive swimmers (11 male, 3 female; age 21.2 ± 3.2 y) recorded daily monitoring over 15 mo. The self-report group (n = 7) reported general health, energy levels, motivation, stress, recovery, soreness, and wellness. The combined group (n = 7) recorded sleep quality, perceived fatigue, total quality recovery (TQR), and heart-rate variability. The week-to-week change in mean weekly values was presented as coefficient of variance (CV%). Reliability was assessed on 3 occasions and expressed as the typical error CV%. Week-to-week change was divided by the reliability of each measure to calculate the signal-to-noise ratio. The diagnostic characteristics for both groups were assessed with receiver-operating-curve analysis, where area under the curve (AUC), Youden index, sensitivity, and specificity of measures were reported. A minimum AUC of .70 and lower confidence interval (CI) >.50 classified a “good” diagnostic tool to assess performance change.
The total number of races included in the analysis, along with the number of improvements (change < SWC) and decrements (change > SWC) in performance are reported in Tables 1 and 2, respectively. The self-report group had a total of 143 performances included in analysis, recording 27 ± 8 races/athlete. The combined group had a total of 143 performances included in analysis, recording 27 ± 6 races/athlete.
The week-to-week CV%, reliability, and signal-to-noise ratio are reported in Table 3. For both groups a total of 280 training weeks (40 wk/athlete) were analyzed for the week-to-week CV% respectively. For both groups, 4 weeks were excluded from analysis due to no training. The seven athletes in the self-report group had a total of 252 athlete training weeks included in analysis. The combined group had a combined 243 training weeks included for the subjective measures (sleep, fatigue, and TQR) and 226 training weeks for HRV across all 7 athletes. Of all measures assessed there was a “good” signal-to-noise ratio for the following variables in the self-report group: soreness (3.1), general health (3.0), wellness% (2.0), and motivation (1.6). The combined group had a “good” signal-to-noise ratio for sleep (2.6), TQR (1.8), fatigue (1.4), R-R interval (2.5), and LnRMSSD:RR (1.3).
The lack of discriminatory ability of any single monitoring variable brings into question the utility of this approach to accurately assess performance change. The “good” signal-to-noise ratio of the numerous monitoring tools assessed shows the variables potential to monitor athletes’ fitness and fatigue. However, both the type of athlete and sport should be considered in establishing a monitoring system. A multidimensional monitoring system to account for variations in fatigue and an athlete’s performance capacity should therefore be considered.
Monitoring the general health of athletes was the only variable that met the criteria to assess decrements in performance. We recommend from the results of both the AUC and the Youden index that caution be taken with the interpretation of any single monitoring tool to assess performance change. Further, it is appropriate that monitoring tools are validated with each sport. The smaller CV% in swimmers compared with team sports suggests that even subtle variations in subjective and HRV measures may be important in these athletes.
Subjective questionnaires, heart-rate variability, training monitoring, swimming.
Crowcroft, S., McCleave, E., Slattery, A. & Coutts, A, J. (2017). Assessing the measurement sensitivity and diagnostic characteristics of athlete-monitoring tools in national swimmers. International Journal of Sports Physiology and Performance, 12, S5-95-100.