Author(s): WANG, B., BURNLEY, M., BLAGROVE, R.C., Institution: LOUGHBOROUGH UNIVERSITY, Country: UNITED KINGDOM, Abstract-ID: 760

INTRODUCTION:
The speed-duration relationship describes the hyperbolic relationship between running speed and time to task failure. Critical speed (CS) is the asymptote of this relationship, defining the boundary between heavy and severe exercise intensity domains, below which a metabolic steady state can be achieved. The curvature constant D' represents a finite distance capacity available when running at speeds >CS before task failure. These parameters are strongly related to middle-distance running performance, with CS broadly reflecting sustainable oxidative energy production and D' the finite capacity for substrate-level phosphorylation. Although laboratory-based constant-speed time-to-exhaustion (TTE) protocols are often considered the gold standard for assessing the speed-duration relationship, the reproducibility of CS and D' derived from TTE trials has not been examined. Therefore, this study aimed to assess the test-retest reliability of CS and D' using TTE trials in well-trained runners.
METHODS:
Five male (age 25 ± 6 y; maximal oxygen uptake (VO2max) 68 ± 6 mL/kg/min) and 5 female (age 26 ± 7 y; VO2max 57 ± 5 mL/kg/min) well-trained distance runners attended the laboratory on five occasions, each separated by 3-7 days. The first visit characterised blood lactate responses to increased running speed and quantified VO2max. Participants subsequently completed 4 constant-speed TTE trials across visits 2-3 (60 min passive recovery between runs within visits) on a motorised treadmill (1% gradient) at speeds spanning 50% delta to sVO2max to elicit exhaustion between ~2-15 min. The same 4 speeds were repeated in the identical order for each participant during visits 4-5 to allow test-retest analysis. CS and D' were estimated using three two-parameter models (distance-time, inverse-time, and hyperbolic). Reliability was quantified using typical error (TE), coefficient of variation (CV), intraclass correlation coefficients (ICCs), and systematic differences.
RESULTS:
No systematic differences were observed between test and retest (P > 0.05). CS demonstrated excellent reliability across models (ICCs 0.985-0.992; TEs 0.8-1.1%; CVs 0.7-1.0%), whereas D' exhibited greater variability (TEs 11.1-13.9%; CVs 9.3-10.9%) and moderate (hyperbolic model; ICC 0.692) to good (distance-time and inverse-time models; ICCs 0.783-0.813) reliability, with 95% confidence intervals spanning poor to excellent.
CONCLUSION:
CS derived from severe-intensity constant-speed TTE trials under controlled laboratory conditions represents a highly reliable performance threshold, such that relatively modest intervention- or condition-induced changes are likely to be detected. The greater variability observed in D' indicates its higher sensitivity to model selection and inherent performance fluctuations during TTE trials. Therefore, caution is warranted when using D' to monitor small performance changes or evaluate intervention effects in well-trained runners.