Reliability of the durability concept in professional cyclists: a field-based study
Protocol
n=18 professional cyclists (27 ± 4 y); one cycling season of training and competition data
Durability defined as the decline in power output after accumulated work
Highest maximum mean power (MMP) values for 10 s and 1, 5, 10 and 20 min recorded after different levels of accumulated work (0-40 kJ/kg)
Repeatability from the two highest MMP values per cyclist for each duration and level of accumulated work: standard error of measurement (SEM) and intra-class correlation coefficient (ICC)
Key Findings
A progressive decline of MMP values compared with the non-fatigued state after higher levels of accumulated work, particularly after 20 kJ/kg (p < 0.001)
All MMP values showed high repeatability under fatigue: all SEM < 5% and ICC > 0.90; the lowest repeatability was for the shortest efforts (10-s MMP)
Confirmed separately for training sessions and for competitions, with lower repeatability there (SEM < 8%, ICC > 0.80)
Conclusion: the measure of durability appears reliable, which might support its validity for monitoring field-based performance in professional cyclists
Practical Application
In 18 professional cyclists over a season, the best 10-s, 1-, 5-, 10- and 20-min powers produced after 0-40 kJ/kg of prior work were repeatable (SEM < 5%, ICC > 0.90 across all durations; lowest for 10 s), and power declined progressively with accumulated work, clearly from 20 kJ/kg on — so a fatigued power profile pulled from ordinary training and race files is a reliable measurement, no lab needed.
Repeatability was lower when training and competition were treated separately (SEM < 8%, ICC > 0.80), so pool a season of both, and treat a change in a fatigued MMP that is smaller than the SEM (under 5%, up to 8% within a single context) as noise.
Read accumulated work in kJ per kg body mass: the decline became clear past 20 kJ/kg, and the study went to 40 kJ/kg.
Inference, not tested: the 10-s value is the noisiest, so sprint durability needs more samples than 1-20-min durability before a change is trusted.
Limitations
Professional-only cohort (n=18); amateurs untested
Reliability from the two highest values per bin — a best-effort method; the abstract gives no absolute decline percentages
Field measurements carry their own confounders (terrain, cadence, motivation, race context)
How this entry was checked and changed
Why this level
regraded 2026-09-27 from A-: reliability study, n=18
Content correction 2026-09-27
"ICC > 0.85 in typical metrics" -> all SEM < 5% and ICC > 0.90 (training and competition separately: SEM < 8%, ICC > 0.80).
"Power decline after 1500-2500 kJ is a practical marker" -> the study expressed accumulated work in kJ/kg (0-40 kJ/kg) with a clear decline from 20 kJ/kg.
"Field methods can replace laboratory methods", "individual variability is relatively stable" -> reworded to the abstract's claim (reliable, supports validity for field monitoring). Added n=18, age 27 ± 4, durations 10 s / 1 / 5 / 10 / 20 min. Level A- -> B; Tier and Sport lines added.
2026-10-01: cross-reference repointed from a merged framework to the one that replaced it (SYNTH-01→31, 05→34, 09→70, 12/15→32, 19→26).