REF-752 · Lamberts RP, Swart J, Noakes TD, Lambert MI — 2011 · British Journal of Sports Medicine, · Paper · PubMed
A 15-minute submaximal cycle test (LSCT) was highly reliable in its second and third stages (ICC 0.85-1.00, typical error 1.3-4.4%), and mean power in stage 3 correlated with 40-km time-trial and peak power output at r = 0.91-0.94
Far better than heart-rate recovery did (r = 0.55-0.68)
Protocol
Seventeen well-trained male road cyclists each completed four peak power output tests and four 40-km time trials, and performed the Lamberts and Lambert Submaximal Cycle Test (LSCT) before every one of them.
The LSCT clamps heart rate, not power: three stages at fixed, predetermined heart rates (6 min at 60%, 6 min at 80%, 3 min at 90% of maximal heart rate), then 1 min of recovery in which heart-rate recovery (HRR) is measured. Power, speed, cadence and RPE are the outputs.
Key Findings
Stages 2 and 3 were highly reliable: intraclass correlations 0.85-1.00, typical error of measurement 1.3-4.4%.
Mean power in stage 2 correlated with performance at r = 0.80-0.89, and in stage 3 at r = 0.91-0.94 (peak power output and 40-km TT).
HRR correlated with performance more weakly, r = 0.55-0.68.
The authors conclude the test predicts peak and endurance performance from submaximal power, RPE and HRR, and detects meaningful change more precisely than VO₂max.
Practical Application
The strongest single number in a submaximal test is the output at a fixed internal load — power at a clamped heart rate, or heart rate at a clamped power — not heart-rate recovery. HRR related to performance far more weakly (r 0.55-0.68 against 0.91-0.94 for stage-3 power); read it as a supporting variable, never as the headline.
Put the test at the end of an identical warm-up and before any hard work, so every repeat starts from the same state; only the later, harder stages were reliable enough to track.
A change inside roughly 1.3-4.4% of power is within the test's own error — do not report it as a change in form.
Limitations
n = 17, men only, one research group; correlations are between athletes, which does not prove the test tracks change within one athlete (see REF-754, REF-753 for within-athlete use).
Laboratory conditions; field execution adds heat, wind and gradient noise.
The heart-rate-clamped design requires a measured maximal heart rate (REF-3494).