REF-732 · Valenzuela, Alejo, Montalvo-Pérez, Gil-Cabrera, Talavera, Lucia & Barranco-Gil, 2021 · Frontiers in Physiology, 2021;12:676484 · Paper · PubMed
Critical power and lactate-threshold markers agree strongly on average and disagree widely on the individual: correlations r = 0.81-0.98, but limits of agreement of ±22 to ±52 W
Protocol
n = 17 male recreational cyclists (33 ± 5 y; peak power output 4.5 ± 0.7 W/kg). Critical power (from three trials of 1, 5 and 20 min) was compared against four lactate-threshold (LT2) markers determined in a maximal incremental test: the fixed onset of blood lactate accumulation, and three individualised markers (Dmax-exp, Dmax-pol, LTΔ1).
Key Findings
Correlations were strong: r = 0.81-0.98.
Differences were trivial-to-small and non-significant for most markers (Hedges' g = 0.01-0.17, bias 1-9 W, p > 0.05).
Dmax-exp correlated best but sat significantly above CP (Hedges' g = 0.43, bias 20 W, p < 0.001) — the strongest correlation was not the most interchangeable marker.
The limits of agreement were wide: ±22 W (Dmax-exp) to ±52 W (Dmax-pol) across markers. Magnitude-based inferences rated the between-marker power differences (>1% but <5%) as unclear to most likely practically meaningful.
Authors' conclusion: LT2 markers show strong association and overall trivial-to-small differences with CP, but given the wide limits of agreement and the likelihood of meaningful individual differences, caution should be used when treating them interchangeably.
Practical Application
A strong correlation between two threshold markers is not permission to swap them for one athlete. ±22 to ±52 W is the width of a training zone; the group mean agreeing tells you nothing about whether this rider's numbers agree.
Say which marker a number came from. "Threshold" alone is ambiguous; a CP value and an LT2-by-Dmax-exp value for the same rider are different prescriptions — the mean gap here was 20 W, and the individual gap can be anywhere inside ±22-52 W.
Do not re-derive one from the other across a season. If an athlete's zones were set from CP, keep testing CP; switching marker mid-season produces a change that looks like fitness and is measurement.
The best-correlating marker was systematically biased upward. Choosing a marker by correlation alone will pick one that tracks CP nicely while sitting 20 W above it — a recipe for zones that are quietly too hard.
Recreational cyclists here; in trained and elite riders the agreement question is not settled by this study.
Limitations
n = 17, male, recreational (Tier 2), single laboratory. Trained and elite riders are not represented, and agreement between markers may differ with training status.
Cross-sectional: no test-retest reliability for CP or for the LT markers, so part of the wide limits of agreement is measurement noise rather than true disagreement between constructs.
Neither CP nor any LT marker is validated here against MLSS, the physiological reference — the study compares markers with each other.
Cycling only, laboratory protocols; field-derived estimates carry additional error on top of what is reported here.
Four LT markers among many in use; a marker not tested may agree better or worse.
How this entry was checked and changed
Content correction 2026-09-27
Illustrative "CP 265 W vs LT2 285 W" example -> replaced with the abstract's own figures (mean bias 20 W, LoA ±22-52 W).
Added from the abstract: CP protocol (1-, 5-, 20-min trials), Hedges' g values, and the magnitude-based-inference note (differences >1% but <5%, unclear to most likely meaningful).
Cross-references to duplicate stubs repointed to their canonical entries (REF-129->REF-044, REF-200->REF-024); the stubs were emptied on 2026-09-27.