REF-1326 · Galán-Rioja et al., 2020 · Sports Medicine · Paper · PubMed

How far below my critical power is my first threshold?

Pooled over 9 small studies, VT1 sat about 30% below CP and MLSS about 11% below; correlated, not interchangeable. Abstract only, with statistical flaws (B).

Main caveat: Abstract only. Which studies, which sports, which CP models and protocols, the participants and the risk of bias could not be checked.
limited evidenceMeasured on recreational to highly trained athletes

Galán-Rioja, González-Mohíno, Poole and González-Ravé asked how closely critical power (CP), proposed as the boundary between the heavy and severe domains (the maximal metabolic steady state), corresponds to the maximal lactate steady state (MLSS), the first and second ventilatory thresholds (VT1, VT2) and the respiratory compensation point (RCP).

They searched Scopus and Web of Science on 2 October 2019, screened 356 articles and included 9 studies with 104 participants, pooling correlation coefficients with a random-effects model. The pooled correlation across all thresholds was r = 0.73.

By pair: CP and RCP r = 0.80 (95% CI 0.40 to 1.21), CP and MLSS r = 0.77 (0.36 to 1.18), CP and VT1 r = 0.76 (0.31 to 1.21), CP and VT2 r = 0.39 (−0.37 to 1.15, p = 0.31). Despite the correlations, the thresholds under- or over-estimated CP: VT1 by 30% and MLSS by 11% below, RCP by 6% and VT2 by 21% above.

The authors conclude that, if CP is the best estimate of the heavy-severe transition, none of these thresholds, as determined in the analysed studies, is synonymous with it.

Our reading: the direction is solid and fits the base (MLSS somewhat below CP, REF-168; strong correlation is not interchangeability, REF-732), and it gives a pooled anchor for where the first threshold sits below CP — about 70% of it (our arithmetic) — but the statistics in the abstract have visible errors, so the percentages are approximate and say nothing about one athlete.

Protocol

Search: Scopus and Web of Science, 2 October 2019; 356 articles analysed.
Inclusion (as printed): "studies were randomized controlled trials" and studies that included interrelations between CP and VT1, VT2, MLSS or RCP; duplicates and studies not meeting the criteria excluded.
Included: 9 studies, n = 104 participants in total; who they were (sex, training status, sport) and how CP and each threshold were determined are not given in the abstract.
Analysis: random-effects weighted meta-analysis of correlation coefficients; subgroup analysis by threshold; mean percentage under- or over-estimation of CP by each threshold.
Not in the abstract: risk-of-bias assessment, heterogeneity statistics, CIs for the percentage differences, how the percentages were computed (relative to CP or to the threshold), the CP model and protocols used in the included studies.
Funding and interests: not stated in the abstract or the PubMed record; Crossref lists no funder.

Key Results

CP compared withPooled r (95% CI)Z, pMean difference from CP
RCP0.80 (0.40 to 1.21)Z = 3.90, p = 0.0001over by 6%
MLSS0.77 (0.36 to 1.18)Z = 3.71, p = 0.0002under by 11%
VT10.76 (0.31 to 1.21)Z = 3.32, p = 0.0009under by 30%
VT20.39 (−0.37 to 1.15)Z = 1.01, p = 0.31over by 21%
All thresholds0.73 (p printed "> 0.00001")——
Authors' conclusion: regardless of significant correlations, CP differs significantly from each threshold; none should be considered synonymous with CP.
Ratios, if each percentage is taken relative to CP (our arithmetic; the abstract does not say which way it was computed): VT1 about 70% of CP, MLSS about 89%, RCP about 106%, VT2 about 121%; VT1 about 79% of MLSS.

Mechanisms

Measured: nothing new; pooled correlations and mean differences from the included studies.
Authors' interpretation: CP marks the heavy-severe boundary (the maximal metabolic steady state); MLSS, VT1, VT2 and RCP are different constructs, at least as determined in the included studies, and should not be used as synonyms for CP.
Inference: VT1 marks a different boundary (moderate to heavy), so its "30% underestimate" of CP is mostly the width of the heavy domain, not a measurement error; the high correlation reflects that fitter people have higher values of every threshold. VT2 and RCP are often treated as one event (as in REF-1323); that they differ here by about 15 points suggests the included studies defined them differently (inference).

Practical Application

How far below my critical power is my first threshold? Pooled over 9 small studies, VT1 sat about 30% below CP and MLSS about 11% below; correlated, not interchangeable. Abstract only, with statistical flaws (B).
The pooled ladder (our arithmetic, group means): VT1 about 70% of CP and about 79% of MLSS — close to the 78% of MLSS measured in 14 well-trained men (REF-1323). Where CP and a 20-min FTP agree on average (275 vs 278 W in REF-040, but +30 to −36 W apart individually), this puts VT1 near 70% of FTP; REF-1325 gives about 73% by another route. Both are averages across different samples.
Do not read MLSS, RCP or VT2 as CP: MLSS sat below CP and RCP and VT2 above it on average. For heavy-severe prescription the base anchors on CP (SYNTH-27, SYNTH-53); for the easy ceiling it anchors on a measured first threshold (SYNTH-51).
For one athlete, measure: the abstract gives no individual spread; other entries show limits of agreement of tens of watts between threshold markers (REF-732; REF-1323).

Limitations

Abstract only. Which studies, which sports, which CP models and protocols, the participants and the risk of bias could not be checked.
Statistics that do not add up: three of the four pooled CIs for r run above 1.0 (up to 1.21), which a correlation cannot do; the overall result is printed "p > 0.00001"; the percentage differences have no CI, and whether each is relative to CP or to the threshold is not stated.
Inclusion criterion: "randomized controlled trials" does not describe threshold-comparison studies, which are cross-sectional; how the criterion was applied cannot be told from the abstract.
Small evidence base: 9 studies, 104 participants, about 12 per study on average (our arithmetic).
Published debate: comments by Keir, Mattioni Maturana, Iannetta and Murias (Sports Med 2021;51:367-368) and by García-Tabar and Gorostiaga (Sports Med 2021;51:2011-2013), each answered by the authors (51:369-370; 51:2015-2016); their content was not read for this entry.
Funding and interests: not stated in the abstract or PubMed record; no funder on Crossref.
How this entry was checked and changed
Why this level

a systematic review with meta-analysis, but of 9 small measurement studies (104 participants in all), not of controlled trials, which at best equals the rubric's B+ for a measurement study of 100 or more, and is held one step lower again for serious flaws read from the abstract alone: the abstract's pooled confidence intervals for r exceed 1 (impossible for a correlation, which points to pooling without a Fisher z transformation or to a reporting error), the percentage differences carry no CI or heterogeneity statistic, and the inclusion criterion is printed as "randomized controlled trials". Four published letters (two comments, two replies) followed; not read here. Read from the abstract (closed access; OpenAlex lists no open repository copy and Europe PMC has none)