REF-1619 · Hogan, Cross, Tiong, Pratheepan, Blaxland, Machan & Rooney, 2026 · Medicine & Science in Sports & Exercise · Paper · PubMed · Free full text
Maximal fat oxidation (Fatmax) sits below the first threshold, not at it: pooling 158 comparisons from 42 studies (1505 healthy people, 68% of those with a recorded sex male), Fatmax came out lower than the first lactate or gas-exchange threshold (LT/GET) by about 5.78% of VO2max, 4.75 mL/kg/min, 22 W or 11 beats/min (standardised mean difference 0.74, 95% CI 0.32-1.16; I² 95.0%); the gap was larger with test stages of 3 min or less and was not detected with longer stages, and it is a group average, so this review cannot say where one person's Fatmax sits against their LT1 or VT1. Read in full
Main caveat: The authors': healthy, non-obese people only, so not for clinical groups or obesity; heterogeneity stayed high inside subgroups.
Hogan and colleagues (University of Sydney) asked whether the intensity of maximal fat oxidation (Fatmax) is the same intensity as the moderate-to-heavy transition marked by the first lactate threshold or the gas-exchange (first ventilatory) threshold, which they group as LT/GET, and which features of the people and the tests change the answer.
An earlier meta-analysis of 14 studies had found no significant difference (effect size −0.23, 95% CI −0.56 to 0.09), and its follow-up found LT/GET intensities often fell outside the predefined range within ±10% of maximal fat oxidation, with no consistent directional bias.
This review searched four databases from inception and kept any study of healthy, non-obese people that reported both markers in the same units from graded exercise; 42 studies qualified, 30 of them not in the earlier reviews, giving 158 effect sizes.
In a three-level random-effects model Fatmax sat lower than LT/GET by a standardised difference of 0.74 (95% CI 0.32-1.16, P = 0.0010), which the authors express as about 5.78% of VO2max, 4.75 mL/kg/min, 22 W or 11 beats/min.
Heterogeneity was high (I² 95.0%, mostly between studies by the text's split, which Table 6's variances contradict); leave-one-out estimates ran 0.651-0.807 and no publication bias was detected.
Run one at a time, the moderator analyses found the gap clear in men, women, people aged 20 or over and active to elite groups, larger against GET (0.905) than against LT (0.524), larger after a 7-12-h fast (1.069) than after a fast under 6 h (0.374, not significant), and not significant, with point estimates below zero, when the stages were long (Fatmax test −0.108; VO2max test −0.543) against 1.241 and 0.869 with short stages.
The authors conclude that Fatmax usually falls below LT/GET, that the two should not be swapped without thought, and that apparent agreement under some test conditions may be a method artefact; they recommend standardising food for 48 h and a 10-12-h overnight fast for fasted tests, the Jeukendrup and Wallis exercise equations, and third-order polynomial or sine-model fitting for Fatmax.
Our reading: the direction is clear at group level; the moderator claims are weaker than the abstract makes them, because no subgroup was tested against another and several "no difference" subgroups carry the same gap with a wider interval.
Participants
Pooled sample: 1505 participants in 42 studies from 18 countries (27 studies, 64%, from Europe); 958 men, 402 women, 44 reported only as a mixed group, 101 with sex not recorded; 22 studies men only, 13 both sexes, 3 women only.
Age: study mean ages 13.9 to 48.5 years; pooled weighted mean 28.1 ± 10.9 years; by study, 15% under 20, 40% 20-29, 41% 30-39 and 4% 40-49 years.
Fitness (the review's labels): elite 448 (30%), trained 525 (35%), active 407 (27%), sedentary 125 (8%), assigned from the study authors' descriptions and, where possible, VO2max against sex-, age- and mode-specific norms.
Health: apparently healthy, BMI < 30 kg/m², no known cardiometabolic disease or cancer, no medication affecting substrate use; only control-group data from clinical studies and only pre-intervention data from intervention studies.
Women's physiology: of 16 studies with women, 7 gave no menstrual or contraceptive details, 4 recorded regular cycles without controlling phase, 5 tested in the follicular or the luteal phase; 5 gave details of oral-contraceptive use (in 4 the women were not using it; 1 compared users with non-users).
Protocol / Methods
Search: PubMed, MEDLINE, Embase and Scopus from inception to 9 July 2022, updated 20 January 2025; keywords for lactate, ventilatory, gas-exchange, metabolic, aerobic or anaerobic threshold combined with Fatmax, maximal or peak fat oxidation, lipomax or fuel utilisation. 4,200 database records plus 3 from other sources, 2,806 after duplicates, 349 full texts, 42 included (66 were potentially eligible; 34 authors were asked for missing data, 10 replied, 24 studies dropped).
Inclusion: peer-reviewed original research; cross-sectional design, or the control or baseline data of an intervention; LT/GET from a graded test by blood or plasma lactate, GET or VT; Fatmax from a graded protocol; mean and SD or SE for both, in identical units (VO2 absolute, relative or as % of VO2max; W, W/kg or % of maximal power; heart rate, % of HRmax or % of heart-rate reserve).Excluded: any threshold, Fatmax or VO2max from prediction equations, reference values or extrapolation; LT defined as 4 mmol/L, a point that earlier work, the authors say, showed to mark the maximal lactate steady state.
How Fatmax was found in the studies: fat oxidation from gas exchange by eight stoichiometric approaches (Frayn 1983 in 20 studies, 48%, by the text but 50.0% in Table 2; Jeukendrup and Wallis 2005 in 23.8%; no details in 11.9%); Fatmax as the highest measured stage in 24 studies (57%), otherwise third-degree (11.9%) or second-degree (7.1%) polynomials, curve smoothing, the sine model (4.8% each), Mader's approach, visual inspection or not stated.
How LT/GET was found: 20 studies used LT, by 12 methods (most often the first rise in lactate above baseline, 6 studies; others include the log-log method, the highest intensity before the onset of blood lactate accumulation, the test's lowest lactate + 1.5 mmol/L, and 2 mmol/L read from a fitted curve); 26 used GET, by 13 methods (first sustained rise in VE/VO2 with VE/VCO2 flat 11.9%, Wasserman ventilatory equivalents 9.5%, V-slope 7.1%, others); 4 studies measured both.
Tests: cycle ergometer in 30 studies (71%), treadmill in 12 (29%); 26 (62%) found both markers in one graded test to exhaustion, most often with 3-min stages (10 studies); 16 (38%) used a separate VO2max test and a submaximal Fatmax test (Fatmax stages 2-20 min, most often 6 min); 21 metabolic carts plus one Douglas-bag set-up, the Oxycon Pro most common (9 studies).
Food and fasting before the test: fasting stated in 32 studies (76%), from a 2-h fast on the test morning to an overnight 10-12 h (an overnight fast of 10-12 or 7-10 h in 20 studies, 48%); unclear in 10 (24%); usual diet in 19 (45%); a carbohydrate-rich diet beforehand in 12 (29%); meals set by a nutritionist in 7 (17%); food recall in 10 (24%).
Model: three levels (participants within a sample, effect sizes within a study, studies), restricted maximum likelihood, Hedges' g with 95% CI; a positive effect means Fatmax lower than LT/GET. The three-level model fitted better than a two-level one (AIC 425 vs 513, BIC 434 vs 519; likelihood ratio χ² = 89.65, P < 0.0001).Multilevel I² (≤25% low, ≤50% moderate, >75% high); funnel and contour-enhanced funnel plots, Egger's test, trim and fill; leave-one-out sensitivity; univariate moderator models for 23 study, participant, nutrition, protocol and analysis features by the text's count (Table 6 lists 22). The full statistical description is in Supplementary Material 3, which we did not read.
Risk of bias: RoB 2 for 11 randomised crossover trials, the NIH tool for 3 pre-post studies without control, AXIS for 28 cross-sectional studies; rated by three reviewers with a third to settle differences; agreement κ = 0.90 over 766 items.
Funding and interests: the authors report no conflicts of interest; the article text carries no funding statement.
Key Results
Pooled difference, Fatmax below LT/GET (42 studies, 158 effect sizes):
| Unit | Fatmax lower by (95% CI) |
|---|---|
| Standardised mean difference (Hedges' g) | 0.74 (0.32-1.16), P = 0.0010; forest plot prints 0.74 [0.33, 1.15] |
| % of VO2max | 5.78 (2.50-9.06) |
| VO2, mL/kg/min | 4.75 (2.05-7.45) |
| Power, W | 22 (9-34) |
| Heart rate, beats/min | 11 (5-18) |
Heterogeneity: I² = 95.0%, 12.3% within studies and 82.7% between studies; single effects ranged from −4.10 (Fatmax well above the threshold) to 11.62 (a 4-person sample with SDs of 0.8).
Sensitivity and bias: leave-one-out pooled effects 0.651-0.807; Egger intercept 1.77, t = 1.14, P = 0.13; trim and fill added no studies.
Risk of bias as rated by the authors: all 11 crossover trials low overall; the 3 pre-post studies moderate; the AXIS tool gives no overall rating, but all 28 cross-sectional studies stated their aims, one justified its sample size, and about 65% did not make clear whether the sample represented its target population.
Where the unit figures come from (our arithmetic): each unit's interval equals the standardised interval multiplied by a fixed figure (about 7.8% of VO2max, 6.4 mL/kg/min, 30 W, 15 beats/min), so all four unit figures, not only the watts and beats, read as the pooled standardised gap converted with a typical spread, not as separately pooled differences; the moderator by unit gave 0.747 (VO2), 0.750 (power) and 0.769 (heart rate).
Moderators, each run on its own (Table 6; standardised gap, positive = Fatmax lower; k = number of effect sizes):
| Moderator | Subgroup (k): gap (95% CI) |
|---|---|
| Sex | men (103): 0.855 (0.413-1.297); women (32): 0.979 (0.493-1.464); not recorded (20): −0.612 (−2.171 to 0.947) |
| Age | under 20 y (36): 0.206 (−0.372 to 0.783; printed 0.7833); 20-29 y (72): 0.943 (0.289-1.597); 30 y or more (50): 0.663 (0.296-1.030) |
| Fitness | elite (85): 0.720 (0.299-1.141); trained (43): 0.600 (0.069-1.131); active (17): 0.835 (0.258-1.412); sedentary (13): 0.886 (−0.357 to 2.129) |
| Menstrual cycle | controlled (11): 1.370 (−0.003 to 2.743); not controlled (13): 0.326 (−0.771 to 1.423); unclear (11): 1.099 (0.411-1.787) |
| Oral contraceptives | users excluded (12): 0.944 (−0.189 to 2.078); unclear (22): 0.815 (0.326-1.305); users included (1): no estimate |
| Fasting | under 6 h (52): 0.374 (−0.273 to 1.021); 7-12 h (63): 1.069 (0.496-1.641); unclear (43): 0.538 (−0.601 to 1.676) |
| Usual diet kept | yes (38): 0.214 (−0.476 to 0.904); no (68): 1.083 (0.376-1.791); unclear (52): 1.219 (0.570-1.868) |
| Carbohydrate-rich diet | yes (71): 0.765 (−0.216 to 1.746); no (16): 0.348 (−0.259 to 0.954); unclear (71): 0.913 (0.317-1.509) |
| Standardised meals / food recall | meals yes (27): 0.745 (−0.445 to 1.935), no (131): 0.741 (0.286-1.196); recall yes (31): 0.670 (−0.412 to 1.752), no (127): 0.763 (0.312-1.214) |
| Threshold marker | GET/VT (80): 0.905 (0.384-1.427); LT (78): 0.524 (0.020-1.029) |
| Mode | cycle ergometer (135): 0.748 (0.238-1.258); treadmill (23): 0.724 (−0.061 to 1.509) |
| Fatmax test | same test as VO2max (123): 0.683 (0.175-1.190); separate test (35): 0.841 (0.078-1.604) |
| Constant increments | yes (85): 0.693 (0.183-1.203); no (38): 0.654 (−0.514 to 1.822) |
| Stage length, VO2max test | short (126): 0.869 (0.404-1.333); long (32): −0.543 (−1.468 to 0.383) |
| Stage length, Fatmax test | short (103): 1.241 (0.690-1.792); long (55): −0.108 (−0.824 to 0.609) |
| Gas analysis | breath by breath (90): 0.843 (0.283-1.403); portable (30): −0.468 (−1.720 to 0.784); spirometry (21): 1.694 (0.354-3.034); mixing chamber (8): 0.348 (−0.628 to 1.323); several (6): 1.149 (0.299-1.999) |
| Fat-oxidation equation | Frayn 1983 (84): 0.945 (0.484-1.407); Jeukendrup and Wallis 2005 (26): 0.998 (−0.100 to 2.096); others (19): 0.403 (−0.892 to 1.697); unclear (29): −0.483 (−1.728 to 0.763) |
| Fatmax method | highest measured stage (80): 0.824 (0.207-1.440); polynomial (35): 1.109 (0.555-1.664); sine model (8): 0.708 (−0.093 to 1.510); other model (9): 0.072 (−1.985 to 2.128); unclear (25): 0.103 (−0.580 to 0.786) |
| LT1 method | first rise above baseline (25): 0.624 (−0.136 to 1.383); highest intensity before the rise (24): 0.461 (−0.377 to 1.299); log or curve fitting (18): 0.720 (0.471-0.970); fixed lactate (8): 2.001 (−0.557 to 4.559) |
| GET method | ventilatory equivalents (63): 0.439 (−0.366 to 1.244); V-slope (16): 1.442 (0.678-2.206) |
Short and long stages: the Results define short stages as 3 min or less; the Discussion places the comparable results in separate-test protocols with stages of 4 min or more. These are the only two moderators whose subgroup intervals do not overlap (our reading of Table 6).
What the tests test: each moderator's F-test has as many numerator degrees of freedom as subgroups with an estimate (for example sex F(3, 37) = 5.786, P = 0.002; fasting F(3, 39) = 5.503, P = 0.003), and all 22 in Table 6 were significant (P 0.0001 to 0.020; the text counts 23); this fits a test that the subgroup gaps are all zero, not that they differ from each other (our reading). No subgroup was contrasted with another.
Authors' physiological reading: the idea that lactate inhibits fat oxidation rests on mechanisms studied in animals and in vitro, such as GPR81-mediated suppression of adipose lipolysis, which they say is driven by intramuscular lactate, a signal that need not rise together with blood lactate; since Fatmax usually falls before the blood-lactate transition, they suggest other factors start the decline in fat oxidation and lactate may add to it at higher intensity.
Practical Application
Maximal fat oxidation sits below the first threshold, not at it: across 42 studies and 1505 healthy people Fatmax averaged about 22 W, 11 beats/min or 6% of VO2max under LT1/VT1. This is a group average with I² of 95%; how far apart the two sit for one person is not known from this review. Inference: Zone 2 set just under a measured LT1/VT1 mostly runs above Fatmax, and riding at Fatmax is a notch easier than the top of Zone 2.
Do not use one as the other for one athlete. The authors offer LT/GET as a stand-in for Fatmax under certain test conditions when substrate oxidation cannot be measured directly.Our reading of their Table 6 is narrower: the gap was not significant, with point estimates below zero, with long stages (−0.108 and −0.543) and with portable analysers (−0.468), and shrank with a usual diet (0.214), age under 20 (0.206) or a fast under 6 h (0.374), all with wide intervals; treadmill tests (0.724) and sedentary people (0.886) carried the same gap as everyone else with a wider interval, and those conditions may reflect method rather than physiology, as the authors themselves allow.
Reading a lab report. In the pooled studies, Fatmax from tests with stages of 3 min or less sat well below LT/GET and Fatmax from longer stages did not (gap 1.241 against −0.108, a comparison between groups of studies, not within the same people), and the fat-oxidation equations and curve fits gave different gaps too.To track change, repeat the test with the same protocol, cart, equation, fasting and food; the authors recommend standardising food for 48 h, a 10-12-h overnight fast for a fasted test, the Jeukendrup and Wallis exercise equations, and a third-order polynomial or sine-model fit rather than the single highest stage.
Inference (our arithmetic with [REF-451]): a Fatmax test repeated three times in 10 trained men had a typical error of 9.6% of its oxygen uptake, about 6 points of VO2max at the 62.5%-of-VO2max Fatmax of its 55-man sample, which is as large as the pooled gap of 5.78 points; one test can therefore put Fatmax at, above or below a person's VT1 without the physiology differing.
Women: the gap ran the same way in women (0.979) as in men (0.855). The authors' suggestion that LT/GET may line up with Fatmax in women not taking oral contraceptives rests on a non-significant subgroup whose point estimate is larger than the overall gap (0.944, −0.189 to 2.078, 12 effects); we read that as not shown.
Not from this paper; the base's own practice: set easy and Zone 2 work from a measured first threshold (REF-025 defines Zone 2 as just below it; SYNTH-51 keeps easy work comfortably below it); a Fatmax test is not needed to train the aerobic base; a narrative review finds no support for Zone 2 as the optimal intensity for fat-oxidative or mitochondrial capacity (REF-1092), and PHYS-02 does not treat Fatmax as the intensity that trains fat oxidation best.
Limitations
The authors': healthy, non-obese people only, so not for clinical groups or obesity; heterogeneity stayed high inside subgroups; moderators could only be run one at a time, never together; too few studies over 55 years, so the top age band is 30 or more; summary data only, so no individual-level agreement; no study corrected for the response time of gas-exchange kinetics in ramp or short-stage tests; only one study of oral-contraceptive users.
Certainty: no GRADE or other certainty rating. The 28 cross-sectional studies, two thirds of those included, have no overall risk-of-bias rating, and one of them justified its sample size. In Figure 2A four crossover trials carry a high-risk domain (deviations from intended interventions) yet all 11 are rated low overall, which the usual RoB 2 rule would not allow.
Effect size: standardised differences from group means and SDs of paired, within-person data; the forest-plot values reproduce Hedges' g for two independent groups with variance 2/n + g²/(4n) (our arithmetic: Ghorbani 2014 (1) −4.10 [−5.82, −2.37], González Haro 2007 (2) 11.62 [5.76, 17.48] and Dunst 2024 −1.55 [−2.40, −0.71] all match), so the pairing of the two markers in the same people was ignored.Very small samples with tiny SDs give extreme effects (11.62 from 4 people).
Units: inclusion required both markers in the same intensity units, yet Table 6 lists 8 effects in mmol/L, 1 in km/h and 1 in muscle oxygen saturation; how a lactate concentration served as an intensity unit is not explained.
Moderators: no subgroup was tested against another, so "no significant difference" in a subgroup reflects a wide interval, not demonstrated agreement.
Reporting (things that do not add up): the abstract lists sedentary people under 20, treadmill tests and two-session protocols as conditions with a smaller gap, but Table 6 gives sedentary 0.886, treadmill 0.724 and a separate Fatmax test 0.841 (significant), and sedentary and under-20 are separate subgroups; the Results text gives the cycle ergometer as n = 84, 0.945 (0.484-1.407), which is the Frayn equation row, while Table 6 gives 135 effects, 0.748 (0.238-1.258); the subgroup with an unclear carbohydrate intake (interval 0.317-1.509) is printed with P = 0.22; the pooled interval is 0.32-1.16 in the text and 0.33-1.15 in the forest plot; eight equations in the Results, thirteen in the Discussion; the shortest fast is 2 h in the Results and 90 min in the Discussion; and Table 6's level-2 and level-3 variances (about 1.5-1.7 and 0.24 where a moderator explains nothing) look swapped against the text's 12.3% within-study and 82.7% between-study split (our reading: with the columns swapped and a typical sampling variance near 0.1, 0.24 and 1.6 reproduce roughly 12% and 83%); and the text counts 23 moderators and 20 studies (48%) using Frayn where Table 6 lists 22 and Table 2 gives Frayn 50.0%.
Not read: the supplementary study table, risk-of-bias tool details, statistical methods and funnel plots.
How this entry was checked and changed
Why this level
a systematic review and meta-analysis whose risk of bias we read in the full text, not A because the pooled data are measurements, not controlled-trial comparisons (the rubric's A row), and held at A- by serious heterogeneity (I² 95.0%, single effect sizes from −4.10 to 11.62), 28 of 42 studies being cross-sectional with sample size justified in one, no GRADE certainty rating, and a group-level comparison only; no marker step, because Fatmax and the thresholds are themselves the measured subject of this agreement question (rule 3 exception)