REF-1122 · Norte et al., 2026 · Scandinavian Journal of Medicine & Science in Sports, 36, e70326 · Paper · PubMed
Graded carbohydrate ingestion up to 120 g·h−1 attenuates the reduction in critical power following 3 h of moderate-intensity exercise in a dose-dependent manner
Norte et al.: n = 16 trained cyclists and triathletes (15 men, 1 woman) rode 3 h at 95% of the gas exchange threshold (169 ± 14 W, ~60-65% VO2max per the authors) three times, drinking 0 (water), 60 or 120 g/h of a 1:0.8 maltodextrin:fructose drink, then did a 3-min all-out test (end-test power = CP estimate, work above it = W′). Fresh CP was 277 ± 27 W.
After the ride it was 236 W on water (−14.7%), 257 W at 60 g/h (−7.4%) and 266 W at 120 g/h (−4.2%); each step up in dose kept significantly more CP (60 vs 0: +20 W; 120 vs 60: +9 W). No dose stopped the fall. W′ fell ~13-22% and did not differ between the three feeding conditions.
"Dose-dependent" here means a stepwise ordering across three doses; 90 g/h was not tested, and the second 60 g/h step bought less than half of what the first did.
Protocol
Randomised, fully counterbalanced crossover; 18 recruited, 2 withdrew, n = 16 analysed (15 male, 1 premenopausal naturally menstruating female; age 35 ± 9 years; body mass 76.8 ± 10.9 kg).
Visits: characterisation (incremental lactate test + ramp test for VO2max, GET and PPO; 3-min test familiarisation); a fresh 3-min all-out test (control CP and W′); then three experimental trials at least 6 days apart.
Before each experimental trial: 48 h of reduced training, 24 h of no exercise and a pre-packaged diet of 8.0 g/kg carbohydrate, 2.0 g/kg protein, 1.0 g/kg fat; breakfast with 2.0 g/kg carbohydrate ~2-2.5 h before the ride.
Ride: 180 min at 95% GET (169 ± 14 W; 24 ± 3 kJ/kg of work). Drink: 150 mL every 20 min (1.35 L total) of water, 20 g or 40 g of carbohydrate per serving (60 or 120 g/h; ~13.3% and ~26.7% solutions), no electrolytes; water ad libitum on top. Water was not taste-matched, and the paper reports no blinding of participants to condition.
3-min all-out test ~3-5 min after the ride, ergometer resistance fixed from the rested ramp test; CP = mean power of the last 30 s (end-test power), W′ = work above it. Participants saw only cadence; researchers were blinded to power until all trials were done.
Measured during the ride every 30 min: gas exchange (substrate oxidation, gross efficiency, O2 cost), heart rate, RPE, capillary lactate and glucose, GI symptoms (0-10 scale).
Funding: Portuguese Foundation for Science and Technology (doctoral grant). Drink products supplied by Peak Supps; one author (Shepherd) is listed with a Precision Fuel & Hydration affiliation, which the conflict-of-interest statement does not mention.
Key Results
| Outcome | Fresh (control) | 0 g/h (water) | 60 g/h | 120 g/h | Comparisons |
|---|---|---|---|---|---|
| CP (3-min end-test power) | 277 ± 27 W | 236 ± 30 W | 257 ± 28 W | 266 ± 29 W | all three below control (p < 0.05); 60 vs 0: MD 20 W, p < 0.001, d = 1.92; 120 vs 60: MD 9 W, p = 0.006, d = 1.01 |
| CP fall vs fresh | — | 14.7% ± 7.1% | 7.4% ± 4.7% | 4.2% ± 3.7% | stepwise with dose |
| W′ (work above end-test power) | 15.5 ± 3.6 kJ | 12.0 ± 4.0 kJ | 13.41 ± 3.7 kJ | 12.7 ± 3.6 kJ | fell after the ride (p < 0.001); no difference between feeding conditions (all p > 0.05) |
| W′ fall vs fresh | — | 22.0 ± 20.1% | 12.6 ± 16.5% | 17.8% ± 12.9% | not dose-ordered |
| %VO2max reached in the 3-min test | 97.7% ± 2.2% | 90.1% ± 3.4% (below fresh, p < 0.001) | 94.3% ± 3.7% | 96.7% ± 2.6% | rose with dose (60 > 0, p = 0.025; 120 > 60, p = 0.036) |
| Whole-body CHO oxidation, mean | — | 1.84 ± 0.28 g/min | 2.16 ± 0.15 g/min | 2.31 ± 0.14 g/min | 120 > 60 > 0 (all p ≤ 0.021) |
| Total CHO / fat oxidised in 3 h | — | 331 ± 51 g / 99 ± 25 g | 389 ± 28 g / 73 ± 28 g | 415 ± 23 g / 57 ± 20 g | CHO up, fat down with dose |
| Blood glucose | — | lower | higher than water from 60 min | higher than water from 60 min | 60 vs 120 no difference at any time point |
| RPE late in the ride | — | higher | lower | lower | water above both fed trials from 150 min |
| Cumulative GI score, as printed [range] | — | 6 [0–56] | 13 [0–78] | 18 [0–81] | 120 > water (p = 0.005); 120 vs 60 not different; only 4 participants ever scored ≥ 4 |
Heart rate, VO2, energy expenditure, gross efficiency and the O2 cost of cycling did not differ between feeding trials. Fluid intake did not differ (1868 to 1946 mL means).
Mechanisms
Measured: carbohydrate oxidation rose and fat oxidation fell stepwise with dose; in the third hour carbohydrate supplied more energy at 120 than at 60 g/h (p = 0.008). On water, fat became the main fuel at ~150 min; at 60 and 120 g/h it never did.
Measured: the fatigued 3-min test reached a higher %VO2max as dose rose (90% → 94% → 97%), alongside the higher CP.
Measured: blood glucose did not differ between 60 and 120 g/h at any time point, yet CP did. Inference, not tested: the extra CP at 120 g/h is not explained by blood glucose concentration; whole-body carbohydrate use is the measured difference.
Measured: heart rate, VO2, energy expenditure, gross efficiency and O2 cost did not differ between feeding trials; blood lactate was higher at 120 g/h than on water but did not differ between 60 and 120 g/h. Authors' interpretation: performance durability (the power-duration curve) dissociated from "physiological resilience" markers at this intensity.
Authors' interpretation, not measured (no biopsies, no liver data): liver glycogen sparing and a small muscle glycogen sparing with higher intakes possibly contributed (they cite the pooled crossover analysis of muscle glycogen, REF-1123). Note: REF-1123 found a small sparing with carbohydrate versus placebo but no moderation by ingestion rate, so it does not support more muscle sparing at 120 than at 60 g/h.
Authors' interpretation: W′ fell regardless of carbohydrate, possibly through depletion of intramyofibrillar glycogen; not measured.
Practical Application
Does eating more on a long ride hold my CP? Partly, and more so at 120 g/h: in 16 trained riders after 3 h steady, CP fell 15% on water, 7% at 60 g/h and 4% at 120 g/h. No dose stopped the fall; 90 g/h was not tested; W′ fell with no difference between doses.
Late-ride targets still come down, even when fed well: at 120 g/h the end-test CP was 266 vs 277 W fresh. Hard intervals placed after 3 h should be set below fresh CP, not at it (SYNTH-32). The authors make the same point: zones from rested tests overstate the fatigued boundary.
Fuelling did not detectably protect W′ late in the ride: W′ was 12.0-13.4 kJ in every trial versus 15.5 kJ fresh, with no difference between doses (REF-145 found the same at 60 g/h). Not from this study, the base's own practice (SYNTH-28, REF-131): make the late surge count once rather than repeating it.
Where this sits on the grams-per-hour ladder: it is a single trial in which 120 g/h beat 60 g/h on a fatigued-capacity outcome (not race time), without participant blinding to dose, in trained riders who had carbohydrate-loaded for 24 h and drank the dose as a 26.7% solution. It does not show 120 beats 90 g/h. The athlete's own row still comes from SYNTH-02, earned through gut training.
The gut cost was small here but not zero: 120 g/h gave higher cumulative GI scores than water (not higher than 60 g/h), with fullness, regurgitation and cramps a little higher; most scores stayed below 4/10. Gut-training history was not reported, so a rider who has not practised the dose should not expect the same.
Limitations
One trial, n = 16, 15 of them men; one 3-h moderate preload at 95% GET in a laboratory (ambient conditions not reported). Magnitudes (−15 / −7 / −4%) belong to this protocol, not to a race or to recreational riders.
Water was the zero-carbohydrate condition and was not taste-matched, and the paper reports no blinding of participants to dose (60 vs 120 g/h were unflavoured 13.3% vs 26.7% drinks; that riders could tell them apart is inference). RPE was higher on water. The 3-min all-out test depends on effort, so expectation could contribute to every contrast, including 120 vs 60.
Only three doses; "dose-dependent" is a three-point ordering. 90 g/h, the current guideline anchor, was not tested, so whether 90 g/h preserves CP as well as 120 g/h is unknown (the authors say so).
CP came from one 3-min all-out test with the ergometer resistance set from the rested state; the authors suggest the fatigued fall may be underestimated. It is not a multi-trial CP model.
No muscle or liver glycogen and no exogenous oxidation tracer; mechanisms are inferred from whole-body gas exchange.
Carbohydrate came as a drink after a 24-h 8 g/kg load and a 2 g/kg breakfast; solid food, gels, heat, and athletes without gut training were not studied. Products were supplied by a supplement company and one author is affiliated with a fuelling company.
How this entry was checked and changed
Why this level
held one step below the rubric's B+ for crossovers ≥ 15: single randomised crossover, n = 16, water control and no reported participant blinding on an effort-dependent test; read from the full text (Europe PMC, CC BY)
Full text (checked 2026-10-13)
Sixteen trained cyclists and triathletes (15 males, 1 female); following participant withdrawal (n = 2) the final sample was 16 (full text).
CP 236 ± 30 W (0 g/h), 257 ± 28 W (60 g/h), 266 ± 29 W (120 g/h) vs 277 ± 27 W control; 60 vs 0 MD 20 W, p < 0.001, d = 1.92; 120 vs 60 MD 9 W, p = 0.006, d = 1.01 (full text).
Relative CP reduction 14.7% ± 7.1%, 7.4% ± 4.7% and 4.2% ± 3.7% for 0, 60 and 120 g/h (full text).
W′ control 15.5 ± 3.6 kJ; 12.0 ± 4.0, 13.41 ± 3.7 and 12.7 ± 3.6 kJ after the ride (all p > 0.05 between conditions) (full text).
%VO2max in the fatigued test 90.1% ± 3.4%, 94.3% ± 3.7%, 96.7% ± 2.6% vs 97.7% ± 2.2% fresh (full text).
CHO oxidation 1.84 ± 0.28, 2.16 ± 0.15, 2.31 ± 0.14 g/min; total CHO 331 ± 51, 389 ± 28, 415 ± 23 g (full text).
Ride power 169 ± 14 W at 95% GET; 24 ± 3 kJ/kg of work over 180 min (full text).
Cumulative GI scores 6 [0–56], 13 [0–78], 18 [0–81]; 120 g/h vs water p = 0.005 (full text).