REF-1554 · Simonsen et al., 1991 · Journal of Applied Physiology · Paper · PubMed

How much carbohydrate do weeks of doubles need?

In 22 rowers (10 women) on hard twice-daily training, 10 g/kg/day gave +65% glycogen and +10.7% trial power against flat and +1.6% on 5 g/kg. Tests fuel, not doubling.

Main caveat: Abstract only, and cut off. Arm sizes, confidence, the comparison test and sex splits could not be read.
moderate evidenceMeasured on trained to highly trained athletes

Twenty-two collegiate rowers were randomly assigned to a high-carbohydrate (10 g/kg body mass/day) or moderate-carbohydrate (5 g/kg/day) diet through 4 weeks of hard twice-daily rowing, with protein at 2 g/kg/day and fat adjusted to hold body mass. Mornings were 40 min at 70% of peak oxygen uptake.

Afternoons were either three 2,500-m time trials, used to measure power, or intervals at 70-90% of peak oxygen uptake. Mean daily training came to 65 min at 70% and 38 min at or above 90%. Mean muscle glycogen rose 65% on the high diet (P < 0.05) and held at 119 mmol/kg on the moderate diet.

Mean time-trial power rose 10.7% on the high diet and 1.6% on the moderate one (P < 0.05, as printed for the pair). The authors conclude that the high diet promotes more glycogen and more power during this training. Our reading: when two sessions a day last for weeks, the higher carbohydrate arm stored more glycogen and produced more power in training.

What it isolates is the fuel; the double is common to both arms.

Protocol

Design: random assignment to two diets for 4 weeks; arm sizes are not in the abstract.
Who: 12 male and 10 female collegiate rowers.
Diet: 10 or 5 g carbohydrate per kg per day; protein 2 g/kg/day; fat adjusted to maintain body mass.
Training: 40 min at 70% peak oxygen uptake in the morning; three 2,500-m time trials or interval training at 70-90% in the afternoon. Mean daily 65 min at 70% and 38 min at or above 90%.
Measures: muscle glycogen and power in the time trials within training.
Not readable: the abstract is cut off in PubMed and Crossref at the end of the conclusion, so anything after it is unknown; no timing of biopsies, no test outside training, no rowing race, funding beyond a PubMed tag of non-US-government support, and no interests statement.

Key Results

MeasureHigh carbohydrate (10 g/kg/day)Moderate carbohydrate (5 g/kg/day)P
Mean muscle glycogen, 4 wk+65%constant at 119 mmol/kg< 0.05 (high)
Mean power, time trials, 4 wk+10.7%+1.6%< 0.05, as printed for both figures
Reading the P value. The abstract attaches a single "P less than 0.05" to both power figures. Whether it marks each within-group rise or the between-group gap cannot be told from the abstract. The authors' stated conclusion is a difference between diets.
Our arithmetic: 65 + 38 = 103 min of mean daily training.
Not a low-carbohydrate arm. On 5 g/kg/day power was up 1.6% and glycogen steady. It is a comparison of a very high and a moderate intake.

Mechanisms

Measured: more stored glycogen and more power in the high arm. Whether one caused the other is the authors' inference; the design randomised diet, not glycogen.
Not measured, as far as the abstract shows: glycogen before the second session itself, energy availability, perceived exertion, adaptation markers, health markers, performance away from training.

Practical Application

How much carbohydrate do weeks of doubles need? In 22 rowers (10 women) on hard twice-daily training, 10 g/kg/day gave +65% glycogen and +10.7% trial power against flat and +1.6% on 5 g/kg. Tests fuel, not doubling.
Good for: by transfer from rowing, a reason to raise carbohydrate when two hard sessions a day last for weeks. Output within training rose more on 10 g/kg/day, as the authors conclude; the abstract's P value does not settle whether that gap itself was tested.
Good for: setting the 5 g/kg/day figure against a sustained double load: glycogen stayed flat and power rose little on it.
Not good for: whether to double at all. Both arms did, and there is no once-a-day arm.
Not good for: a figure for athletes doing short or easy doubles. The load was about 38 min a day at or above 90%, for 4 weeks.
Not good for: a race or a test off training. Power was measured in time trials within training; the abstract shows no rested test.
Not good for: advice by sex. Women were in the sample; the abstract gives no result by sex.

Limitations

Abstract only, and cut off. Arm sizes, confidence, the comparison test and sex splits could not be read.
Rowing, whose muscle mass and task differ from cycling and running; nothing on either here.
Power measured in training, with no rested test shown in the abstract, and the diet probably not hidden from the rowers (our inference).
Fat adjusted to hold body mass, so the two diets very likely differ in fat as well as carbohydrate (our inference).
A 1991 study. Methods beyond the abstract could not be read, and no energy availability measure is in the abstract.
No health or menstrual data readable despite 10 women in the sample.
How this entry was checked and changed
Why this level

parallel RCT with n = 22 (the rubric's 20-39 band). Read from the abstract only, and PubMed and Crossref both cut it off at 250 words; no open copy in PubMed Central or Europe PMC. It sits at the low edge: arm sizes, the test comparing arms and results by sex could not be read, and the abstract says nothing on blinding of the diet