Very low carbohydrate intake in gestational diabetes: slightly better glucose, poorer nutrition

Design

Secondary analysis of baseline data from the DiGest randomised trial

Population

215 women with gestational diabetes

Primary outcome

Nutrient intake, CGM time above range, birth weight by carbohydrate intake

Effect

<120 g/day: TAR -6.8% (95% CI -13.4 to -0.2); birth weight -9.6 centiles (-18.7 to -0.6)

In the DiGest trial, 215 women with gestational diabetes recorded a 24-hour dietary recall and wore masked continuous glucose monitors before they were randomised. About a third ate the commonly recommended minimum of 175 g of carbohydrate a day or more, a third ate 120 to 175 g, and a third ate under 120 g.

Women eating under 120 g a day spent 6.8 percentage points less time above 120 mg/dL (95% CI -13.4 to -0.2) and had babies about 9.6 birth-weight centiles lighter (95% CI -18.7 to -0.6), with no difference in large- or small-for-gestational-age rates. They also ate less fibre, fewer B vitamins and vitamin E, and fewer minerals.

This is a cross-sectional look at baseline diet, not a test of advising carbohydrate restriction. The glucose benefit is modest and the interval sits close to zero. The nutrient shortfall is the more robust finding and the one to act on in clinic.

• Ask women with gestational diabetes how much they have cut carbohydrate, not only whether they follow the diet.
• Where intake is very low, check for adequate fibre, B vitamins, vitamin E and minerals.
• Do not advise going below the usual carbohydrate floor on the strength of this analysis.
• Involve a dietitian when a woman has restricted carbohydrate heavily on her own.
• In Indian practice, cereal-based diets mean carbohydrate cuts often remove the main fibre source too.

Why it matters

Women often over-restrict after diagnosis; this shows what that costs in nutrients for a small glucose gain.

Don't overread it

Observational baseline data — it cannot show that advising low-carbohydrate diets improves outcomes.

The statistics, in plain English

The confidence interval for time above range runs from -13.4 to -0.2, so the true effect could be close to nothing. The birth-weight difference was in centiles and did not translate into fewer large or small babies, so its clinical meaning is unclear.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.