Continuous glucose monitoring has transformed type 1 diabetes care, and its role in gestational diabetes has remained unsettled — partly because the pregnancies are short, the glycaemic targets tight, and the comparator, finger-prick testing four times a day, is already reasonably effective. This meta-analysis pooled 21 randomised and observational studies covering 5,650 pregnant women, with low to moderate risk of bias.
Glycaemic measures improved modestly but consistently. Mean blood glucose fell by 0.24 mmol/L (95% CI -0.41 to -0.06), coefficient of variation by 0.78 percentage points, mean amplitude of glycaemic excursions by 0.22 mmol/L, and time above 7.8 mmol/L by 2.19 percentage points (-3.78 to -0.60). Clinical outcomes moved more than those numbers suggest: reductions of 8 to 35% across caesarean delivery, macrosomia, neonatal hypoglycaemia and hyperbilirubinaemia.
The finding to sit with is the 21% increase in medication use. That is not an adverse effect — it is the mechanism. Continuous monitoring reveals post-prandial excursions that four daily finger-pricks miss, which prompts earlier escalation to metformin or insulin, which is where the outcome benefit comes from. Reading it as 'CGM medicalises pregnancy' inverts the causation.
Where this lands in Indian practice depends entirely on cost. Gestational diabetes is common here, screening is widespread under national programmes, and a sensor costs several times what a month of test strips does. But the argument is not all one way: the outcomes avoided — caesarean delivery, a macrosomic baby, a neonatal unit admission for hypoglycaemia — are themselves expensive, and in a self-funding family the comparison is not sensor cost against zero. A reasonable middle position is targeted use: women whose finger-prick profiles look acceptable but whose babies are growing above expectation, or women whose adherence to four-times-daily testing is clearly failing.
The evidence has real limits. Pooling randomised with observational studies weakens the whole, the glycaemic differences are small in absolute terms, and outcome reductions spanning 8 to 35% across four different endpoints is a wide and imprecise range.
- Consider CGM in gestational diabetes where finger-prick monitoring is failing or growth is discordant with the readings.
- Expect and plan for more medication — that is how the outcome benefit is produced, not a side effect.
- The glycaemic differences are small; the clinical outcome reductions are the argument.
- Randomised and observational studies were pooled together, which weakens the certainty.
- Where CGM is unaffordable, the transferable lesson is to look harder at post-prandial values.
The statistics, in plain English
A mean glucose difference of 0.24 mmol/L is small enough that it would be easy to dismiss, and the clinical outcome reductions are much larger than that difference alone would predict. The likely explanation is in the medication figure: continuous monitoring changes clinical decisions, and the decisions change outcomes. Two cautions on certainty. Pooling observational studies with randomised ones means confounding from the observational arm carries into the summary estimate, since women choosing or offered CGM may differ systematically. And a range of '8 to 35%' spanning four separate outcomes is a summary of several different estimates, not a confidence interval around one.
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.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for obstetrics & gynaecology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free