- Design
- retrospective cohort study at a tertiary referral centre with modified Poisson regression, clinicians unblinded to ultrasound findings
- Population
- 21,743 singleton pregnancies undergoing routine 35-37 week ultrasound; screen-positive defined as estimated fetal weight at or above the 90th centile
- Primary outcome
- diagnostic accuracy for large-for-gestational-age and macrosomia, mode of delivery, and composite adverse maternal and neonatal outcomes
- Effect
- sensitivity 34.9% (33.2-36.6) LGA and 35.6% (33.5-37.9) macrosomia; screen-positive intrapartum caesarean aRR 1.47 (1.30-1.67); false positive vs true negative maternal composite aRR 1.28 (1.11-1.48), neonatal aRR 1.35 (0.72-2.53)
This retrospective cohort followed 21,743 singleton pregnancies through routine 35-37 week ultrasound at a tertiary centre, with clinicians unblinded to the results — as they are in practice. Sensitivity for large-for-gestational-age was 34.9% (95% CI 33.2 to 36.6) and for macrosomia 35.6% (33.5 to 37.9); specificity was 97.4% and 95.6%. So the scan misses about two-thirds of large babies while rarely calling a normal-sized one large.
What happened next is the point. Screen-positive pregnancies were less likely to attempt labour (adjusted RR 0.87, 95% CI 0.84 to 0.90) and more likely to end in intrapartum caesarean (aRR 1.47, 95% CI 1.30 to 1.67), with more composite adverse maternal (aRR 1.43, 95% CI 1.32 to 1.55) and neonatal outcomes (aRR 2.37, 95% CI 1.85 to 3.05).
The comparisons that isolate the label from the baby are the ones that matter. False positives — normal-sized babies called large — had more operative vaginal delivery (aRR 1.29, 95% CI 1.07 to 1.57) and more maternal morbidity (aRR 1.28, 95% CI 1.11 to 1.48) than true negatives, with no difference in neonatal outcomes (aRR 1.35, 95% CI 0.72 to 2.53). False negatives — genuinely large babies not identified — had less induction and less caesarean than true positives, and no excess neonatal harm (aRR 0.77, 95% CI 0.56 to 1.06). Being labelled large caused intervention; not being labelled did not cost the baby anything measurable.
This is observational, from one tertiary centre, and confounding by indication cannot be excluded — but the false-positive and false-negative comparisons are specifically designed to test the label rather than the size, and they are hard to explain away. It supports moving from universal term growth scanning to a risk-based approach, and where a scan is done, treating a raised estimate as weak evidence rather than as a diagnosis.
- Question universal third-trimester growth scanning at term in low-risk women; consider a risk-based approach
- Where a scan is done, weight the estimate as poor evidence — it misses about two-thirds of large babies
- Do not let a screen-positive result by itself steer a woman away from attempting labour
- Audit your own unit's caesarean rate among screen-positive but normally grown babies
- This is observational from a single tertiary centre; the false-positive comparison, not the headline, is what carries it
The statistics, in plain English
Sensitivity of 34.9% means the test misses roughly two out of three large babies; the high specificity means it rarely mislabels a normal baby, but with large babies uncommon, a positive result is still frequently wrong. The neonatal composite in screen-positives (aRR 2.37) is not evidence of harm from screening, because those pregnancies genuinely include more large and complicated babies — that is why the false-positive comparison matters, and there the neonatal interval (0.72 to 2.53) shows no difference while the maternal one (1.11 to 1.48) does. Adjusted relative risks from a retrospective cohort where clinicians acted on the test cannot fully separate the label from the reasons it was assigned.
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