Suspecting a big baby at term changes management more than it changes outcomes

Design

retrospective cohort study with modified Poisson regression, clinicians unblinded to scan results

Population

21,743 singleton pregnancies having routine ultrasound at 35 to 37 weeks at a tertiary centre

Primary outcome

diagnostic accuracy for large-for-gestational-age, mode of delivery, composite adverse maternal and neonatal outcomes

Effect

sensitivity 34.9%; screen-positive intrapartum caesarean aRR 1.47 (1.30 to 1.67); no neonatal difference in false positives (aRR 1.35, 0.72 to 2.53)

A retrospective cohort of 21,743 singleton pregnancies having routine ultrasound at 35 to 37 weeks tested how well a screen-positive estimated fetal weight - at or above the 90th centile - predicted a large baby, and what happened when it was wrong. Clinicians could see the scan results, which is the point of the study.

Sensitivity was 34.9% (33.2 to 36.6) for large-for-gestational-age and 35.6% for macrosomia, with specificity around 96% to 97%: the scan misses about two-thirds of big babies. Screen-positive pregnancies were less likely to attempt labour (aRR 0.87, 0.84 to 0.90) and more likely to have an intrapartum caesarean (aRR 1.47, 1.30 to 1.67). The telling comparison is the false positives: against true negatives they had more operative vaginal delivery (aRR 1.29, 1.07 to 1.57), more intrapartum caesarean (aRR 1.25, 1.0 to 1.55) and more adverse maternal outcomes (aRR 1.28, 1.11 to 1.48), with no difference in adverse neonatal outcomes (aRR 1.35, 0.72 to 2.53). False negatives - big babies nobody suspected - had fewer inductions and caesareans and no increase in neonatal harm (aRR 0.77, 0.56 to 1.06).

That is a labelling effect: the suspicion, not the size, drove the intervention. It argues against universal late-pregnancy screening for fetal largeness and for a risk-based approach - scan where there is a reason, such as diabetes or a clinical discrepancy, and where the result will inform something other than the threshold itself. When a scan does come back at the 90th centile, the conversation should start from the sensitivity, not the centile.

• Do not order a third-trimester growth scan purely to exclude a big baby in a low-risk pregnancy
• When the estimate is high, counsel with the sensitivity: two-thirds of large babies are missed anyway
• Separate the indication for delivery from the scan result - diabetes, prior shoulder dystocia, clinical findings
• Watch your own induction threshold after a screen-positive result; that is where the intervention creeps in
• Audit local practice for the gap between suspected macrosomia and birthweight at or above 4,000 g

Why it matters

The scan is not changing what happens to the baby - it is changing what happens to the mother.

Don't overread it

A retrospective cohort at one tertiary centre cannot prove the label caused the intervention, and it is not a trial of withholding the scan.

The statistics, in plain English

Sensitivity of 34.9% means the scan identifies roughly one in three large babies; specificity near 97% means it rarely flags a normally grown one, but with large babies uncommon, most flags are still wrong. The neonatal confidence intervals in the false-positive and false-negative comparisons both cross 1.0, so no neonatal difference was shown either way - which is what makes the extra caesareans hard to justify. This is observational, and the adjusted relative risks cannot fully remove the reasons a clinician chose to intervene.

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.