Could masking gestational age estimation during scanning improve detection of small-for-gestational-age fetuses? A...

Antenatal detection of small-for-gestational-age fetuses improves outcomes and reduces perinatal mortality rates. However, ultrasonographic estimation of fetal weight is subject to several potential sources of error. One potential source of error is subconscious operator bias towards "normal" measurement values for gestational age (observer-expectancy bias).

We aimed to determine whether the sensitivity of small-for-gestational-age detection is improved by removing real-time display of estimated gestational age during measurement of the abdominal circumference in the third trimester.

This retrospective evaluation (November 2014-May 2018 inclusive) included all singleton infants liveborn at ≥28 weeks gestation in a single United Kingdom obstetrics center. In the preintervention phase, real-time estimated gestational age was displayed to sonographers as they measured fetal abdominal circumference (the key determinant of estimated fetal weight with the use of the INTERGROWTH 21st fetal weight equation) in the third trimester. In the postintervention phase, real-time gestational age information was removed on selected ultrasound machines. Accuracy of birthweight percentile estimation was assessed before and after intervention, both in the full cohort comprising all eligible scans and in a subcohort that was scanned within 4 weeks of delivery. We assessed the accuracy of small-for-gestational-age detection using the sensitivity, positive likelihood ratio, and area under the receiver-operator curve.

Of the 18,342 eligible pregnancies, 9342 (51%) had a third-trimester growth scan. The sensitivity of ultrasonographic estimation of fetal weight for antenatal detection of small-for-gestational-age babies did not change significantly between the before and after intervention phases (31.5% confidence interval, 27.1-36.2 vs 31.7% confidence interval, 20.2-45.0). Although the sensitivity for small-for-gestational-age detection was higher in the subcohort that was scanned within 4 weeks of delivery than in the full cohort (P<.001), there was no significant difference between the before and after intervention phases (58% confidence interval, 50-66 vs 65% confidence interval, 43-84). With the use of an estimation of the abdominal circumference percentile rather than estimated fetal weight percentile significantly decreased the sensitivity for small-for-gestational-age detection in all groups (P<.01), but there was no difference between the before and after intervention phases.

Blinding operators to the estimated gestation of the fetus during abdominal circumference measurement does not significantly alter the antenatal detection rate of small-for-gestational-age babies. The observer-expectancy effect is therefore unlikely to be a significant contributor to the error that is associated with ultrasonographic estimation of fetal weight.