Limitations in prediction of gestational age and birth weight by ultrasonographic methods

Abstract

The index of predictability of estimated gestational age (EGA) and birth weight (BW) from biparietal diameter (BPD) was tested in 1333 normal pregnancies; 96% of the patients were black. The correlation BPD and actual weeks to delivery was 0.834, r2 =69.5%. Confidence limits were established and a nomogram for EGA was derived. Our nomogram and two similar ones were compared. The overall correlation between them was good, but that for specific BPD and EGA was not. In addition, our BPD values were smaller for corresponding gestational ages in other nomograms. The overall accuracy of estimated delivery date was +/- 1.6 weeks. We were not able to predict birth weight at all from a single, random BPD measurement. We have demonstrated the serious limitations that exist in the prediction of gestational age from previously established nomograms. We infer that this error is even greater when homogenous populations are so studied. We suggest that any published BPD-EGA or BPD-BW nomogram be given close, intensive statistical scrutiny, within the population studied, before it is used. When such scrutiny is not possible, serial sonographic determinations, or longitudinal use of growth charts, would be more acceptable.