Objective To study prospectively the prediction power, at different gestations, of clinical and ultrasound measurements for fetal size in diabetic pregnancy.
Setting A large combined obstetric diabetic clinic in a teaching hospital.
Participants One hundred and eighty-one pregnancies in which women had scans at least two of three specific time points and who were delivered of singletons after 34 weeks: 73% were pre-gestational insulin-dependent diabetics, the others were pre-gestational White class A or gestational diabetics.
Interventions Clinical estimates of fundal height and fetal size and ultrasound estimates of abdominal circumference and head circumference were routinely carried out at gestational ages of 28, 34 and 38 weeks or before delivery.
Main outcome measures Standardised birthweight, corrected for gestation and parity. The relation with clinical and ultrasound measurements was investigated using multiple linear regression and the capability of the measurements to predict macrosomic births (>95th centile of normals) using receiver-operator characteristic curves.
Results All measurements are poor predictors of eventual standardised birthweight. Prediction improves with closeness to delivery. Prediction is significantly improved by adding ultrasound to clinical information, but at 34 weeks or later this only contributes 8 % of the variance. There is no difference in the prediction power for macrosomia between clinical and ultrasound measurements.
Conclusions Even regular serial scanning and clinical examination will not always diagnose the macrosomic fetus in diabetic pregnancy. In our hands, clinical examination is as predictive as ultrasound measurements. Ultrasound does add to clinical prediction power but only to a small extent. Ultrasound should be used in a selected way, as defined by clinical findings, and with recognition and understanding of the errors and biases involved.