Clinical vs. sonographic estimate of birth weight in term parturients. A randomized clinical trial

To determine the relative accuracy of clinical and sonographic birth weight estimation among term parturients (> or = 37 weeks) and to assess the performance of the two techniques in identifying newborns with weights of < 2,500 g vs. > or = 2,500 g or < 2,500 g vs. at least 4,000 g.

The sample size for this randomized clinical trial was based on the assumption that 50% of clinical predictions are within 10% of birth weight. Thus, 700 parturients were necessary to show a difference of 10% with sonographic estimates (alpha = .05, beta = .02). Inclusion criteria were singletons with a reliable gestational age of > or = 37 weeks, admitted for delivery and with no known fetal anomalies. Physicians who were unaware of previous sonographic estimates obtained the estimates. Student t and chi 2 tests were used; relative risk (RR) and 95% confidence intervals (CIs) were calculated. Receiver-operating characteristic (ROC) curves were constructed to compare the two techniques' ability to differentiate between abnormal (birth weight < 2,500 g and > 4,000 g) and normal (2,500-3,999 g). P < .05 was considered significant.

Over 30 months, 758 term parturients were recruited; of them, 391 had clinical estimates and 367, sonographic. The two groups were similar in gestational age, prepregnancy and intrapartum body mass index, station of the presenting part, actual birth weight and frequency of newborns with weights < 2,500 g or > or = 4,000 g. Predictions based on clinical examination were significantly more likely to be within 10% of actual weight (58%) than those derived from ultrasound examination (32%; P < .0001; RR, 1.65; 95% CI, 1.43, 1.69). The areas under the ROC curves indicated that both techniques had a similar ability to differentiate normally and abnormally grown fetuses (P > .05).

Among term parturients, clinical estimates had significantly higher accuracy than ones derived sonographically.