Accuracy of Sonographically Estimated Fetal Weight Near Delivery in Pregnancies Complicated With Diabetes Mellitus

Objectives—The purpose of this study was to evaluate the accuracy of sonographic estimations of fetal weight (FW) and signed percent error between pregnant patients with and without diabetes mellitus (DM). Methods—We conducted a retrospective cohort study of all singleton nonanom-alous live births who delivered after 34 weeks and received a sonographic estimation of FW within 2 weeks of delivery at the University of Cincinnati Medical Center between 2008 and 2011. Our primary outcome compared the DFW and signed percent error between DM and non-DM pregnancies. Sensitivity and specificity were calculated for the prediction of FW greater than 4000 g in each study group. Linear regression analysis assessed correlation coefficients, R² val-ues, and variance of the DFW by live birth weight. Results—The mean DFWs were 62 and 103 g for non-DM and DM pregnancies, respectively (P =.04). However, the signed percent error (mean ± SD, 1.7% ± 9.8% versus 2.6% ± 9.9%; P =.15) was similar between the study groups. Linear regression comparing the DFW to the live birth weight revealed a weak correlation in DM (r = 0.34; R² = 0.11) and non-DM pregnancies, (r = 0.17; R² = 0.03) preg-nancies. Overall sensitivity for the prediction of FW greater than 4000 g was poor (0.41 and 0.62 in non-DM and DM pregnancies). However, the specificity was high (0.97 and 0.99 for both groups). Conclusions—Although DM alters the biometric measurements of the fetus with increasing thoracoabdominal size, there are no clinically significant alterations in the accuracy of sonography for FW prediction when performed near delivery. Sonography is highly specific for birth weight greater than 4000 g, which is helpful for delivery planning and management.