Diagnostic accuracy of ultrasound in predicting birth-weight discordance in twin pregnancy: systematic review and meta-analysis

Objective To investigate the diagnostic performance of ultrasound in predicting birth-weight (BW) discordance in twin pregnancy. Methods Electronic databases PubMed, EMBASE and CINAHL were searched using combinations of MeSH terms 'birth weight', 'discordance', 'twins' and ` ultrasound'. Summary estimates of sensitivity, specificity, positive and negative likelihood ratios and diagnostic odds ratios for the predictive accuracy of sonographically estimated fetal-weight (EFW) discordance and abdominal-circumference (AC) discordance in predicting BW discordance were computed using hierarchical summary receiver-operating characteristics curves. Results Twenty studies (including 5826 twin pregnancies) were included. EFW discordance >= 20% had a sensitivity of 65.4% and specificity of 90.8% in predicting BW discordance >= 20%. The predictive performance of ultrasound performed within 1 month, 2 weeks and 3 days before birth was 61.4%, 72.3% and 78.9%, respectively. EFW discordance >= 25% had a sensitivity of 57.7% and specificity of 95.2% in predicting BW discordance >= 25%. The sensitivity of EFW discordance >= 25% detected within 1 month, 2 weeks and 3 days before birth was 60.0%, 75.0% and 60.3%, respectively, while the corresponding values for specificity were 97.7%, 96.2% and 87.3%, respectively. In view of the different cut-offs reported among studies, it was not possible to perform comprehensive data synthesis for each AC discordance cut-off. The optimal diagnostic performance of AC discordance was for prediction of BW discordance >= 25%, with a sensitivity of 70.8% and specificity of 86.4%. Conclusion Ultrasound EFW discordance has an overall moderate accuracy in predicting BW discordance in twin pregnancy