LAUSR

Abstract Objective This study aimed to examine whether the addition of fetal growth velocities, based on serial scans, compared with estimated fetal weight (EFW) alone, improves the prediction accuracy of birth weight (BW). Study Design From the prospective National Institute for Child Health and Human Development Fetal Growth Studies, Singletons (2009–2013) that recruited women at low-risk for aberrant fetal growth, we included 2,397 participants with two to five ultrasounds between 15 and 41 weeks. Fetal growth velocity ( V ) was calculated between successive visits. Linear regression estimated the relationship between EFW and BW: non-V model, absent all growth velocities, included EFW at visit 5, interval from visit 5 to delivery, their cubic polynomials, and delivery gestational age; all-V model added all available velocities between research visits (i.e., V2 + V3 + V4 + V5) to the non-V model; single-V model added only the velocity between visits 4 and 5. Results Non-V performed the worst, with the highest mean absolute error of 197 g and the lowest R 2 , 0.63. The two best performing models were all-V and single-V, with the lowest mean absolute error (192 g for both), the highest R 2 of 0.65 (for both), and the greatest proportion of predicted BWs within 10% of actual BW (84.7% for both). Conclusion Integrating fetal growth velocities into BW estimating formulas improved BW prediction accuracy, which has the potential to guide clinical intervention decisions that rely on EFW. Key Points Decreasing the BW prediction error is essential. Single EFW assessment to predict BW is standard. Adding third-trimester growth velocity improved accuracy. Growth velocities also improved small- and large-for-gestational age predictions. These improvements reach a population level.