Population-specific validation of fetal weight formulas: Evidence from a multicenter near-delivery cohort.

INTRODUCTION

Accurate ultrasound estimated fetal weight (EFW) supports key perinatal decisions. This study aimed to compare the accuracy of a population-specific formula with that of a widely used global model in pregnancies examined near delivery.

MATERIAL AND

METHODS

This was a secondary analysis of a prospective Japanese cohort study to develop a new fetal ultrasound biometry reference chart. Fetal ultrasound measurements were obtained by Japan Society of Ultrasonics in Medicine-certified sonographers or under their supervision and submitted to the coordinating center, where EFW was calculated using the Shinozuka and Hadlock-3 formulas. Analyses were restricted to cases with EFW available from both formulas. The primary analysis included examinations within 7 days before delivery; a sensitivity analysis restricted the interval to ≤3 days. A subgroup analysis was performed in fetuses classified as small for gestational age (SGA).

RESULTS

In the primary analysis (n = 310), the median signed error was -75.8 g for Shinozuka and -223.3 g for Hadlock-3 (Wilcoxon p < 0.001). The median absolute percentage signed error was 4.81% for Shinozuka and 7.93% for Hadlock-3 (Wilcoxon p < 0.001). The proportion within ±10% of BW was 83.9% with Shinozuka versus 61.6% with Hadlock-3 (McNemar p < 0.001). In the sensitivity analysis (n = 176), similar findings were observed. In the SGA subgroup (n = 40), the signed error and absolute percentage error were significantly lower with the Shinozuka formula than with Hadlock-3.

CONCLUSIONS

In this multicenter cohort examined near delivery, the population-specific Shinozuka formula produced EFW closer to BW and a higher proportion within ±10% than Hadlock-3, which showed greater systematic underestimation. These findings support the use of locally validated, population-specific EFW formulas when available, particularly for clinical decision-making near delivery.