Improving the accuracy of screening for large-for-gestational-age fetuses: a multicenter observational study.

Abstract

Objectives: Large-for-gestational-age (LGA) fetuses present significant maternal and neonatal risks. However, prenatal screening is prone to inaccuracies, leading to potentially unnecessary interventions. This study aims to evaluate the factors influencing the accuracy of third-trimester screening for LGA fetuses to improving diagnostic accuracy.

Methods: A prospective, multicenter cohort study was conducted involving low-risk pregnancies from three hospitals. Screening was analyzed using ultrasound-based fetal weight estimation (EFW), abdominal circumference (AC) and symphysial fundal height (SFH) measurements. EFW and AC were assessed either during the routine third-trimester ultrasound or during an additional growth ultrasound when available. Newborns were classified as LGA based on AUDIPOG growth curves. Screening performance was assessed using sensitivity, specificity, predictive values, diagnostic odds ratios (DOR), Youden's index and accuracy. We also evaluated composite screening tests combining biometric parameters with maternal clinical risk factors and influence of gestational age at the time of growth ultrasound to identify the optimal timing for screening.

Results: Among 2,217 women, risk factors such as high BMI and gestational diabetes increased suspicion of LGA fetuses, contributing to both true and false positive results (p<0.001). No single ultrasound parameter demonstrated superior diagnostic performance. Third-trimester ultrasound showed a sensitivity of 37 % [31-44 %] and a specificity of 94 % [93-95 %], while growth ultrasound improved sensitivity to 65 % [57-74 %] but reduced specificity to 82 % [79-85 %]. SFH measurements did not enhance screening performance. Overestimation of fetal weight was observed in 56.89 % (95/167) of cases, with errors exceeding 10 % in 26.95 % (122/167) of newborns. Combined screening using fetal biometry and maternal clinical risk factors showed high specificity but poor sensitivity, limiting their utility as standalone tools for detecting macrosomia.

Conclusions: This study underscores the impact of operator bias in LGA screening, with risk factors influencing measurements. The modest performance of ultrasound-based screening highlights the inherent limitations of current methods. These findings call for cautious labeling of LGA fetuses and development of management strategies to address the challenges of imprecise screening.