New score to optimize ultrasound fetal weight estimation: a pilot study

Abstract

Introduction: Fetal weight estimation is important for obstetrical decisions; accuracy of ultrasound fetal weight estimations has not been studied. Objective: To determine ultrasound fetal weight estimations precision by developing a score based on clinical, ultrasound and maternal parameters. Design: Cross sectional study. Setting: Guillermo Almenara Irigoyen National Hospital, Lima, Peru. Participants: 130 pregnant women and their fetuses. Interventions: Clinical, ultrasound and maternal data were collected. We dichotomized predictions as actual birth weight less or equal to ±10% (correct) and more than ±10% (erroneous). We analyzed clinical, ultrasonographic and maternal variables using ROC curves in order to identify those with more influence in ultrasound accuracy. With such identified variables we developed a logistic regression model to establish a score that would estimate the probability of variation, having a ±10% of real fetal birth weight cut point. Main outcome measures: Precision of a score that relates gestational age and both fetal abdominal circumference and biparietal diameter, clinical fetal weight estimation and maternal weight. Results: The resulting score had a good capacity of discrimination of ultrasonographic fetal weight estimations according to their variation to ±10% of real birth weight (area under the curve ROC: 0,76 p<0,001). Three groups resulted: high performance 20%, conventional performance 44,4%, and low performance 71,2%. Conclusions: The new score proposed estimates the probability of variation of fetal weight above ±10% of real birth weight, with three groups of performance: high, conventional and low.