A RISK-PREDICTION MODEL FOR PLACENTA ACCRETA SPECTRUM SEVERITY FROM STANDARDIZED ULTRASOUND MARKERS

We aimed to generate a model to predict the risk of a woman having normal, abnormally adherent (AAP) or abnormally invasive placentation (AIP) based on the presence of recently codified ultrasound (US) markers and disease definitions of placenta accreta spectrum (PAS). We recruited women with anterior low-lying placenta or placenta previa and a history of previous caesarean delivery to a prospective cohort study. US markers of abnormal placentation were recorded on a standardized pro forma. The presence and International Federation of Gynecology and Obstetrics grade of PAS was evaluated clinically and histologically at delivery. Markers demonstrating a predictive relationship to PAS were incorporated into a logistic regression model. A total of 106 women were included, of whom 42 (40%) were normal, 24 (23%) had AAP and 40 (38%) had AIP. A model including just four key variables (loss of clear zone, abnormal placental lacunae, placental bulge and blad-der wall interruption) was shown to reliably predict presence and severity of PAS, with an optimism-corrected C-index of 0.901. A simple model incorporating four US markers can predict likelihood and severity of PAS with high accuracy. This is the first time this has been demonstrated using the recently codified definitions of the US signs and disease definitions. Further work will see our model applied prospectively to a large patient cohort, ide-ally through a smartphone-based application, for external validation. (E-mail: william.sargent@nhs. net) (c) 2022 World Federation for Ultrasound in Medicine & Biology. All rights reserved.

Automated summary

The study aimed to develop a predictive model for the risk of normal, abnormally adherent, or abnormally invasive placentation based on ultrasound markers and disease definitions. A prospective cohort study recruited women with specific placenta conditions and recorded ultrasound markers. The presence and severity of placenta accreta spectrum (PAS) were evaluated, and a logistic regression model was created. The model, using four key ultrasound markers, showed high accuracy in predicting PAS presence and severity.