Clinical Risk Factors and a Prediction Model for Placenta Accreta Spectrum Among Women Without Prior Cesarean Delivery: A Single-Center Cohort Study
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Placenta accreta spectrum (PAS) is an important cause of severe maternal morbidity. Although prior cesarean delivery is a well-established risk factor, PAS also occurs in women without prior cesarean section (CS), in whom risk may be underestimated. This study evaluated routinely available clinical factors associated with PAS in this population and developed a clinical-history-based prediction model.
Methods
We conducted a retrospective cohort study of women without prior CS who delivered at Peking University Third Hospital, China, from January 1, 2022, to December 31, 2023. PAS was diagnosed according to the 2019 International Federation of Gynecology and Obstetrics clinical and/or histopathological criteria. Multivariable logistic regression was used to identify independent risk factors. Model performance was assessed using receiver operating characteristic curves, calibration, decision curve analysis, and stratified 5-fold cross-validation. Analyses were repeated after stratification by placenta previa status.
Results
Among 11,148 women without prior CS, 236 had PAS. Independent risk factors in the overall cohort were placenta previa, operative hysteroscopy, uterine curettage, in vitro fertilization, and multifetal pregnancy. The overall clinical prediction model showed an area under the curve of 0.838 (95% confidence interval, 0.81-0.87), with stable performance in internal validation. In stratified analyses, model discrimination was lower among women without placenta previa (area under the curve, 0.734) and those with placenta previa (area under the curve, 0.647).
Conclusions
In this single-center cohort, routinely available clinical history was associated with PAS risk among women without prior CS. The proposed model may help identify patients who warrant targeted PAS imaging or specialist assessment, but external validation and integration with imaging features are needed before broad clinical implementation.
