Objective: To validate the use of placenta accreta risk-antepartum (PAR-A) score as a predictive tool of clinical outcomes of placenta accreta spectrum (PAS).
Methods: This is a prospective study, conducted in six PAS specialized centers in six different countries. The study was conducted between October 1, 2020 and March 31, 2021. Women who were provisionally diagnosed with PAS during pregnancy were considered eligible. A machine-learning-based PAR-A score was calculated. Diagnostic performance of the PAR-A score was evaluated using a receiver operating characteristic curve, for perioperative massive blood loss and admission to intensive care unit (ClinicalTrials.gov identifier NCT04525001).
Results: Of 97 eligible women, 86 were included. PAS-associated massive blood loss occurred in 10 patients (11.63%). Median PAR-A scores of massive blood loss in the current cohort were 8.9 (interquartile range 6.9-14.1). In predicting massive blood loss, the area under the curve of PAR-A scores was 0.85 (95% confidence interval [CI] 0.74-0.95), which was not significantly different from the original cohort (P = 0.2). PAR-A score prediction of intensive care unit admission was slightly higher compared with the original cohort (0.88, 95% CI 0.81-0.95; P = 0.06).
Conclusion: PAR-A score is a novel scoring system of PAS outcomes, which showed external validity based on current data.
Keywords: machine learning; obstetric emergency; obstetric hemorrhage; placenta previa; placental morbidity; risk stratification.
© 2021 International Federation of Gynecology and Obstetrics.