• Jung-Won Kim, Yoon-Kyung Lee, Ji-Hyun Chin, Seon-Ok Kim, Mi-Young Lee, Hye-Sung Won, and Woo-Jong Choi.
• Department of Anesthesiology and Pain Medicine, Catholic Kwandong University of College of Medicine, International St. Mary's Hospital, Incheon Metropolitan City, Korea.
• J Anesth. 2017 Aug 1; 31 (4): 593-600.

PurposeIt is important to predict massive postpartum hemorrhage in patients with placenta previa totalis (PPT) and a method that accurately predicts this event is needed. The present study developed a scoring system that predicts massive transfusion in patients with PPT.MethodsThis single-center retrospective cohort study comprised 238 patients with PPT who underwent caesarean section between January 2004 and December 2010. Massive transfusion was defined as the transfusion of ≥8 units of packed red blood cells within 24 h after delivery. Multivariate regression analysis was used to estimate the risks of massive transfusion. A probability score model was then constructed and tested for performance. Subsequently, the model was validated in other patients with PPT (n = 117).ResultsThirty-one patients (13.0%) underwent massive transfusion. Ultrasound suspicion of placental adhesion, previous caesarean section, gestational age <37 weeks, sponge-like appearance of the cervix, and anterior placenta were all independent predictors of massive transfusion. The performance for the score model revealed good calibration (Hosmer-Lemeshow chi-squared 1.64; P = 0.44), and its discrimination (the area under the receiver operating characteristic for this model was 0.84) was better than when suspicion of placental adhesion was used alone (0.67; P < 0.001). In the validation set, the performance was 0.88.ConclusionThe scoring system developed using the five independent risk factors had better performance to predict massive transfusion in patients with PPT than when suspicion of placental adhesion was used alone. However, further large-scale studies are warranted to clarify the usefulness and accuracy of this model.

20 keywords...

hide…