Clinical Prediction of Retained Products of Conception: Combining Obstetric History and Ultrasound for Improved Accuracy in Severe Postpartum Hemorrhage

Abstract

Background The current challenge is how to improve the management of postpartum hemorrhage (PPH) to reduce the maternal mortality rate further. This study aimed to investigate whether a combined specific obstetric history and ultrasonographic findings can improve the predictive accuracy of retained products of conception (RPOC) with severe PPH. Methods This retrospective study included 56 patients who were diagnosed with RPOC. We extracted the following clinical data: obstetric history of second-trimester miscarriage, the time at which there was clinical suspicion of RPOC after the previous pregnancy (TIME), grayscale ultrasonographic finding (RPOC long-axis length [SIZE]), and color Doppler ultrasonographic finding based on the Gutenberg classification (RPOC hypervascularity). In this study, we defined cases requiring blood transfusion therapy or transcatheter arterial embolization as severe PPH. The patients were divided into two groups according to the presence or absence of severe PPH. The predictors of severe PPH were evaluated using logistic regression models. Model A comprised a combination of second-trimester miscarriage and TIME, Model B comprised a combination of Model A and long-axis SIZE, and Model C comprised a combination of Model B and RPOC hypervascularity. Results The multivariable analysis showed that long-axis SIZE was the only significant predictor of severe PPH (odds ratio [OR], 10.38; 95% confidence interval [CI], 2.06-63.86) independent of second-trimester miscarriage, TIME, and RPOC hypervascularity. The c-statistic was higher in Model C (OR, 0.863; 95% CI, 0.731-0.936) than in Model A (OR, 0.723; 95% CI, 0.551-0.847) and Model B (OR, 0.834; 95% CI, 0.677-0.923). Conclusion Combining a specific obstetric history (second-trimester miscarriage and TIME) and ultrasonographic findings (long-axis SIZE and RPOC hypervascularity) improves the predictive accuracy of RPOC with severe PPH. This prediction model may be a useful clinical screening tool for RPOC with severe PPH.