Abstract
BACKGROUND: Although pregnancy complicated by liver cirrhosis is rare, women with cirrhosis experience increased adverse pregnancy outcomes. This study aimed to evaluate pregnancy outcomes in women with liver cirrhosis and develop a predictive model using maternal factors for preterm birth in such pregnancies. METHODS: A retrospective analysis was conducted on pregnancy outcomes of a cirrhosis group (n = 43) and a non-cirrhosis group (n = 172) in a university hospital between 2010 and 2022. Logistic regression evaluated pregnancy outcomes, and a forward stepwise logistic regression model was designed to predict preterm birth in pregnant women with cirrhosis. The model's predictive performance was evaluated using the receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC). RESULTS: The incidence of cirrhosis during pregnancy was 0.06% (50/81,554). Pregnant women with cirrhosis faced increased risks of cesarean section, preterm birth, intrahepatic cholestasis of pregnancy, thrombocytopenia, and postpartum hemorrhage. In pregnant women with cirrhosis, preterm birth risk significantly increased at an incidence rate of 46.51% (20/43). According to the prediction model, the key predictors of preterm birth in pregnant women with cirrhosis were intrahepatic cholestasis of pregnancy and total bilirubin. The model demonstrated accurate prediction, with an AUC of 0.847, yielding a model accuracy of 81.4%. CONCLUSIONS: Pregnant women with cirrhosis face a heightened risk of adverse obstetric outcomes, particularly an increased incidence of preterm birth. The preliminary evidence shows that the regression model established in our study can use the identified key predictors to predict preterm birth in pregnant women with cirrhosis, with high accuracy.