凶险性前置胎盘孕妇产后大出血风险预测模型的建立与验证

Establishment and verification of risk prediction model for postpartum hemorrhage in pregnant women with pernicious placenta previa

目的 构建凶险性前置胎盘(PPP)孕妇产后大出血风险预测模型并检验其预测效能。方法 选取2021年1月至2022年6月该院收治的196例PPP孕妇为研究对象,根据是否发生产后大出血将其分为产后大出血组(n=72)与无产后大出血组(n=124)。收集孕妇相关临床资料,采用单因素分析和多因素logistic回归分析孕妇发生产后大出血的影响因素。基于筛选的影响因素,采用R3.5.3软件绘制PPP孕妇产后大出血风险列线图预测模型,并应用受试者工作特征(ROC)曲线和Hosmer-Lemeshow分别检验其预测效能和拟合优度。另纳入2022年7月至2023年6月该院收治的98例PPP孕妇对模型进行外部验证。结果 多因素logistic回归分析结果显示,孕妇年龄增加、剖宫产次数(≥2次)、胎盘粘连、胎盘植入、前置胎盘类型(部分型/中央型)、距前次剖宫产时间(<3年)、前次剖宫产前置胎盘、前次剖宫产的医院分级(二级及以下)均是PPP孕妇发生产后大出血的独立危险因素(P<0.05)。基于以上8个因素建立PPP孕妇产后大出血风险列线图预测模型,该模型的ROC曲线下面积(AUC)为0.816(95%CI:0.795~0.838),灵敏度为78.45%,特异度为86.38%,经Hosmer-Lemeshow检验该模型具有较好的拟合度(χ^(2)=5.126,P=0.358)。外部验证:98例PPP孕妇中35例发生产后大出血,经该模型预测发生产后大出血32例,AUC为0.786(95%CI:0.723~0.801),灵敏度为72.48%,特异度为83.49%,经Hosmer-Lemeshow检验该模型具有较好的拟合度(χ^(2)=5.496,P=0.382)。结论 基于孕妇年龄、剖宫产次数、胎盘粘连、胎盘植入、前置胎盘类型、距前次剖宫产时间、前次剖宫产前置胎盘、前次剖宫产的医院分级构建的PPP孕妇产后大出血风险列线图预测模型,对于PPP产后出血有一定的临床预测价值。

Objective To construct a risk prediction model for postpartum hemorrhage in pregnant women with pernicious placenta previa(PPP)and test its predictive performance.Methods A total of 196 PPP pregnant women admitted to the hospital from January 2021 to June 2022 were selected as the study objects.According to whether postpartum hemorrhage occurred or not,they were divided into the postpartum hemorrhage group(n=72)and group without postpartum hemorrhage(n=124).The clinical data of pregnant women were collected,and the influencing factors of postpartum hemorrhage were analyzed by univariate analysis and multivariate logistic regression model.Based on the selected influencing factors,R3.5.3 software were used to draw a nomogram model for predicting the risk of postpartum hemorrhage of PPP pregnant women,and its predictive performance and goodness-of-fit was tested by receiver operating characteristic(ROC)curve and Hosmer-Lemeshow model,respectively.In addition,98 PPP pregnant women admitted to the hospital from July 2022 to June 2023 were included for external verification of the risk prediction model.Results Multivariate logistic regression analysis results showed that the increased age of pregnant women,number of cesarean sections(≥2),placenta adherence,placenta accretion,type of placenta previa(partial/central),time since last cesarean section(<3 years),placenta previa in previous cesarean section,and hospital classification of previous cesarean section(grade 2 and below)were all independent risk factors for postpartum hemorrhage in PPP pregnant women(P<0.05).Based on the above eight independent risk factors,a nomogram model was established to predict the risk of postpartum hemorrhage in PPP pregnant women.The area under the ROC curve(AUC)of the model was 0.816(95%CI:0.795-0.838),the sensitivity was 78.45%,and the specificity was 86.38%.Hosmer-Lemeshow test showed that the model had a good fit(χ^(2)=5.126,P=0.358).External validation:among 98 PPP pregnant women,postpartum hemorrhage occurred in 35 cases,and 32 cases were predicted by this model,with an AUC of 0.786(95%CI:0.723-0.801).The sensitivity was 72.48%,and the specificity was 83.49%.Hosmer-Lemeshow test showed that the model had a good fit(χ^(2)=5.496,P=0.382).Conclusion Based on the age of pregnant women,number of cesarean sections,placenta adherence,placenta accretion,type of placenta previa,time since last cesarean section,placenta previa and hospital classification in previous cesarean section,the PPP nomogram model for predicting postpartum hemorrhage risk has been established,and has certain clinical predictive value for PPP postpartum hemorrhage.