产后出血危险因素回顾性分析及风险预测模型的建立

Retrospective analysis of risk factors for postoperative hemorrhage and establishment of risk prediction model

目的 分析产后出血危险因素,建立产后出血风险预测模型并评估其预测效能。方法 收集2017—2022年于该院定期产检的884例产妇临床资料,将472例分娩并发生产后出血的产妇纳入出血组,另外516例分娩且未发生产后出血的产妇纳入未出血组。采用单因素分析、Lasso回归分析、logstic多因素回归模型筛选产后出血的影响因素,建立产后出血风险列线图预测模型,并采用校准曲线、受试者工作特征(ROC)曲线及决策曲线分析(DCA)评估模型的预测效能。结果 472例产后出血产妇中多胎妊娠36例,流产及引产68例,最终368例纳入出血组。单因素分析显示,既往子宫手术史、合并凶险性前置胎盘、胎盘滞留、本次分娩剖宫产、孕次≥3次、既往流产史、巨大儿、合并妊娠期糖尿病、合并妊娠期高血压疾病、合并自身免疫系统疾病对产后出血有影响(P<0.05)。进一步利用Lasso回归分析筛选logistic多因素回归分析变量,结果显示:10个危险因素全部被纳入多因素logistic回归分析,最终筛选出既往子宫手术史、合并凶险性前置胎盘、巨大儿、合并妊娠期高血压疾病、既往流产史、合并自身免疫系统疾病6个产后出血的独立危险因素(P<0.05),本次分娩剖宫产为产后出血的保护因素(P<0.05)。基于独立影响因素绘制产后出血风险列线图预测模型,其校正曲线与理想曲线较为接近,ROC曲线下面积(AUC)为0.853(95%CI:0.826~0.879),DCA曲线及临床影响曲线显示阈值在0.2以上,模型具有较好的净获益,表明模型具有较好的应用价值。结论 建立的列线图预测模型可较好地预测产后出血发生风险,具有一定临床实用价值。

Objective To analyze the risk factors of postpartum hemorrhage,and to establish a postpartum hemorrhage risk prediction model and evaluate its prediction efficiency.Methods Clinical data of 884 pregnant women who had regular antenatal check-ups in the hospital from 2017 to 2022 were collected.A total of 472 women who gave birth and suffered from postpartum hemorrhage were included in the hemorrhage group,and 516 women who gave birth and did not suffer from postpartum hemorrhage were included in the non-hemorrhage group.Univariate analysis,Lasso regression analysis and logistic multifactor regression model were used to screen the influencing factors of postpartum hemorrhage,and a nomogram model for postpartum hemorrhage risk prediction was established.Calibration curve,receiver operating characteristic(ROC) curve and decision curve analysis(DCA) diagram were used to evaluate the diagnostic efficiency of the prediction model.Results Among the 472 cases of postpartum hemorrhage,36 cases of multiple pregnancy,68 cases of abortion and induced labor,and 368 cases were included in the hemorrhage group.Univariate analysis showed that the previous history of uterine surgery,dangerous placenta previa,placenta retention,cesarean section in this delivery,≥3 pregnancies,previous abortion history,macrosomia,gestational diabetes mellitus,gestational hypertension disease,and autoimmune disease had an impact on postpartum hemorrhage(P<0.05).Lasso regression analysis was further used to screen the variables of logistic multivariate regression analysis,and the results showed that all 10 risk factors were included in multivariate logistic regression analysis,six independent risk factors for postpartum hemorrhage including previous history of uterine surgery,dangerous placenta previa,macrosomia,pregnancy-induced hypertension,previous abortion history,and autoimmune diseases were finally screened out(P<0.05),and cesarean section in this delivery was a protective factor for postpartum hemorhage(P<0.05).A nomogram model for predicting postpartum hemorrhage risk was developed based on independent influencing factors,the calibration curve was close to the ideal curve,and the area under ROC curve(AUC) was 0.853(95%CI:0.826-0.879),DCA curve and clinical impact curve showed the threshold value above 0.2,the model had a good net benefit,indicating that the model had good application value.Conclusion The nomogram model of postpartum hemorrhage risk prediction can predict the risk of postpartum hemorrhage well,and has certain clinical value.