基于产科资料和血清学指标构建足月妊娠胎膜早破胎儿/新生儿并发症的预测模型

Construction of a prediction model for fetal/neonatal complications in full-term pregnancy with premature rupture of membranes based on obstetric data and serological indicators

目的基于产科资料和血清学指标构建足月妊娠胎膜早破(PROM)胎儿/新生儿并发症的预测模型,以期为临床早期预测及预防胎儿/新生儿发生并发症提供参考。方法将2019年1月至2020年12月该院收治的240例足月PROM产妇分为建模组162例和验证组78例,收集所有研究对象的临床资料并检测血清人绒毛膜促性腺激素(β-HCG)、C-反应蛋白(CRP)、金属蛋白酶-3(MMP-3)水平。根据是否发生并发症将建模组分为发生组和未发生组。采用多因素Logistic回归分析建模组足月PROM胎儿/新生儿发生并发症的影响因素,构建列线图预测模型,并绘制受试者工作特征(ROC)曲线、决策曲线对该模型进行评估验证。结果建模组中发生并发症60例(发生组),未发生并发症102例(未发生组)。发生组PROM孕周37周0~6 d、内检次数≥3次、阴道炎、破膜至分娩时间24~72 h产妇占比及血清β-HCG、CRP、MMP-3水平高于未发生组(P<0.05);多因素Logistic回归分析结果显示,PROM孕周37周0~6 d、内检次数≥3次、破膜至分娩时间24~72 h及血清β-HCG、CRP、MMP-3水平升高是足月PROM胎儿/新生儿发生并发症的独立危险因素(P<0.05)。基于以上指标构建建模组中足月PROM胎儿/新生儿发生并发症的列线图预测模型,该模型在建模组和验证组中的一致性指数分别为0.891、0.877。ROC曲线显示,该列线图在建模组和验证组中预测的曲线下面积分别为0.976(95%CI:0.858~0.990)、0.938(95%CI:0.875~0.962)。决策曲线显示,当建模组高风险阈值在0.0~0.8时、验证组高风险阈值在0.0~0.8时,该列线图具有较好的临床净收益。结论足月PROM胎儿/新生儿发生并发症的影响因素包括PROM孕周、内检次数、破膜至分娩时间和血清β-HCG、CRP、MMP-3水平,基于以上因素构建的列线图预测模型对高危人群具有一定的预测效能,可能改善妊娠结局。

Objective To construct a prediction model for fetal/neonatal complications of premature rupture of membranes(PROM)in full-term pregnancies based on obstetric data and serologic indicators,with a view to providing a reference for early clinical prediction and prevention of fetal/neonatal complications.Methods A total of 240 cases of full-term PROM maternal patients admitted to the hospital from January 2019 to December 2020 were divided into a modeling group(162 cases)and a validation group(78 cases).Clinical data of the maternal patients were collected,serum human chorionic gonadotropin(β-HCG),C-reactive protein(CRP)and metalloproteinase-3(MMP-3)levels were detected.The modeling group was subdivided into occurrence group and non-occurrence group aoccording to whether the complications occurred or not.Multivariate Logistic regression analysis were used to identify factors that contribute to fetal/neonatal complications in full-term PROM maternal patients in the modeling group,and a nomogram prediction model was constructed and evaluated by plotting the receiver receiver operating characteristic(ROC)curves and decision-making curves.Results Complications occurred in 60 cases(occurrence group)in the modeling group,and 102 cases were without complications(non-occurrence group).The proportions of pregnant women with PROM at 37 weeks and 0-6 d of gestation,with≥3 internal examinations,with vaginitis,and with a rupture of membranes to delivery time of 24-72 h,as well as serumβ-HCG,CRP,and MMP-3 levels in the occurrence group were higher than those in the non-occurrence group(P<0.05).Multivariate Logistic analysis results showed that PROM at 37 weeks and 0-6 d of gestation,with≥3 internal examinations,with rupture of membranes to delivery time of 24-72 h and elevated serumβ-HCG,CRP,MMP-3 levels were independent risk factors for fetal/neonatal complications in full-term PROM(P<0.05);based on the above indicators,a predictive nomogram model of complications in term PROM fetuses/neonates in the modeling group was constructed.The consistency indices of this model in the modeling group and validation group were 0.891 and 0.877,respectively.ROC curves showed that area under the curve predicted by this nomogram model in the modeling group and validation group were 0.976(95%CI:0.858-0.990)and 0.938(95%CI:0.875-0.962),respectively;and the decision curves showed that,when the high-risk in the modeling group threshold was 0.0-0.8,and the nomogram model had a better net clinical benefit when the high-risk threshold was 0.0-0.8 in the validation group.Conclusion The factors influencing the occurrence of complications in full-term PROM fetuses/neonates include gestational age,number of internal examinations,time from rupture of membranes to delivery and serumβ-HCG,CRP,MMP-3 levels,and the nomogram prediction model constructed based on the above factors has certain efficacy in predicting high-risk populations,which may improve the outcome of pregnancy.