剖宫产术后足月妊娠阴道试产预测模型初探

Development of a model for predicting success of trial of labor after cesarean delivery at term

目的建立适合我国剖宫产术后足月妊娠阴道试产风险的预测模型。方法回顾性分析2012年1月1日至2014年12月31日在深圳市妇幼保健院产科住院分娩的所有剖宫产术后阴道试产(TOLAC)足月妊娠孕妇的病历资料,利用Logistic回归分析方法,筛选出剖宫产术后阴道分娩(VBAC)的影响因素,建立预测模型,进行模型拟合优度检验,并绘制受试者工作特征曲线(ROC)。结果共纳入研究对象531例,其中阴道分娩组448例,试产失败转剖宫产组83例。试产成功率为84.4%,子宫破裂2例(0.38%)。多因素Logistic回归分析显示剖宫产妊娠间隔时间、新生儿出生体重以及胎膜早破是TOLAC失败的独立影响因素,其OR值分别为2.79、1.002和0.244(P均<0.05)。Logistic回归预测方程表达式为:P=1/[1+exp(2.4×新生儿出生体重+1.03×剖宫产妊娠间隔时间-1.41×胎膜早破-10.24)]。通过Hosmer-lemeshow检验,该模型的拟合程度好(χ~2=123.45,P=0.996),模型总的预测准确率为86.77%。ROC曲线下面积(AUC)为0.794(χ~2=65.79,P=0.000)。结论新生儿出生体重越大、剖宫产妊娠间隔时间>7年、无胎膜早破是TOLAC失败的独立危险因素。由这三个因素建立的剖宫产术后再次妊娠阴道试产风险预测模型预测效能较好,准确率较高。

ObjectiveTo establish a suitable model for predicting the success of trial of labor after cesarean section （TOLAC） during the pregnancy at term. MethodsData for all deliveries at term with one cesarean delivery history in Shenzhen Maternity and Child healthcare hospital during 1 January 2012 to 31 December 2014 were reviewed. Variables associated with VBAC were identified and used to build a model to predict the outcome of TOLAC with multivariable logistic regression. Godness of fit and accuracy of the model were evaluated by ROC. ResultsA total of 531 women met inclusion criteria and underwent TOLAC. Of the women who underwent trial of labor, 448 （84.4%） had a successful VBAC, 83 failed, and 2 （0.38%） had uterine rupture. Multivariable logistic regression analysis showed that previous cesarean section （CS） time interval, neonatal birth weight （BW） and premature rupture of membranes （PROM） were independent factors affecting TOLAC outcome, and their Odds Ratios were 2.79, 1.002 and 0.244, respectively. The Logistic regression model was expressed as follows： P=1/[1＋ exp （2.4×neonatal BW＋ 1.03× previous CS time interval -1.41×PROM -10.24）]. The Hosmer-lemeshow test showed that the model fitted well （χ2=123.45, P=0.996）, and the prediction accuracy of the model was 86.77%. The model performed well with an AUC of 0.794 （P=0.000）. ConclusionsA predictive model, which contains three variables （previous CS time interval, neonatal BW and PROM）, has been developed and its prediction efficiency and accuracy are satisfactory. The larger birth weight, the longer time interval from previous CS, and the absence of PROM are more likely to be failed in TOLAC.