盆底超声在孕期尿失禁中的预测价值

Predictive value of pelvic floor ultrasound for urinary incontinence during pregnancy

目的:探讨盆底超声对孕期尿失禁(Urinary Incontinence,UI)的预测价值。方法:对2016年7月—2017年1月于两家三甲医院产科门诊就诊的358名孕妇进行问卷调查及经会阴盆底三维超声检查。对单因素分析有组间差异的指标,采用受试者工作特征(ROC)曲线分析预测价值,多因素logistic回归分析建立联合预测模型,比较ROC曲线下面积(AUC)>0.7的单个指标及联合预测模型的预测效能。结果:排除无效数据16例,共342名孕妇纳入分析。孕期UI的发生率为39.47%。单个超声指标的AUC在0.576~0.716之间,预测效能一般。其中,静息状态下的β角、裂孔面积(hiatalarea,HA),Valsalva状态下的膀胱颈位置,膀胱颈的下移距离的预测效能较差(AUCs≥0.6,P<0.05)。Valsalva状态下的HA及静息状态与Valsalva状态下HA的差值(ΔHA)的AUC均>0.7,预测效能可接受(P<0.05)。通过logistic逐步回归分析,静息状态下的β角及ΔHA纳入回归模型,ROC曲线分析发现模型联合预测的AUC为0.721(P<0.05)。当比较Valsalva状态下的HA、ΔHA及联合预测模型的AUC时,均未发现存在统计学差异(P>0.05)。结论:静息状态下的β角及HA,Valsalva状态下的膀胱颈位置及HA,膀胱颈的下移距离,ΔHA均可用于孕期UI的预测,未发现盆底超声联合预测效能优于单个指标的预测效能。

Objective:To assess the predictive value of pelvic floor ultrasound for urinary incontinence(UI)during pregnancy.Methods:From July 2016 to January 2017,358 pregnant women from two hospitals underwent an interview and pelvic floor transperineal ultrasound assessment.For the parameters with difference between groups by single factor analysis,ROC curve was used to analyze the predictive value,and multivariate logistic regression analysis was used to establish the joint predictive model.The predictive value of the predictive model and single parameter with area under curve(AUC)>0.7 was compared.Results:After excluding 16 women with invalid data,342 pregnant women were included.The prevalence rate of UI during pregnancy was 39.47%.The AUC of single parameter ranged from 0.576 to 0.716,and the predictive efficiency was general.The predictive efficiency ofβangle,hiatal area(HA)at rest,bladder neck vertical position(BNVP)at Valsalva maneuver(VM),bladder neck descent(BND)was poor(AUCs≥0.6,P<0.05).The AUCs of HA at VM and the difference(ΔHA)between HA at rest and HA at VM were>0.7,and the predictive efficiency was acceptable(P<0.05).Theβangle andΔHA at rest were included in the regression model by the logistic regression analysis.The ROC curve analysis showed that the AUC predicted by the model was 0.721(P>0.05).No statistical difference was found when comparing the AUCs of HA at VM,ΔHA and the combined prediction,respectively(P>0.05).Conclusions:βangle,HA at rest,BNVP,HA at VM,BND andΔHA can be used for predicting UI during pregnancy.It is not found that pelvic floor ultrasound combined prediction is superior to single parameter prediction.