胎儿心脏大血管畸形超声漏诊病例特征及可能风险因素分析

Missed Diagnosis of Fetal Cardiac Great Vessel Anomalies by Ultrasound:Case Characteristics and Analysis of Potential Factors

目的探讨胎儿心脏大血管畸形超声漏诊病例临床特征,分析其漏诊发生可能风险因素。方法2013年5月—2023年5月行产前胎儿心脏筛查孕妇15462例,其中真阳性胎儿386例,假阴性胎儿58例,漏诊率13.06%。将444例孕妇按照4∶1进行随机分配,352例归入训练集,92例归入验证集。将训练集分为真阳性组(n=308)、假阴性组(n=44)。另外选取60例正常胎儿作为对照组,真阳性胎儿183例作为观察组。采用多因素Logistic回归分析和风险比例模型分析影响胎儿心脏大血管超声诊断结果假阴性的独立危险因素、计算一致性指数并构建预测模型,通过受试者工作特征(ROC)曲线、十字交叉试验和Nagelkerke R2指标评估模型效能。结果无心电图监测、无妊娠早期血流追踪、无血流频谱监测、非四腔心(4CV)+左心室流出道(LVOT)+右心室流出道(RVOT)+三血管(3VV)+三血管气管(3VT)心动图切片类型为影响胎儿心脏大血管畸形超声诊断结果假阴性的独立危险因素(P<0.05,P<0.01),由此4项指标构建的预测模型在训练集和验证集中具有最高的一致性指数,分别为0.792、0.789。模型评价结果显示ROC曲线下面积在训练集和验证集中分别为0.847(95%CI 0.812,0.879,P<0.05)、0.857(95%CI 0.832,0.891,P<0.05),模型区分度良好;十字交叉试验结果显示训练集和验证集各参数均十分接近,认为预测模型稳定性较高;Nagelkerke R2=0.627,认为模型拟合优度较高,预测能力较强。结论无心电图监测、无妊娠早期血流追踪、无血流频谱监测、非4CV+LVOT+RVOT+3VV+3VT心动图切片类型为影响胎儿心脏大血管畸形超声诊断结果假阴性的独立危险因素。由心动图切片类型、妊娠早期血流追踪、血流频谱监测、心电图监测4项指标构建的预测模型具有较高的区分度、稳定性和预测能力。

Objective To explore the clinical characteristics of missed diagnosis of fetal cardiac great vessel anomalies by ultrasound and to analyze the possible risk factors for missed diagnosis.Methods A total of 15462 pregnant women who underwent prenatal fetal cardiac screening in our hospital from May 2013 to May 2023 were enrolled,including 386 true positive fetuses and 58 false negative fetuses,with a missed diagnosis rate of 13.06%.Among them,444 patients were randomly assigned in a 4∶1 ratio,with 352 assigned to the training set and 92 to the validation set.The training set was divided into a true positive group(n=308)and a false negative group(n=44).Additionally,60 normal fetuses were selected as the control group,and 183 true positive fetuses were included in the observation group.Multivariate logistic regression analysis and risk ratio modeling were used to analyze the independent risk factors affecting the false negative results of fetal cardiac great vessel anomaly by ultrasound diagnosis.The consistency index was calculated and a predictive model was constructed.The performance of a model was evaluated using receiver operating characteristic(ROC)curves,cross-validation and Nagelkerke R2 index.Results The absence of electrocardiographic monitoring,early pregnancy blood flow tracking,blood flow spectral monitoring,and non-4CV+LVOT+RVOT+3VV+3VT cardiac image slice type were identified as independent risk factors affecting false negative results of fetal cardiac great vessel anomaly by ultrasound diagnosis(P<0.05,P<0.01).The predictive model constructed using these four indicators had the highest C index in the training set and the validation set,with values of 0.792 and 0.789,respectively.The results of the model evaluation showed the area under the ROC curve(AUC)was 0.847(95%CI 0.812,0.879,P<0.05)in the training set and 0.857(95%CI 0.832,0.891,P<0.05)in the validation set,indicating good discrimination of the model.The results of the cross-validation experiment showed that the parameters for both the training and validation sets were very close,indicating a high stability of the model.The Nagelkerke R2 value was 0.627,indicating a goodness of fit and strong predictive capability of the model.Conclusion The absence of electrocardiographic monitoring,early pregnancy blood flow tracking,blood flow spectral monitoring,and non-4CV+LVOT+RVOT+3VV+3VT cardiac image slice type are independent risk factors affecting false negative results of fetal cardiac great vessel anomaly by ultrasound diagnosis.The predictive model constructed using these four indicators,cardiac slice type,early pregnancy blood flow tracking,blood flow spectral monitoring,and electrocardiogram monitoring,has good discrimination and accuracy.