宫颈锥切术后孕妇早产预测模型的建立

目的: 建立宫颈锥切术后孕妇发生早产的预测模型，并对模型的临床价值进行初步评估。方法: 回顾性分析浙江大学医学院附属妇产科医院有宫颈锥切术史的73名孕妇妊娠病历资料，以此为训练数据集采用Logistic回归分析筛选早产的影响因素并建立预测模型。选择宫颈锥切术后45名孕妇作为验证数据集，采用ROC曲线进行宫颈锥切术后孕妇早产预测模型的临床价值评估。结果: 对于宫颈锥切术后孕妇，年龄（OR=1.20，95% CI：1.01~1.43，P < 0.05）和孕中期子宫颈长度（OR=0.06，95% CI：0.01~0.21，P < 0.01）是早产的独立预测因素。通过Logistic回归分析建立回归模型Logit（P）=1.408-2.903×孕中期子宫颈长度+0.186×孕妇年龄。训练数据集使用构建的模型预测早产的AUC值为0.93（95%CI：0.87~0.99），该模型预测早产的最佳阈值为-1.512，敏感度、特异度、约登指数、阳性预测值、阴性预测值、准确率分别为91.7%、81.5%、0.732、68.8%、95.7%、84.5%；测试数据集使用构建的模型预测早产的AUC值为0.94（95%CI：0.86~1.00），当最佳阈值为-0.099时，模型预测早产的敏感度、特异度、约登指数、阳性预测值、阴性预测值、准确率分别为92.9%、90.3%、0.832、81.3%、96.5%、91.1%。结论: 以宫颈锥切术后孕妇的年龄和孕中期子宫颈长度建立的早产预测模型可以较准确地预测早产的发生，值得临床进一步研究。
Abstract: Objective: To establish a prognostic model for preterm birth in women after cervical conization, and to evaluate its effectiveness. Methods: Seventy three women after cervical conization in Women's Hospital of Zhejiang University were included for this retrospective study. The influencing factors of preterm delivery were analyzed by Logistic regression analysis and a prognostic model was created. Receiver operating characteristic (ROC) curve was used for evaluation of the predictive ability of the model. Forty five women who underwent cervical conization were included for testing the validity of the model. Results: For women after cervical conization, mother's age (OR=1.20, 95%CI:1.01-1.43, P < 0.05) and cervical length during middle pregnancy (OR=0.06, 95%CI:0.01-0.21, P < 0.01) were independent predictors for preterm birth. The regression model was Logit (P)=1.408-2.903×cervical length+0.186×age. The areas under the ROC curve (AUC) of the training dataset was 0.93 (95%CI:0.87-0.99). The sensitivity, specificity, Youden index, positive predictive value (PPV), negative predictive value (NPV) and accuracy with the cutoff value of -1.512 were 91.7%, 81.5%, 0.732, 68.8%, 95.7% and 84.5% respectively. The AUC of the testing dataset was 0.94 (95%CI:0.86-1.00). The sensitivity, specificity, Youden index, PPV, NPV and accuracy with the cutoff value of -0.099 were 92.9%, 90.3%, 0.832, 81.3%, 96.5% and 91.1%, respectively. Conclusion: The model based on the age and cervical length during middle pregnancy can effectively predict preterm delivery in pregnant women after cervical conization. Key words: Cervical intraepithelial neoplasia/surgery Uterine cervical neoplasms/surgery Hysterectomy Postoperative period Premature birth Forecasting Models, theoretical Retrospective studies