宫颈LEEP术后发生宫颈管粘连风险模型构建及验证

Construction and validation of a risk model of cervical adhesions after cervical loop electrosurgical excision procedure

摘要:

目的 分析子宫颈环形电切术（loop electrosurgical excision procedure，LEEP）术后发生宫颈管粘连的危险因素，建立并验证风险预测模型。方法 本研究前瞻性选取2021年9月至2022年4月遵义市第一人民医院子宫颈鳞状上皮内病变（cervical squamous intraepithelial lesion，SIL）且行LEEP治疗的200例患者作为研究对象。根据LEEP术后有无发生宫颈管粘连分为粘连组（n=45）和未粘连组（n=155），采用单因素和多因素logistic回归分析LEEP术后发生宫颈管粘连的影响因素，基于筛选得到的危险因素建立列线图模型，并对模型的精准度进行验证。结果 年龄（OR=1.152，95%CI=1.063~1.249）、创面出血（OR=6.602，95%CI=2.652~16.436）、创口感染（OR=6.288，95%CI=2.357~16.772）、切除深度（OR=3.383，95%CI=1.367~8.373）是宫颈管粘连发生的独立危险因素。验证结果：一致性指数（concordance index，CI）为0.904；Hosmer-Lemeshow提示该列线图模型预测LEEP术后发生宫颈管粘连的风险预测值与实际观测值比较，差异未显示统计学意义（χ²=1.810，P=0.840）；列线图模型预测LEEP术后发生宫颈管粘连的校正曲线趋近于理想曲线，验证前后的平均绝对误差为0.033；列线图模型预测LEEP术后发生宫颈管粘连的校正曲线和受试者工作特征（receiver operating characteristic curve，ROC）曲线下的面积（area under the curve，AUC）为0.904（95%CI=0.858~0.950），灵敏度为0.933，特异度为0.768。结论 本研究构建的SIL行LEEP术后患者宫颈管粘连发生风险预测模型效果良好，可以有效预测宫颈管粘连的发生，可为早期对患者采取预防性干预措施提供参考。

Abstract:

Objective To analyze risk factors for cervical adhesions after loop electrosurgical excision procedure（LEEP） of the cervix，and to establish and verify a risk prediction model for it.Methods We prospectively selected 200 patients with cervical squamous intraepithelial lesions（SILs） who received LEEP treatment in The First People’s Hospital of Zunyi from September 2021 to April 2022. According to the presence or absence of cervical adhesions after LEEP，they were divided into adhesion group（n=45） and non-adhesion group（n=155）. Univariable and multivariable logistic regression analyses were performed to determine risk factors for the occurrence of postoperative cervical adhesions. A nomogram model was established based on the selected risk factors，followed by accuracy verification.Results The independent risk factors for postoperative cervical adhesions included age［odds ratio（OR）=1.152，95%CI=1.063-1.249］，cervical wound bleeding（OR=6.602，95%CI=2.652-16.436），cervical wound infection（OR=6.288，95%CI=2.357-16.772），and the depth of excision（OR=3.383，95%CI=1.367-8.373）. The verification results showed that the CI was 0.904. The Hosmer-Lemeshow test showed no significant difference between the nomogram model-predicted value and the observed value of the risk of cervical adhesions after LEEP（χ²=1.810，P=0.840）. For predicting the occurrence of cervical adhesions after LEEP，the calibration curve of the nomogram model was close to the ideal curve，with the mean absolute error before and after verification being 0.033. The area under the receiver operating characteristic curve of the nomogram model predicting the occurrence of postoperative cervical adhesions was 0.904（95%CI=0.858-0.950），with a sensitivity of 0.933 and a specificity of 0.768.Conclusion The established risk prediction model performs well in predicting the occurrence of cervical adhesions in patients with SILs undergoing LEEP，which can provide a reference for early preventive intervention measures for these patients.