妊娠期糖尿病产妇围术期反流误吸预测模型的构建与验证 [中文引用][英文引用]

目的  探讨妊娠期糖尿病剖宫产产妇围术期反流误吸的危险因素，并建立妊娠期糖尿病产妇围术期反流误吸的预测模型。方法  采取便利抽样法，选取 2019 年 1 月— 2023 年 7 月南京某三级甲等综合医院妊娠期糖尿病剖宫产围术期产妇 960 例作为训练集，另选取 2023 年 8 月— 2024 年 4 月同一医院妊娠期糖尿病剖宫产围术期产妇 320 例作为验证集。运用Logistic 回归分析妊娠期糖尿病剖宫产产妇围术期反流误吸的危险因素，以 R 软件建立反流误吸预测模型，采用 Bootstrap 法重复抽样 1 000 次做模型验证，采用受试者操作特性（receiver operating characteristic，ROC）曲线与校准曲线评估模型的可靠性和预测效果。结果  最终纳入妊娠期糖尿病剖宫产产妇围术期反流误吸模型的预测变量为体质量指数（body mass index，BMI）≥28kg/m2、术前严格禁食禁饮、全身麻醉、急诊剖宫产手术、使用吸入麻醉剂、手术持续时间 >1h 及产前抗生素使用。训练集和验证集的 ROC 曲线下面积分别为 0.855（95%CI：0.781~0.929）、0.842（95%CI：0.771~0.915），训练集、验证集的校正曲线均与实测值曲线基本一致。结论  肥胖、术前严格禁食禁饮、全身麻醉、急诊剖宫产手术、使用吸入麻醉剂、手术持续时间 >1h 及产前抗生素使用是妊娠期糖尿病剖宫产产妇围术期反流误吸的影响因素，本研究构建的妊娠期糖尿病产妇围术期反流误吸预测模型准确率较高。

Objective To investigate the risk factors of perioperative aspiration to the pregnant women with gestational diabetesmellitus（GDM）and caesarean section，hence to develop a nomogram model for prediction of perioperative aspiration in the pregnantwomen with caesarean section. Methods A convenience sampling method was used to select a total of 960 pregnant women who hadGDM and received caesarean section in our hospital from January 2019 to July 2023. The data obtained from the pregnant women wereused as the training set. Further 320 pregnant women who had GDM and received caesarean section in our hospital between August2023 and December 2024 were assigned as the validation set. Logistic regression was used to analyse the risk factors of perioperativeregurgitation and aspiration. A nomogram model of perioperative regurgitation and aspiration for the pregnant women with GDM wasestablished using the R software. The Bootstrap method was adopted to conduct repeated sampling for 1，000 times for model validation. The receiver operating characteristic（ROC）curve and the calibration curve were used to evaluate the reliability and prediction of themodel. Results The included predictive variables in the model were：the strict preoperative fasting and water deprivation，generalanaesthesia，emergency caesarean section，BMI ≥ 28kg/m2，administration of inhaled anaesthetics，operation time over 1 hour，andadministration of prenatal antibiotics. The calibration curves of both the training set and the validation set were found comprehensivelyconsistent with the ideal curve. The areas under the ROC curves of the training set and the validation set were 0.855（95%CI：0.781-0.929）and 0.843（95%CI：0.771-0.915），respectively. Conclusion The predictive variables for perioperative regurgitation and aspirationin the pregnant women with GDM include obesity，strict preoperative fasting and water deprivation，general anaesthesia，emergencycaesarean section，administration of inhaled anaesthetics，over 1 hour of operation time and administration of prenatal antibiotics. Thenomogram model established in this study has a high accuracy and a high clinical value.