【摘要】 目的 构建并验证经口气管插管患者发生口腔黏膜压力性损伤的风险预测模型。方法 纳入2022年12月至2023年7月浙江省某三级甲等医院ICU的400例经口气管插管患者为模型训练集，2023年8-10月的152例为模型外部验证集。采用logistic回归筛选口腔黏膜压力性损伤的预测变量。绘制受试者工作特征曲线下面积、校准曲线和临床决策曲线来评估预测模型的预测性能。采用Bootstrap法对模型进行内部验证。结果 logistic回归结果显示，年龄、躁动-镇静程度量表评分、急性生理与慢性健康评分Ⅱ、血清白蛋白水平、气管插管操作者、气管导管固定方式和人工气道留置时间是口腔黏膜压力性损伤的影响因素(P<0.05)。训练集和验证集的受试者工作特征曲线下面积分别为0.913(95%CI:0.873～0.954)和0.956(95%CI:0.922～0.981);Hosmer-Lemeshow检验结果表明，模型预测值与实际值吻合较好，模型校准度较好(χ~2=1.702,P=0.427)。Bootstrap法重抽样结果显示，受试者工作特征曲线下面积为0.908(95%CI:0.852～1.105),灵敏度为94.9%,特异度为60.8%,准确率为83.7%。结论 本研究构建的风险预测模型具有较好的区分度、校准度和临床应用价值，可为精准识别经口气管插管患者口腔黏膜压力性损伤高风险人群提供参考依据。更多还原

【Abstract】 Objective To construct and validate a risk prediction model for oral mucosal membrane pressure injury in patients with oral tracheal intubation. Methods 400 patients with oral tracheal intubation in a tertiary-level hospital in Zhejiang Province from December 2022 to July 2023 were included as the model training set, and 152 patients from August to October 2023 were included as the external validation set of the model. Logistic regression was used to screen predictive va-riables for oral mucosal membrane pressure injury. The area under curve(AUC) of the receiver operating characteristic, cali-bration curve, and clinical decision curve were used to evaluate the predictive performance of the prediction model. The model was validated internally using the Bootstrap method. Results The logistic regression showed that age, Richmond Agitation-Sedation Scale score, Acute Physiology and Chronic Health Evaluation Ⅱ, serum albumin level, tracheal intubation operator, tracheal tube fixation method, and artificial airway retention time were influencing factors for oral mucosal membrane pressure injury(P<0.05). The AUCs of the training set and the validation set were 0.913(95%CI: 0.873 to 0.954) and 0.956(95%CI: 0.922 to 0.981). Hosmer-Lemeshow test showed that the predicted value of the model was in good agreement with the actual value, and the model calibration degree was good(χ~2=1.702, P=0.427).The Bootstrap resampling showed that the AUC for the subject was 0.908(95%CI:0.852 to 1.105), with a sensitivity of 94.9% and a specificity of 60.8%, and an accuracy of 83.7%. Conclusion The risk prediction model constructed in this study has good discrimination, calibration, and clinical application value. It can provide reference for accurately identifying high-risk patients with oral mucosal membrane pressure injury after oral tracheal intubation.更多还原