非计划重返ICU风险预测模型的构建与验证

Development and verification of the risk predict model of unplanned readmission to comprehensive ICU

目的:分析综合ICU患者非计划重返ICU的现状和危险因素,构建并验证非计划重返ICU风险预测模型。方法:收集2015年1月至2019年4月镇江市某三级甲等医院综合ICU出科患者的相关资料,将患者资料分为建模分组数据和验证分组数据,应用二元Logistics回归分析建模分组患者数据来构建风险预测模型并通过ROC曲线进行拟合度检验,验证分组数据以检验预测模型的灵敏度和特异度。结果:共纳入805名患者资料,42例患者于初次转出ICU后非计划重返ICU,重返率为5.2%,最常见的重返原因是呼吸衰竭。二元Logistics回归结果显示,年龄(OR=1.029)、为呼吸系统疾病的入科诊断(OR=5.625)、二次插管(OR=12.290)和序贯器官功能衰竭(SOFA)评分(OR=1.368)是非计划重返ICU的独立危险因素(P<0.05),ROC曲线下面积(AUC)=0.884,约登指数最大值为0.658,对应的灵敏度为91.9%,特异度为73.9%,截断值为0.033。独立的数据验证结果显示,AUC=0.806,灵敏度为80.0%,特异度为62.3%。结论:本研究构建的风险预测模型对综合ICU患者非计划重返的发生风险具有较好的预测效果,可帮助医务人员在患者转出ICU前进行重返风险评估,为预防性治疗和护理干预提供参考。

Objective: To analyze the current situation and risk factors of unplanned readmission of comprehensive ICU, and to develop and verify the risk predict model of unplanned readmission to ICU. Methods: The data of patients discharged from comprehensive ICU of a tertiary grade A hospital in Zhenjiang from January 2015 to April 2019 was collected. The data was divided into modeling set and validation set. Binary Logistics regression was used to analysis the data, and the model was examined by ROC curve. The grouped data was used to verify the sensitivity and specificity of the model. Results: A total of 805 patients were included, 42 patients readmitted to the ICU after the initial transfer within 30 days. The readmission rate was 5.2%. The main cause of readmission was respiratory failure. The results of binary Logistics regression showed that age(OR=1.029), admission diagnosis of respiratory diseases(OR=5.625), secondary intubation(OR=12.290), and SOFA score(OR=1.368) are independent risk factors for unplanned readmission to ICU(P<0.05), AUC=0.884. The maximum value of the Youden index was 0.658, and the corresponding sensitivity was 91.9%, the specificity was 73.9%. The cutoff value was 0.033. The independent data verification result showed that AUC=0.806, the sensitivity was 80.0%, and the specificity was 62.3%. Conclusion: The risk prediction model had a good predictive effect on the risk of unplanned readmission to comprehensive ICU, which can evaluate the risk before transferring from the ICU, and provide reference for preventive treatment and nursing intervention.