摘要
目的建立老年重症肺炎患者出现Ⅱ级以上急性胃肠功能损伤(AGI)的预测模型,并对模型进行评价和内部验证。方法回顾性分析合肥市第二人民医院2019年6月至2022年5月间收治的268例年龄>65岁的重症肺炎患者(训练集207例,验证集61例)。收集所有患者的年龄、性别、基础疾病、肺炎严重程度评分(PSI评分)、镇静镇痛药物用量、机械通气时间等16项指标。在训练集中通过logistic回归分析后建立预测老年重症肺炎患者出现Ⅱ级以上AGI的模型,绘制受试者工作特征(ROC)曲线和校正曲线评估模型的可靠性,并通过验证集数据对模型进行内部验证。结果训练集207例重症肺炎患者中有50例治疗过程中出现Ⅱ级以上AGI,通过logistic回归分析后建立预测模型为:L=SOFA评分×0.181+PSI评分×0.066+丙泊酚用量×0.607+瑞芬太尼用量×1.187,L>19.288时,即可认为此重症肺炎患者有93.24%的概率会出现Ⅱ级以上AGI。对预测模型进行评价,ROC曲线显示模型的区分度较好,AUC=0.960,H-L检验提示(χ^(2)=7.39,P=0.496>0.05),模型拟合度良好。模型预测的灵敏度为82.00%,特异度为96.82%。预测模型在验证集中表现同样稳定,AUC=94.58%(灵敏度81.25%,特异度93.33%),H-L检验提示(χ^(2)=4.51,P=0.808>0.05),预测准确率为90.16%。结论建立的老年重症肺炎患者出现Ⅱ级以上AGI的预测模型可运用于临床,为临床早期预防提供一定参考。
Objective To establish a prediction model of acute gastrointestinal injury(AGI)above grade II in elderly patients with severe pneumonia,and to evaluate and validate the model internally.Methods A retrospective analysis was performed on 268 patients aged>65 years with severe pneumonia admitted to the Second People's Hospital of Hefei from June 2019 to May 2022(207 cases in the training set and 61 cases in the verification set).Sixteen indicators,including age,sex,underlying disease,pneumonia Severity index(PSI)score,dosage of sedative and analgesic drugs,and mechanical ventilation time of all patients were collected.After logistic regression analysis in the training set,a model was established to predict AGI above gradeⅡin elderly patients with severe pneumonia.Receiver operating characteristic(ROC)curve was drawed and correction curve was used to evaluate the reliability of the model.The model was internally validated by validation set data.Results Among 207 patients with severe pneumonia in the training set,50 patients developed AGI above grade Ⅱ during treatment.The prediction model was established by logistic regression analysis as follows:When L=Sequential Organ Failure Assessment(SOFA)×0.181+PSI score×0.066+propofol dosage×0.607+reifentanil dosage×1.187,L>19.288,it can be considered that patients with severe pneumonia have a 93.24%chance of developing grade Ⅱ or above AGI.The ROC curve showed that the model was well differentiated,AUC=0.960.H-L test indicated(χ^(2)=7.39,P=0.496>0.05)the model fit was good.The sensitivity and specificity of the model were 82.00%and 96.82%respectively.AUC=94.58%(sensitivity 81.25%,specificity 93.33%),H-L test indicated(χ^(2)=4.51,P=0.808>0.05)the prediction accuracy was 90.16%.Conclusions The prediction model for AGI after severe pneumonia in elderly patients can be used clinically to help predict the occurrence of AGI in elderly patients with multiple injuries.
作者
赵晶晶
王菁
胡志航
詹圆
邹立群
姚莉
Zhao Jingjing;Wang Jing;Hu Zhihang;Zhan Yuan;Zou Liqun;Yao Li(Department of Intensive Care Medicine,the Second People's Hospital of Heifei,Hefei 230011,China)
出处
《中国医师杂志》
CAS
2023年第4期560-564,共5页
Journal of Chinese Physician
基金
合肥市科技攻关项目(合科[2020]8号)。
作者简介
通信作者:姚莉,Email:15056966307@163.com。
