Logistic回归联合ROC曲线模型预测综合病房老年感染新冠病毒患者危重风险及分级护理研究  

Logistic Regression combined ROC curve model to predict the critical risk and graded care of elderly patients with novel coronavirus in comprehensive wards

作　　者：莫敏 虞敏 文仪 冷婷 MO Min;YU Min;WEN Yi;LENG Ting(The Guidong People’s Hospital of Guangxi Zhuang Autonomous Region,Guangxi543000,China)

机构地区：[1]广西壮族自治区桂东人民医院,广西梧州543000

出　　处：《中医药临床杂志》2023年第12期2414-2420,共7页Clinical Journal of Traditional Chinese Medicine

摘　　要：目的:回归分析预测法,构建三甲医院综合病房感染新冠病毒老年患者危重风险预测模型,确定分级护理策略,制定护理流程清单。方法:选取广西某三级甲等医院综合病房2022年12月28日—2023年1月30日109例感染新冠病毒老年患者作为研究对象,将其分为非重症组40例和重症组69例,通过单因素分析和多因素logistic回归分析对老年感染新冠病毒患者危重的危险因素进行筛选,结合多因素logistic回归分析结果构建风险预测模型,确定分级护理策略,制定护理流程清单。结果:筛选变量后,建立回归方程模型,logit(P)=(-19.552)+0.931×性别+1.060×是否合并基础病史+0.162×空腹血糖+0.017×C反应蛋白+2.759×降钙素原+0.404×呼吸+1.102×跌倒坠床风险,以预测概率P的约登指数最大时作为临界值,计算新冠病毒感染老年患者危重症的预测概率为0.65,灵敏度为0.79,特异度为0.82。多因素logistic回归分析显示,合并基础疾病的新冠病毒感染老年患者病情危重风险越大,空腹血糖值、C反应蛋白值、降钙素原水平越高,危重风险越高,且呼吸频率越快、跌倒坠床评分越高,风险越大。结论:患者性别、是否合并基础疾病、空腹血糖、C反应蛋白、降钙素原、呼吸、跌倒坠床风险构建的logistic回归和ROC曲线模型,对新冠病毒感染的老年患者危重症疾病的发生能起到较好的预测作用。Objective:regression analysis and prediction method to construct the critical risk prediction model of elderly patients infected with novel coronavirus in the comprehensive ward of grade A hospital,so as to determine the hierarchical nursing strategy and make a list of nursing process.Methods:select Guangxi tertiary hospital comprehensive ward on December 28,2022-30 January 2023,109 cases will be coronavirus elderly patients as the research object,will be divided into the severe group 40 cases and severe 69 cases,through univariate analysis and multivariate logistic regression analysis of elderly infection will be coronavirus patients critical risk factors screening,combined with multivariate logistic regression analysis results build risk prediction model,determine the hierarchical nursing strategy,develop nursing process list.Results:After screening the variables,a regression equation model was established,and whether logit(P)=(-19.552)+0.931 gender+1.060 was combined with basic medical history+0.162 fasting blood glucose+0.017 C reactive protein+2.759 procalcitonin+0.404 respiratory+1.102 falling risk.When the prediction probability P was maximum,the prediction probability of critical illness in novel coronavirus infection was 0.65,the sensitivity was 0.79,and the specificity was 0.82.Multivariate logistic regression analysis showed that the greater the risk of critical illness,the higher the fasting blood glucose value,C reactive protein value,procalcitonin level,the higher the critical risk,and the faster the respiratory rate,the higher the fall score,the greater the risk.Conclusion:logistic regression and ROC curve model of patient sex,combined underlying disease,fasting blood glucose,C-reactive protein,procalcitonin,respiratory,and falling risk can play a good role in predicting the occurrence of critical disease in elderly patients with novel coronavirus infection.

关 键 词：新冠病毒感染 老年患者 重症风险 预测模型 分级护理策略 

分 类 号：R259[医药卫生—中西医结合]