ICU综合征多因素Logistic回归分析及风险模型的建立

Multivariate Logistic analysis of ICU sydrome and establishment of risk model

目的探讨ICU综合征发生的主要危险因素，建立Logistic风险模型，并评价其预测ICU综合征发生的灵敏度、特异度和准确度。方法收集2007年6-9月入住ICU的危重症患者的数据，包括年龄、性别、文化程度、宗教信仰、家庭收入、医疗费报销情况、是否有家庭成员缺失、原发病、患病时间、性格类型、APACHE Ⅱ评分等，采用SPSS 11．0统计软件包进行单因素分析，取其中有统计学意义的变量，做多因素Logistic回归分析。结果年龄、性别、家庭收入、医疗费报销情况、是否有家庭成员缺失、是否经历同室患者死亡、住ICU时间、是否呼吸机治疗、睡眠情况、APACHE Ⅱ评分、患病时间、性格类型等因素与ICU综合征发生有关。Logistic回归分析共有6个因素进入模型，按作用强弱依次为是否呼吸机治疗、住ICU时间、睡眠情况、性别、患病时间和曾经历同室患者死亡。此模型判断ICU综合征的准确度为82．37％，灵敏度为79．42％，特异度为85．13％。结论曾使用呼吸机、住ICU时间、睡眠情况、性别、患病时间和曾经历同室患者死亡是ICU综合征发生的主要危险因素，所建立的风险模型能较好预测ICU综合征的发生。

Object To investigate the main risk factors of ICU sydrome,and establish Logistic regression model to evaluate the accuracy rating,sensitivity and specificity for predicting ICU sydrome. Methods Investigated all patients hospitalized in ICU from June to July of 2007. Data collected include age, gender,religion,family income, education degree, primary disease, character type, APACHE II score and et al.Case-control study was used to retrospectively investigate the clinical data of ICU sydrome.Firstly,single factor analysis was used in 21 clinical variables. Then , Multivariate Logestic regression analysis was used in selected variables. Results Twelve clinical variables ,including age, gender, family income, health care cost wiping out degree,whether or not having family member＇death , whether or not encountering other patients＇death,length of stay in ICU, whether or not receiving ventilator theraphy, sleep quality,score of APACHE II,time with disease, character type.By Logestic regression analysis,it was revealed that there were 6 significant factors for ICU sydrome in diminishing sequence as whether or not receiving ventilator theraphy, length of stay in ICU, sleep quality, gender, time with disease and whether or not encountering other patients＇death. Logestic regression equation was as following： P=l/[ 1 ＋ Exp ∑ （31.425 ＋2.611X1＋8.188X2-2.677X3＋7.474X4＋4.312X5＋6.153X6） ]. The accuracy rating,sensitivity and specificity of the Logestic model was 82.37 %,79.42 % and 85.13% respectively. Conehusions Whether or not receiving ventilator theraphy, length of stay in ICU, sleep quality, gender, time with disease and whether or not encountering other patients＇death were the main risk factors of ICU sydrome. The Logestic regression model based on these factors is reliable in predicting ICU sydrome.