呼吸道合胞病毒感染儿童住院费用影响因素的回归树分析

 目的 利用回归树模型分析和预测呼吸道合胞病毒(respiratory syncytial virus, RSV)感染儿童住院费用的影响因素,为卫生决策提供依据。方法 采用苏州儿童医院2003—2009年完全随机抽样抽取的959例呼吸道合胞病毒感染住院儿童的数据,通过SPSS 18.0,检验用多元线性模型预测住院费用影响因素的适用条件,并构建回归树模型。结果 多元线性回归模型不适用于预测日平均住院费用的影响因素,回归树模型从10个变量中筛选出4个变量:C反应蛋白(C reactive protein,CRP)水平、年龄、是否有哮喘、是否有心肺功能疾病,其中CRP水平是最重要的影响因素,其次为年龄、是否有哮喘和是否有心肺功能疾病, 交叉验证的均方误差(mean squared error,MSE)为0.086。当病例的CRP水平>5.760 mg/L时,仅有CRP水平影响其日平均住院费用;当病例的CRP水平≤3.145 mg/L时,年龄也开始影响日平均住院费用;有哮喘或心肺功能疾病的病例,其日平均住院费用较高。结论 回归树模型在流行病学研究中具有较高的应用价值,在估计呼吸道合胞病毒感染儿童的日平均住院费用时,应重点考虑CRP水平、年龄、是否有哮喘和是否有心肺功能疾病四个因素。

Application of regression tree in the analysis of factors affecting hospitalization expenses for children infected by respiratory syncytial virus

Objective To analyze and predict the factors affecting the hospitalization expenses for children infected by respiratory syncytial virus(RSV),and to provide information for government decision.Methods There were 959 children randomly collected from all the children with RSV infection from 2005 to 2009 in Soochow Universith Affiliated Children's Hospital.SPSS 18.0 was used in the data analysis procedure,including testing the conditions for the model of multiple linear regression and building the model of reg...