Hurdle模型及其在居民就诊影响因素中的应用  

Hurdle Model and its Application in Hospitalizing Factors of Residents

作　　者：赵丽华[1,2] 刘桂芬[2] 原静[2] 曾平[2] 

机构地区：[1]太原理工大学理学院数学系,030024 [2]山西医科大学卫生统计教研室,030001

出　　处：《中国卫生统计》2010年第2期149-151,共3页Chinese Journal of Health Statistics

基　　金：山西省自然基金项目(200911005-2)

摘　　要：目的阐明Hurdle模型原理及其在计数资料中的应用。方法利用logit-Poisson Hurdle模型分析两周内居民就诊次数的影响因素。结果 Hurdle模型logistic回归部分结果表明,性别、年龄和经济收入是决定居民是否去医院就诊的主要因素,零截尾Poisson部分结果表明两周内患病次数、由疾病而导致活动天数减少是影响两周内居民就诊次数的主要因素。结论 Hurdle模型是处理零频数过多计数资料一种有效可行的方法 ,其参数估计准确,分析结果切合实际、解释合理。Objective To explore the application of Hurdle model in Health Service Utilization.Methods Analyse the impacting factors about the number of residents＇ treatment within two weeks applying logit-Poisson Hurdle model.Results The part of logistic revealed gender,age and income of the residents become the factors that deciding whether or not to treatment.And the part of zero-truncated Poisson indicated illness（number of illness in past 2 weeks） and actdays（number of days of reduced activity in past 2 weeks due to illness or injury） were related with the number of visit doctor in past 2 weeks.Conclusion Hurdle model was a feasible and effective method to deal with the count data with excess zeros,its parameter estimation was accurate,the analysis result was in line with reality and easy to explain.

关 键 词：Hurdle模型 计数资料 零频数过多 

分 类 号：R195[医药卫生—卫生统计学]