基于分位数回归法的成人白血病患者住院费用影响因素分析

Application of quartile regression method in influencing factor analysis for inpatient cost of adult leukemia

目的应用分位数回归模型分析天津市参保白血病患者住院费用的影响因素,为有效开展医疗费用快速分析,进一步改革和完善医疗保险制度提供方法学应用参考依据。方法收集2003—2013年天津市城市居民基本医疗保险数据库中诊断为白血病的病例5044例,采用分位数回归方法建模进行费用的影响因素分析。结果通过数据建模与分析比较,与多重线性回归模型相比,分位数回归模型在分析能力和估计结果方面更加全面。住院总费用第25、第50、第75分位数回归模型结果显示,年龄和性别对不同分位数的住院总费用影响不同,即并非线性关系(P<0.05);不同性别、年龄段、人员类别、手术与否、住院天数对患者费用影响差异均有统计学意义(P<0.05)。结论分位数回归方法能够更加完整地描述被解释变量条件分布的全貌,也可以分析解释变量是如何影响被解释变量的不同分位数等内容。性别、年龄段、人员类别、手术与否、住院天数等因素可影响不同分位数的住院总费用。

Objective To analyze the influencing factors of hospitalization expenses of leukemia patients insured in Tianjin by quartile regression method,and to provide a reference for effective rapid analysis of medical expenses and further reform and improvement of medical insurance system.Methods Totally 5044 cases diagnosed as with leukemia were collected from Tianjin urban residents basic medical insurance database system from 2003 to 2013,and the quartile regression method was used to analyze the influencing factors of cost.Results Compared with the multiple linear regression,the quartile regression model was more comprehensive in terms of analytical power and estimation results through data modeling and analysis.The results of the 25th,50th and 75th quartile regression model showed that age and gender had different effects on the total hospitalization cost in different quartiles,that is,the relationship was not linear(P<0.05).There was a significant difference in the cost of the patients with different gender,age,type of personnel,operation or not,and hospitalization days(P<0.05).Conclusion The quartile regression method can describe the overall condition distribution of explained variables more completely,and also analyze how explanatory variables affect the different quartiles of the explanatory variables.Factors such as gender,age,type of personnel,operation or not,hospitalization days and so on can affect the total hospitalization cost of different quartiles.