老年性白内障患者住院费用的多因素分析及模型选择

Multiple factor analysis and model selection of hospitalization expenses of senile cataract patients

目的 采用2种多因素分析方法对老年性白内障手术患者住院费用的影响因素进行分析,并将2种分析方法所得结果进行比较,为分析住院费用的影响因素等相关问题提供方法学上的借鉴,为卫生行政部门和社会基本医疗保障体系的住院费用支付政策地制定提供参考依据.方法 收集新疆某三级甲等综合医院2008-2010年老年性白内障手术患者病例,筛选出符合条件的手术患者资料,分别采用多重线性回归和通径分析的方法对其住院费用影响因素进行分析,并采用相同的适配度指标将两种方法所得结果进行比较.结果 2种分析方法的结果均显示住院天数和药费比例是影响住院费用的2个主要因素,通径分析结果显示：术前占床日是导致住院天数延长的主要因素;多个适配度指标显示：通径分析模型优于多重线性回归分析的结果.结论 通径分析方法的结果不仅能筛选出住院费用各影响因素,还能清晰地显示各因素间的相关关系,并区分各因素对住院费用的直接或间接影响.其所得结果与常用的多重线性回归所得结果基本一致,还提供了更加丰富的信息.此外,在数据的回代方面,使用者手中地一份数据文件可同时采用2种多因素分析方法进行分析,使用通径分析不会额外增加使用者的工作量.在今后相关研究中,可尝试采用通径分析方法处理与此类似问题.

Objective To provide methodology reference for the analysis of the hospitalization expense and related re- search, for the administrative departments and social basic medical security system of hospitalization expenses payment to make reasonable related policies, the factors that influence the hospitalization expenses of senile cataract patients were analyzed by two multiple factor analysis methods, and then their results will be compared with each other ones. Methods The hospitalization ex- pense structure information of all the eligible patients with the senile cataract from 2008 to 2010 were collected from an AAA com- prehensive hospital in Xinjiang province, the data were analyzed by multiple linear regression method and path analysis method, the results of the two methods were measured with the same adapter indices. Results The results of two methods showed that the length of hospital stay and the proportion of prescription charge are the major factors that influenced hospitalization expenses; the results of path analysis revealed that the pre-operation days in hospital is the main factors that caused the hospitalization days ex- tended; several adapter indices showed that the result of path analysis model are better than that of muhiple linear regression a- nalysis. Conclusion The path analysis can found out the relationships among the factors, as well as their impact on the hospi- talization costs. The findings of path analysis are generally consistent with those of the more commonly used method-multiple line- ar regression, but with more abundant information. In addition, the path analysis doesn＇t increase the workload of the users. Path analysis should be applied to deal with similar problems in future research.