摘要
目的建立县级公立医院的病例组合模型,为县级公立医院利用病例组合进行医院服务产出指标调整提供工具。方法应用决策树分类与回归树(classification and regression trees,CART)算法,以住院费用为轴心变量和住院费用影响因素为节点变量,建立病例组合模型,并以方差减少量(reduction in variance,RIV)、变异系数(coefficientof variation,CV)和受试者工作特征曲线(receiver operating characteristic curve,ROC)进行模型检验。结果建立433个组合。RIV值10.00%,CV平均值0.88,ROC曲线下面积基本都大于0.70,说明病例组合模型较为合理,组间差异稍弱于以往。结论可通过纳入更多住院费用影响因素和规范临床诊疗行为进一步改善病例组合。经病例组合指数调整后的医疗服务产出指标更具有可比性。
Objective To establish a case mix model in county-level public hospitals for adjusting hospital service output indicators in county-level public hospitals.Methods Case mix model was established with costs as axis variable and influencing factors of costs as node variable using classification and regression tree algorithm.The reduction in variance value(RIV),coefficients of variance(CV)and receiver operating characteristic curve(ROC)were used to test the model.Results Four hundred and thirty-three case mix groups were formed.The RIV value was 10.00%,mean of CV was 0.88,and the area under the ROC curve was more than 0.70,indicating good performance of the classification model and the variance between groups was slightly smaller than before.Conclusion The case mix model can be further improved by incorporating more variables affecting costs and standardizing clinical diagnosis and treatment behavior.The indexes of medical service output adjusted by case mix index are more comparable.
作者
冯旅帆
胡敏
陈文
FENG Lyu-fan;HU Min;CHEN Wen(Center for Pharmacoeconomic Research and Evaluation,School of Public Health,Fudan University,Shanghai 200032,China)
出处
《中国卫生资源》
北大核心
2019年第3期200-203,共4页
Chinese Health Resources
关键词
病例组合
决策树
医疗服务
评价
县级公立医院
case mix
decision tree
medical service
evaluation
county-level public hospital
