基于半参数分位数回归模型的超长住院患者费用研究  

Study on the Costs of Patients with Excessively Long Hospital Stays Based on a Semiparametric Quantile Regression Model

作　　者：朱惠东 郭慧君[1] 杨彦洁[1] 李丹[1] Zhu Huidong;Guo Huijun;Yang Yanjie;Li Dan(Xiyuan Hospital of China Academy of Chinese Medical Sciences,Beijing 100091,China)

机构地区：[1]中国中医科学院西苑医院,北京市100091

出　　处：《中国病案》2025年第1期52-56,共5页Chinese Medical Record

基　　金：中国中医科学院西苑医院管理人才培养工程项目(第二批)(GLRC(MP)202208)。

摘　　要：目的对北京市某三甲中医医院超长住院患者费用的影响因素进行分析,为降低平均住院日和次均住院费用提供数据支持。方法收集2018年1月1日-2023年12月31日出院的超长住院患者住院病案首页信息,建立半参数分位数回归模型,探究不同影响因素对不同水平住院费用的作用效果。结果1160例超长住院患者的住院费用呈偏态分布,中位数为76872元,平均年龄(70±14)岁,平均住院日38天。诊断数量、材料费占比、住院天数、首次住院和输血在10%、30%、50%、70%和90%共5个分位点上的系数均显著为正,中医治疗费占比在5个分位点上的系数均显著为负,入院途径(门诊)在0.1分位点不显著在其余分位点均显著,入院途径(转入)在0.1分位点显著而在其余分位点不显著。年龄的影响效应随着年龄的增加呈现出先降后增趋势。对于药费占比较低和较高的超长住院患者,药费占比的影响效应更强。结论半参数分位数回归结果显示,诊断数量、材料费占比、中医治疗费占比、住院天数、是否首次住院、入院途径和是否输血与超长住院患者费用之间存在显著参数关系,年龄和药费占比与超长住院患者费用之间存在较为显著的非参数关系。另外,不同水平住院费用对各影响因素的响应方向和程度不同。医院应对不同费用水平的超长住院患者采取有针对性的管理措施,增强医疗资源分配效率的同时降低患者经济负担。Objectives This study aims to analyze the factors affecting the cost of prolonged hospitalization in a tertiary Chinese medicine hospital in Beijing to provide data support for reducing the average length of stay and average hospitalization cost per visit. Methods The information on the front page of the hospitalization medical records of patients with extended hospitalization who were discharged from January 1, 2018 to December 31, 2023was collected, and a semi-parametric quantile regression model was established to explore the effects of different influencing factors on different levels of hospitalization expenses. Results The hospital expenses of 1160 extra-long stay patients exhibited a skewed distribution, with a median of 76,872 yuan. The average age of extra-long stay patients was 70±14 years old and average length of hospitalization days was 38 days. The coefficients of the number of diagnoses, the proportion of material cost, length of hospitalization days, whether it was the first hospitalization(yes), whether it had blood transfusion(yes) were all significantly positive at the 5 quantiles of 10%, 30%, 50%,70% and 90%, while the coefficients of the proportion of traditional Chinese medicine treatment cost were significantly negative at the 5 quantiles. The admission route(outpatient) was not significant at the 0.1 quantile and significant at all other quantiles, while the admission route(transfer) was significant at the 0.1 quantile but not significant at all other quantiles. The effect of age showed a trend of first decreasing and then increasing with the increase of age. The effect of the proportion of drug cost was stronger for extra-long stay patients with low and high drug cost ratio. Conclusions The semi-parametric quantile regression results show that there is a significant parametric relationship between the number of diagnoses, the proportion of material fees, the proportion of traditional Chinese medicine treatment fees, the number of days of hospitalization, whether the patient is hospitaliz

关 键 词：超长住院患者 半参数分位数回归模型 住院费用 

分 类 号：R197.3[医药卫生—卫生事业管理]