首页|期刊导航|医学信息|基于决策树CHAID模型的桡骨远端骨折患者住院费用DRGs分组研究

基于决策树CHAID模型的桡骨远端骨折患者住院费用DRGs分组研究OA

Study on DRGs Grouping of Hospitalization Costs for Patients with Distal Radius Fractures Based on the CHAID Decision Tree Model

中文摘要英文摘要

目的 构建基于CHAID决策树模型的桡骨远端骨折(DRF)疾病诊断相关分组(DRGs)体系,优化医疗资源管理,为医保支付改革与临床路径优化提供循证依据.方法 回顾性分析 2018-2023 年某三甲医院 3136 例DRF患者病案数据,通过单因素(Mann-Whitney U)与多因素(多元线性回归)分析筛选住院费用影响因素,应用CHAID算法构建DRGs分组模型,以非参数检验(Kruskal-Wallis H检验)、方差减少量(RIV)和组内变异系数(CV)评估分组效果,并基于中位数与Tuke's fences法(Q3+1.5IQR)制定费用标准及计算病种权重.结果 基于CHAID模型共形成 8 个DRGs组(P<0.01),首层分类节点为治疗方案,次层为住院天数,末层结合年龄分组及有无并发症/合并症分类.组间费用异质性显著(H=3135.00,P<0.001),方差减少量(RIV)为 72.50%,组内变异系数(CV)介于 0.24~0.55,分组效果良好.以中位数及Q3+1.5IQR制定费用标准并计算病种权重,其中DRGs1 组(保守治疗,≤8 d,无并发症/合并症)费用最低,均数为 2942.24 元,DRGs8(手术治疗,>8 d,有并发症/合并症)均费用最高,达DRGs1 的 7.3 倍,病种权重达到 1.56,医疗资源消耗强度最大.共 227 例(7.24%)患者住院费用超限,超限病例费用占总费用的 13.00%.结论 CHAID模型建立的DRGs体系能精准识别费用影响因素,通过差异化管理降低住院费用,为医保支付改革提供量化依据,同时促进桡骨远端骨折临床路径优化与医疗质量提升,具有一定的卫生经济学价值.

Objective To establish a diagnosis-related groups(DRGs)system for distal radius fractures(DRF)based on the CHAID decision tree model,optimizing medical resource management and providing evidence-based support for medical insurance payment reform and clinical pathway optimization.Methods Retrospective analysis was conducted on 3136 DRF patient records from a tertiary hospital(2018-2023).Univariate analysis(Mann-Whitney U test)and multivariate analysis(multiple linear regression)were performed to identify factors influencing hospitalization costs.A DRG grouping model was then constructed using the CHAID algorithm.The grouping performance was evaluated using the Kruskal-Wallis H test,reduction in variance(RIV),and coefficient of variation(CV).Cost standards and disease weights were determined based on the median and Tukey's fences(Q3+1.5IQR).Results Based on the CHAID model,eight DRG groups were established(P<0.01).The primary classification node was treatment modality,the secondary node was hospital day,and the tertiary node incorporated age grouping and presence of complications/comorbidities.Significant inter-group cost heterogeneity was observed(H=3135.00,P<0.001),with a reduction in variance(RIV)of 72.50%and intra-group coefficient of variation(CV)ranging from 0.24 to 0.55,demonstrating effective grouping.Cost benchmarks were defined using median values and Q3+1.5IQR,and case-mix weights were calculated accordingly.DRG1(conservative treatment,hospital day≤8 days,no complications/comorbidities)exhibited the lowest mean cost(2942.24 yuan),while DRG8(surgical treatment,hospital day>8 days,with complications/comorbidities)had the highest mean cost,which was 7.3 times that of DRG1,with a disease weight of 1.56,indicating the greatest consumption of medical resources.A total of 227 patients(7.24%)had hospitalization costs exceeding the upper limit,and the costs of these outlier cases accounted for 13.00%of the total costs.Conclusion The DRGs system established by the CHAID model accurately identifies cost-influencing factors,reduces hospitalization expenses through differentiated management,provides quantitative evidence for medical insurance payment reforms,and simultaneously promotes the optimization of clinical pathways and enhancement of healthcare quality for distal radius fractures,demonstrating significant health economics value.

何彩云;吴柏辉;陈嘉怡

广州中医药大学附属中山中医院创伤骨科,广东 中山 528400广州中医药大学附属中山中医院创伤骨科,广东 中山 528400广州中医药大学附属中山中医院创伤骨科,广东 中山 528400

医药卫生

疾病诊断相关分组桡骨远端骨折决策树CHAID模型住院费用

Diagnosis-related groupsDistal radius fractureCHAID decision tree modelHospitalization costs

《医学信息》 2026 (14)

39-44,6

10.3969/j.issn.1006-1959.2026.14.006

评论