基于CHAID算法的急性胰腺炎患者DRGs分组分析

Analysis on DRGs Grouping of Patients with Acute Pancreatitis Based on CHAID

目的 分析四川省某三级医院急性胰腺炎患者病案首页信息,模拟疾病诊断相关分组(diagnosis related groups, DRGs)进行病例分组,为制定相关住院费用标准及DRGs医保付费改革工作提供参考依据。方法 以单因素秩和检验和多元逐步线性回归对四川省某三级甲等医院2019年1月1日-2021年12月31日出院主要诊断为急性胰腺炎(K85)的1 844例患者病案首页数据进行分析,筛选分类节点变量,纳入决策树模型对急性胰腺炎患者进行DRGs病例组合分析。结果 急性胰腺炎患者住院费用均值为13 736.2元,中位数为11 105.2元。通过住院天数、是否手术治疗、疾病分型和伴并发症/合并症情况4个分类节点将1 844例急性胰腺炎患者共分成7个DRGs病例组合,并制定相应住院费用标准。其中,费用最高组住院费用均值为30 104.4元,中位数为29 434.3元,费用上限为59 180.8元;费用最低组住院费用均值为8 398.5元,中位数为7 890.6元,费用上限为15 183.5元。结论 通过决策树模型建立的急性胰腺炎患者DRGs病例分组合理,研究中制定的住院费用标准和病种权重可作为该地区医保支付改革工作中DRGs分组方案制定和调整的数据参考。

Objective To analyze the home page information of acute pancreatitis patients in a tertiary hospital of Sichuan province,explore the grouping by DRGs and establish related hospitalization cost standards,so as to provide reference for the reform of DRGs medical insurance payment.Methods Single-factor rank-sum test and multiple stepwise linear regression were used to analyze the home page data of 1844 patients with acute pancreatitis(K85)who were discharged from a tertiary hospital in Sichuan Province from January 1,2019 to December 312021.Grouping by DRGs was performed in patients with acute pancreatitis by decision tree model.Results The mean hospitalization cost of patients with acute pancreatitis was 13736.2 yuan,and the median was 11105.2 yuan.A total of 1844 patients with acute pancreatitis were divided into 7 DRGs case groups by four classification nodes:length of stay,surgical treatment or not,severity of illness and complications,and the corresponding hospitalization cost standards were formulated.Among them,the average hospitalization cost of the group with the highest cost was 30104.4 yuan,the median was 29434.3 yuan,and the upper limit was 59180.8 yuan.The average hospitalization cost of the lowest group was 8398.5 yuan,the median was 7890.6 yuan,and the upper limit was 15183.5 yuan.Conclusion DRGs cases of acute pancreatitis established by decision tree model can be grouped reasonably,and the hospitalization cost standard and disease weight developed in the study can be used as data reference for medical insurance payment reform in this area.