目的 探索股骨开放骨折疾病诊断相关分组(diagnosis related groups，DRGs)的方案及费用参考标准。方法 数据来源于四川省某三甲综合医院2019年1 月1日—2024年1 月1日82例股骨开放骨折患者病案首页信息，采用多元线性回归筛选出对住院费用有影响的分类节点变量，运用决策树模型建立股骨开放骨折的DRGs分组方案，并计算出各组的费用标准。结果 股骨开放骨折患者住院平均费用77 946。51元，ICU、手术次数、并发症作为分类节点变量纳入决策树模型，共建立8个DRGs分组及相应的住院费用标准;采用Kruskal～Wallis检验各组住院费用差异有统计学意义(x2=61。440，P=0。000)，变异系数为0。25-0。48，组内同质性好。结论 运用决策树模型构建的股骨开放骨折DRGs分组方案合理，可为其他地区或医院在医院运营管理及DRGs医保付费中提供一定的参考价值。
Research on the diagnosis related groups of open femoral fractures based on decision tree model
Objective To investigate the protocol and expense reference standard of Diagnosis Related Groups(DRGs)for patients associated with open femoral fractures.Methods The medical records of 82 patients with open fracture of femur in a tertiary class hospital of Sichuan Province from January 1st,2019 to January 1st,2024 were collected.Multiple linear regression was performed to screen out the categorized node variables that had an impact on hospitalization costs.The DRGs scheme for pa-tients with open fracture of femur was established by using decision tree model,furthermore,the standards cost of each group were calculated.Results The average hospitalization cost of patients with open fracture of femur was 77 946.51 yuan.The In-tensive Care Unit(ICU),number of operations and complications were included in the decision tree model as categorized node variables.A total of 8 DRGs protocol and corresponding hospitalization cost standards were established.There was statistically significant difference in hospitalization expense among each groups by Kruskal-Wallis test(X2=61.440,P=0.000),and the coefficient of variation was ranged from 0.25 to 0.48,indicating the homogeneity of each groups was well.Conclusion The DRGs scheme for patients with open femoral fractures based on decision tree model was reasonable,which can provide certain reference value for other regions or hospitals regarding hospital operations management system and DRGs medical insurance payment.
DRGsOpen fracturesFemoral fracturesDecision tree modelHospitalization expenses
万方数据
维普
