疾病诊断相关分组(DRGs)在肿瘤门诊特殊疾病病种支付中的应用研究

通过多元线性回归模型筛选对医疗费用有统计学意义的影响因素并筛选费用异常数据,利用基于E-CHAID算法的决策树模型进行DRGs分组,用变异系数、方差减少量及非参数检验验证分组的合理性。通过统计模型剔除11条异常数据后共分为7个DRGs组,经CV、RIV及秩和检验验证后证实分组效果合理且较为稳定。相关部门可以门诊特殊疾病为试点逐步推开门诊DRGs的应用,分组时需从数据分布特点入手,在合理的数据基础上结合疾病特征、治疗方式等因素进行分组并动态调整,将DRGs分组的"事前控制"转变为"事前测算—事中控制—事后调整",提高分组的综合性和实用性。

Study on the application of DRGs in the payment of special disease outpatients in tumor system

The multiple linear regression models were used to screen the statistically significant risk factors of medical expenses and abnormal data. DRGs grouping was performed using the decision tree model based on the E-CHAID algorithm with the variation, variance reduction and non-parametric test coefficients to verify the rationality of groups for preferred results. After eliminating 11 abnormal data by the statistical model, results found that DRGs can be divided into 7 groups. After verifying CV, RIV and rank-sum test, the results showed reasonable and stable DRGs grouping. The relevant departments can gradually carry out the application of DRGs in outpatient payment reform by setting special diseases as an experimental data. When grouping, it is necessary to start with the characteristics of data distribution, combining and dynamically adjusting DRGs according to factors such as disease characteristics and treatment methods based on reasonable data. The DRGs comprehensiveness and practicability can be improved by changing "pre-control" to "pre-estimate, process-control, and post-adjustment".