【摘要】 目的 支付标准的测算是DRG实施中的关键环节，本研究旨在分析比较四种常用的数据裁剪方法，为实际付费做好准备，也为后续的研究与实践提供启示及方法改进的依据。方法 利用中段区间法和缩尾法对某市二级医院2017-2019年出院病例的住院费用离群值进行裁剪，用变异系数、总体方差减少系数、Kruskal-Wallis H统计量及ROC曲线评价不同方法裁剪后对DRG分组效能的影响。结果 四种方法裁剪离群值后对DRG分组效能指标均有改善，但改善程度各有不同，最适宜的裁剪方法为方法2（裁剪上限Q₃+1.5IQR,裁剪下限Q₁-0.5IQR）。综合评价离群值裁剪前后CV、RIV以及ROC曲线的变化，可以得到实现更佳分组效能的数据裁剪方法。结论 恰当裁剪离群值可规避医保基金的不必要浪费、维持医院经济平稳运行、减少患者疾病负担，地方医保局可根据实际住院数据恰当选取裁剪方法。更多还原

【Abstract】 Objective The calculation of payment standards is the key link of the implementation of DRG.This study aimed to analyze and compare the characteristics of the four commonly used data trimming methods, prepare for actual payment and provide enlightenment and method improvement basis for subsequent researches in related fields.Methods The outliers of hospitalization expenses of discharged cases in a city′s secondary hospitals from 2017 to 2019 were trimmed using the interquartile range and winsorize.Using the coefficient of variation, reduction in variance（RIV）,Kruskal-Wallis H and ROC curve to evaluate the improvement of DRG grouping performance by different methods.Results The four methods all improved the DRG grouping performance index after trimming the outliers, but the degree of improvement was different.The author believed that the most appropriate trimming method was method 2（high trim point: Q₃+1.5 IQR,low trim point: Q₁-0.5 IQR）.By comprehensively evaluating the changes of CV,RIV and ROC curves before and after outlier trimming, a data trimming method with better grouping performance could be obtained.The results of this study showed that the second method of trimming data could get better effectiveness.Conclusion Appropriate trimming could avoid unnecessary waste of medical insurance funds, maintain the stable operation of the hospital′s economy and reduce the burden of patients′ disease.The local medical insurance bureau should appropriately select trimming method according to the actual hospitalization data.更多还原