摘要目的:通过构建和验证过度医疗行为线索富集模型，为医保基金安全合理使用提供技术支持。方法:利用等级赋分法构建过度医疗行为线索富集模型。获取2016—2019年北京市5家肿瘤专科医院的职工医保和居民医保患者住院交易记录，验证模型的有效性。将每条交易记录转化为标准化分值V(0~100分)，并根据V值大小将全部交易分为22个组，计算每组的拒付率和富集效率。富集效率随V值增加的变化趋势采用Cochran-Armitage趋势性检验，组间拒付率比较采用 χ2检验，V值与拒付金额的相关性分析采用Pearson检验。 结果:纳入模型验证的职工医保和居民医保患者住院交易记录分别有872 599条和86 356条，各含1 164条和103条拒付交易记录。职工医保和居民医保交易记录V值平均分均为49分，中位数均为50分。当V>0分时，随着职工医保和居民医保交易记录V值的增加，富集效率呈增高趋势( Z＝23.86， P<0.001; Z＝11.02， P<0.001)，不同组间的拒付率差异有统计学意义( χ2＝1 307.16， P<0.001)，V值与职工医保和居民医保拒付金额有相关性( r＝0.29， P<0.001; r＝0.30， P＝0.003)。 结论:基于等级赋分的过度医疗行为线索富集模型通过分析大量医保交易记录，能够聚焦在有疑似过度医疗行为的医保交易上，在医保基金管理方面具有一定的事前指导作用。

abstractsObjective:To construct and verify a model of excessive medical behavior clues enrichment, for the technical support for the safe and rational use of medical insurance fund.Methods:A model of excessive medical behavior clues enrichment was constructed by the rank assignment method. The inpatient transaction records of medical insurance for employees and residents in 5 tumor hospitals in Beijing from 2016 to 2019 were obtained to verify the validity of the model. The patients were grouped according to age and gender, and each transaction record was converted into a standardized score V(0-100 scores), all transactions were divided into 22 groups according to V value. The Cochran-Armitage trend test was used to analyze the variation trend of enrichment rate with the increase of V value. Chi-square test was used to compare the chargeback rates of different groups. The correlation between the standardized score V and the amount of chargeback was tested by Pearson.Results:There were 872 599 and 86 356 hospitalization transactions for employee and resident medical insurance patients, with included 1 164 and 103 chargeback records respectively. The average score and median of V-value of employee and resident medical insurance transactions were scores of 49 and 50 respectively.When V>0, the enrichment rates of the employee and resident medical insurance were on the rise with the increasing of V( Z=23.86, P<0.001; Z=11.02, P<0.001), the refusal rates among different groups was significant different( χ2=1 307.16, P<0.001)and the correlation between V value and the chargeback amount was statistically significant( r=0.29, P<0.001; r=0.30, P=0.003). Conclusions:This study established a clue enrichment model of excessive medical behavior based on the rank assignment method. By analyzing a large number of medical insurance transaction records, the model can focus on the medical insurance transaction with suspected excessive medical treatment behavior, and has a certain guiding role in the management of medical insurance fund.