Rasch分析優良病歷評價之裁判給分與分析

Rasch Analysis of Fine Medical Record Writing Appraised by Judge Scores

新制醫院評鑑標準第五章第5.1 項次規定醫院應貫徹病歷紀錄之完整性， 許多醫院皆定期辦理優良病歷書寫的獎勵與督促，其主要目的即在提昇病歷 書寫的品質水準'也對健保給付時的病歷審查提供有利的辨證o 然而，對於 優良病歷書寫評價所面臨的（1）量尺分數之選擇及C 2 ）評審評分的差異性 分析，卻是甚少見諸研究。鑒於Rasch 分析提供等距分數的比較及更多評審的 訊息，遂以某醫學中心2005 年8 月的6 位醫師審查12 本6 科系病歷資料，利用 Facets 電腦軟體進行Rasch 分析與排名。研究發現Infit MNSQ 及Infit ZStd 可以 偵測出問題評審的行為偏差，可以有證據地提出「發回更審」或「除外計分」 而使評審制度更為公平。原始評分排名與「發回更審」適當處理不致影響排 名， 「除外計分」可致排名產生變化。各科病歷品質水準依序為ICU 、心臟 外科、神經外科、胃腸科、骨科及感染科，外科系醫師較內科系醫師評分嚴 苛COR= 1. 59； 95%CI= 1. 24 ， 2.06） 。在二層面C facets ）的Rasch 模式下的原始 分數會與測量分數排名相同，但在有評審因子干擾及分數調整下之三層面原 始分數未必與排名相同。傳統上在醫院內所做的病歷抽審制度，未將量尺制 度及評審分析列入考量，其病歷品質的排名會受影響。對於每月或2 個月一次 的病歷審核制度，除評比病歷品質排名外，對於評審習性之分析亦應加以討 論，以使病歷評審機制更為健全。

Medical records were requested in compliance with that of being completely and clearly described in chapter 5 of the new version of hospital accreditation standard. Many hospitals regularly conduct the writing reward and the supervision both in promotion quality standard of medical records and in advantage of providing dialectical evidence for the health insurance. Selection regarding fine medical record writing facing those of 1) choice of the scale score and 2) bias of inter-judges was rarely considered and never concerned about. In view of the fact that the Rasch analysis provides interval scores and more statistic messages we used Rasch many-facets computer software to perform an appraisal of stochastically examining 12 medical records by 6 doctor-judges. The results found in this study were that Infit MNSQ and Infit ZStd could detect the problematic judge who inconsistently scored and had the evidence needed to propose recheck or exclusion in this appraisal system. The best qualities of medical record writing were those departments of ICU, cardiac, neurosurgery, gastroenterology, orthopedics and infection. The surgical doctors graded more severe comparing to internal medicine ones (OR=1.59; 95%CI 2.06,1.24). In tradition the medical record sample examining system, not includes the scale alternatives and the analysis of inter-judges, was done as a result of distortion in quality ranking. We suggested that judge behaviors should be analyzed and performed in future appraisals in a bit to avoid rank glitch caused.