Rasch分析健保申報主診斷數量之醫院評比與異常

Rasch Analysis of Major Diagnosis Category of Health Care for Inpatients among Medical Centers in Taiwan

有鑒於美國跳蛙集團將各醫院重要疾病年處理量公佈予民眾作為就醫品 質選擇的參考，中央健康保險局也自2005 年元月起網路公佈各醫院每月申報 的營運狀況，健保高屏分局也公開揭露轄區內醫院服務品質的資訊，遂有必 要研究各醫院的治療領域（MDC' 主要診斷分類） 與特質，以提供更多的就 醫資訊選擇。 利用試題反應理論的Rasch （1960） 分析，將2002 年醫學中心健保住院申 報資料的DRG碼轉歸至所屬的MDC' 並評量出各醫院的（1） MDC 診治量及 （2 ）平均每床MDC 貢獻量之潛能（ latent trait ）水準' 排名分析醫院的年度治 療領域，並揭露各醫院r ；有能力而未達到」或「沒有能力卻是達到」的異常 網路版折線圖分析。 研究發現（1）診治量方面：退輔會醫院最佳，其次為私立醫學院醫院、 公立醫學院醫院及財團法人醫院；六個健保分局內以北區分局最高 （logit=0.3 ） ，其次為高屏、台北、中區、東區及南區。除中區與東區間之外， 分局兩兩相互間主顯著性差異（ p<. 05） 0 （2） 平均每床貢獻量方面： 退輔會 醫院最佳，其次分別為私立醫學院、公立醫學院醫院、及財團法人醫院醫 院，醫院類型間呈顯著性差異（F=52.915； p=.OOO） ；東區分局最高 （ logit=0.33） ，其次分別為北區、中區、高屏、台北及南區。除中區、高屏、 台北等分局間外，分局兩兩相互間主顯著性差異（p<.05） ，六個健保分局間 主顯著性差異（ F=47.824； p=.OOO ） 。

As a view of the fact that US Leapfrog Group Hospital Quality and Safety Survey to provide consumers and purchasers of health care with the information such as Evidence-Based Hospital Referral (EHR) they need to make informed health care choices, we therefore attempted to use the method of Rasch analysis of item response theory to know which field was a hospital in Taiwan mostly expert in to. We found that Veterans-typed General Hospitals surpassed others in those treatments of HIV infection, mental disease and the influence healthy condition. Private-run hospitals performed significantly lower in HIV infection than others. Organ transplants were conducted ultimately frequent in those medical college affiliated hospitals. Rare diseases were mostly treated in government-run hospitals rather than in private-run ones. North Division of National Health Insurance Bureau compared to South Division performed most outstanding in all kinds of major diagnosis category (MDC) for inpatients. This research with Rasch Facets computer program analyzed inpatient-discharged data from Taiwan's 17 medical centers in 2002 and revealed which field was a hospital mostly expert in to. Because the Rasch analysis has those characteristics of transforming examinees' scores into interval-standardized scales, tools with properties of objectivity and sufficiency are available and feasible to social science study in examinees' abilities estimated