目的：運用試題反應理論建立資訊模組，挑出健保住院費用的異常。方法：發展一具可以處理連續變項的方法，於微軟Excel 下完成資訊模組的建立，挑出異常反應的資料，(1) 計算平均費用異常率，(2) 醫院層級及疾病別間的費用異常率及費用結構不穩定度是否不同？(3) 蒐集南部三家醫院2015 年健保住院費用申報資料，分析最多件數的三類主要診斷的費用異常率，哪類中的哪費用別與其他費用別呈低關聯性？結果：結果顯示，發展的資訊模組挑出異常反應資料，(1) 費用異常個案數的比率為11%，(2) 疾病別間呈現統計顯著性的差異（F2,6=5.24，p=0.04），(3) 2015年12 月醫院A 的呼吸疾病362 件住院申報呈現右偏分佈，內部一致性的信度為0.82，手術費及血液透析費與其他費用呈現不相關的現象。結論：本研究利用可以處理連續變項資料的試題反應模式，結合微軟Excel 的程式，協助醫院同仁易於挑出資料的異常，做為進一步分析資料及檢討改善的依據，以提昇管理的效能。

Objectives: To develop an information module to identify statistically significant abnormal medical fees on healthcare reimbursement claims Methods: Based on item response theory, we developed an information module that dealt with continuous variables to (1) compute the rates of abnormal fees, (2) determine whether any difference existed in terms of hospital level or major disease, and (3) collect inpatient fee data on healthcare reimbursement from three hospitals in southern Taiwan in 2015, analyze their rates of abnormal fees for the top three diagnostic categories, and then identify the non-consistencies. Results: The rate of abnormal inpatient medical fees was 11%. Only the one of disease category showed a statistically signifi cant difference (F2,6=5.24, p=0.04). All 362 cases in Hospital A in December 2015 showed a distribution with an internal consistency reliability of 0.82, and non-consistent fees for the fees regarding operation and hemodialysis. Conclusions: Application of the information module we developed to monitor continuous variables is recommended for use by hospital managers to improve the identification of abnormalities and cost-effectiveness.