疾病費用統計是健康服務研究與行政管理決策經常使用的訊息。為了疾病別費用統計製表，我們有必要將上萬個國際疾病分類四位碼與五位碼合併為有限數目的群組，才可能進行有意義的討論與分析。本文的目的有三：首先想回顧過去疾病費用比較研究常用哪些疾病分類群組架構？傳統疾病分類群組架構有哪些優缺點？本文最後再針對目前最新的疾病分類群組架構做更詳細的介紹，並檢討台灣使用此分類架構之可行性。以「章別」作為群組架構的缺失為：（1）分類群組名稱太籠統且不特定；（2）不同群組所包括的分類項數目差異太大；（3）同一章內疾病異質性還是很大。以「基本表」作為群組架構的缺失為：（1）其他殘餘分類項疾病數目太多；（2）無法克服跨章分類的限制。為解決上述缺失，美國醫療照護研究品質局提出「臨床分類軟體」，可以將分散不同章節相同病因的診斷合併，有效減少症狀、徵候及診斷欠明分類項及其他殘餘分類項的數目，因此可以提供較特定有用之訊息提供排序比較。不過，使用此改良分類群組架構時，還是要小心健保申報資料疾病診斷準確度與多重診斷費用歸類的問題。

Cost-of-illness tabulations are commonly used information in health service researches and administration decision-making. In order to tabulate the disease-specific cost of an illness, a scheme of grouping disease categories is needed in order to cluster related 4-digit or 5-digit categories into a meaningful and manageable number for the purposes of analysis and discussion. The aim of this study was to discuss the problems of currently used grouping scheme and to find out the modified grouping schemes that intended to resolve these problems. The problems in using the ‘chapter’ as a grouping schemes include: (1) the titles of the groups are too general and non-specific; (2) the number of categories within each group varies greatly; (3) diseases classified within the same ICD chapter are too heterogeneous. The problems in using the ‘basic’ tabulation list’ as a grouping scheme include: (1) categories and number of disease in other residual groups are too large, and (2) diseases with similar etiology are scattered into different bodily organ system chapters. The Clinical Classification Software (CCS) grouping scheme was developed by the Agency for Healthcare Research and Quality (AHRQ) to solve the above problems. The CCS could efficiently reduce the number of ‘other...’ residual groups and of Chapter XVI ‘signs, symptoms, and ill-defined conditions’ and generally provides more specific information. Nevertheless, we should not overlook the problems of accuracy of diagnosis and multiple diagnoses in claims data.

疾病費用 ； 疾病分類 ； 診斷 ； 申報資料 ； 健康服務研究

cost of illness, classification of disease, diagnosis, claims data, health service research