本研究旨在發展長期照護機構院民的病例組合分類系統，研究資料來自五家養護中心 720 位院民， 以美國 MDS (Minimum Data Set) 評估之院民身心狀況與服務需求為分類變項，並以碼錶登錄每位院民一天 24 小時內接受護理人員與護佐照護時間，及一星期內接受各類專業人員的特別處置時間，測量院民的資源耗用量，再利用 PC-Group 電腦軟體中的 AID (Automatic Interactions Detection) 進行院民資源耗用群歸組。結果發現，若以護理人員與護佐照護處置時間測量資源耗用，病例組合系統可將院民歸為 12 個不同資源耗用組， 分組可解釋 28.4% 的資源耗用變異量，資源耗用最大組的平均資源耗用量為最小組的 3.8 倍; 若以全部工作人員照護時間測量資源耗用，則可將院民分為 9 組，解釋 25.4%資源耗用變異量，資源耗用最大組的耗用量為最小組的三倍左右。在資源耗用的歸類中，院民的身體功能、認知障礙、傷口治療、管灌餵食、疾病罹患、與急性症狀的有無等，均為重要分類變項。以上長期照護病人病例組合系統除可提供制定長期照護支付標準的參考，並可幫助機構內部人力調派與品質保証等管理措施。

　　　　　Two case-mix classification systems for institutional long-term carewere developed based on a sample of 720 residents in 5 longterm care facilities.MDS (Minimum Data Set) evaluation of residents' physical/mental status and careneed were gathered, along with stopwatch measure of resource utilization includ-ing nursing staff care time over a 24-hour period and therapy staff time over al-week period. Using AID analysis, residents were classified into 12 groups thatwere homogeneous in their use of nursing resources. The classification explains28.4% of the variation in nursing staff care time. The mean nursing resource useof groups spans a 3.8-fold range. If resource use was measured by total staffcare time, a 9-group classification system was developed. It can explain 25.4%variance in total care time. The mean total resource use of groups spans a3-fold range. Two variance systems have identified that physical functioning,cognitive impairment, skin ulcer or wound care, feeding tube, presence ofchronic disease, and presence of acute episode are important variables incategorizing residents into different resource use groups. The classificationsystem of long-term care patients can provide a reference for the designoflong-term care payment system, staffing level determination, or qualityassurance schemes.