2026-11-092026-11-09https://scholars.csmu.edu.tw/handle/123456789/37763Background: Comparison of utilization, length of stay, cost and patient profiles among different diseases are basic information for health service research (HSR) and administration planning. Nevertheless, if different schemes for diseases grouping were used, different results and conclusions might be derived. Objective: To illustrate the limitations of currently used schemes for diseases grouping and recommend the best scheme that could be used for routine tabulations. Using the BNHI (Bureau of National Health Insurance) claims data of the year 1988 and 2001 to assess the feasibility of the recommended grouping scheme. Method: First, the producers and users of tabulation of BNHI claims data were interviewed to illustrate the limitations and expectations of diseases grouping scheme. Second, we reviewed the related literature to find out the modified and most frequently used schemes. Third, we further compared the different schemes and selected the optimal one. Forth, the real claims data were used to test the feasibility of the selected optimal scheme. Fifth, we compared the figures of Taiwan with those of the United States to reveal the significant information. Results: The limitations of currently used scheme for diseases groupings included: ？HH？H？？the number of categories within each group varied too great across different groups？HH？HH; ？HH？H？？the some of groups were too general and non-specific and the names of some groups were difficult to understand？HH？HH; ？HH？H？？difficulty in counting due to compromising of different classification axes？HH？HH. The expectation of the recommended grouping scheme could solve not only the above-mentioned problems but also could also provide good comparability across countries and time. Through the literature review we found that CCS is superior over GBD, ICD-9, and ICD-10. While using the claim data for CCS we also found that we have to recode many unqualified codes and redistributing the A-codes to various CCS groupings. Conclusions: We recommended that CCS was the best scheme for diagnoses grouping and could be used for routine tabulation of BNHI related data. Owning to the over-coding, multiple-coding, and inaccurate coding of outpatient claim data, efforts of using CCS should focus on inpatient only. An authorized third party should be consulted regarding to the translation of each CCS and the mechanisms of recoding of unqualified coding. Some of the low incidence CCS groups might be combined for parsimony. Several strategies could be used to promote the using of CCS in Taiwan.研究背景：比較排序不同疾病別使用率、住院天數、費用支出與病患特徵是醫療照護服務研究與行政管理經常使用的重要參考數據。如果使用不同的疾病群組架構進行比較，可能會產生不同的結果。 研究目的：比較目前國際學術界所提出較新群組架構的優缺點，並建議健保局採用一個群組架構作為例行統計報表使用，同時以實際資料應用此架構，進一步發現其中的問題並提出解決之道。 研究方法：首先訪談國內專家瞭解現行使用群組架構之缺失與期望；接下來透過文獻回顧來瞭解疾病分類專業領域提出的較新群組架構與醫療照護服務研究常用的群組架構；第三再評比這些架構決定出最佳架構；最後再以實際資料應用，發現其中是否有技術上的困難。 主要發現：國內專家指出現有群組架構的缺失為：「不同群組所包括的分類項數目差異太大」、「群組名稱太籠統且不夠白話」、「妥協不同分類軸所衍生的計數問題」；期望為：「可同時與國際及過去接軌比較」。文獻回顧顯示創新的群組架構包括：「選擇表」、「感染性疾病跨章群組」、「診斷相關群」、「全球疾病負擔」、「臨床分類軟體」；早期醫療照護服務研究大多使用「章」群組架構，近期文獻則較多後三種創新群組架構。本研究評估最佳群組架構為「臨床分類軟體」。使用「臨床分類軟體」於住院申報資料沒有問題，可以方便與國外或過去接軌比較，但是與過去門診申報資料「基本表」接軌有問題，本研究嘗試以民國90年門診分布為參考，分攤無法配對的「基本表」金額，同時也建立一套針對錯誤譯碼重新調整的機制。 結論及建議： 1.建議使用「臨床分類軟體」進行「住院」申報費用、使用率與住院天數統計，一方面可以直接與美國資料比較，一方面也可參考美國醫療照護研究及品質局出版民眾易懂的手冊，再一方面也可以此資料發表於國際學術期刊分享台灣經驗。 2.由於「門診」涉及多重診斷、過度診斷及醫師不正確譯碼等問題，再加上民國90年以前大多是A-code，因此先暫時不必花太多人力與時間於調整A-code與CCS之配對問題。 3.利用ICD-9-CM配對CCS的過程，正好可以檢視各醫院申報資料診斷譯碼的錯誤問題。同時可能也要透過專業的病歷管理協會建立一套具有公信力的調整機制。 4.關於CCS中文診斷名稱翻譯，也最好能讓專科醫學會審議，同時也要顧及民眾的理解與接受程度。 5.由於CCS有260個群組，數目相當多，同時有許多疾病在台灣的發生率相當低，也有必要考慮合併。 6.CCS既然在學理上有相當多優於其他群組架構，實務上就應該大力推廣，各健保分局及醫院都可以廣泛採用此群組架構。可以透過講習、工作坊與行政措施來進行。國際疾病分類衛生統計健康保險international classification of diseasehealth statisticshealth insuranceProject