傳統死因統計使用單一原死因進行統計製表，許多死亡證明書上死因相關訊息因此無法被充分利用。目標：以多重死因統計方法分析糖尿病死亡率改變，尤其著重於糖尿病與心血管疾病關係之改變。系統抽樣1987, 1992, 1997 三年死亡證明書各兩千份，死亡證明書上每個診斷都譯碼，再以ACME 電腦選擇原死因並進行多種多重死因分析。ACME 選擇的糖尿病死亡數在不同年齡層或不同年代大多高於衛生署統計室譯碼員的選擇。男性死者糖尿病填寫在第二部分的比例逐年增加，但是在女性並無增加。如果同時考慮糖尿病嚴重度與開具者意圖重新計算糖尿病死亡數，發現死亡率改變型態是先上升後下降，不同於傳統原死因死亡率的直線上升。男性糖尿病伴隨心血管疾病的數目隨年代增加，女性則維持不變。與糖尿病並存的心血管疾病最主要是腎臟病，其次是高血壓，第三是缺血性心臟病。雖然ACME 可以標準化不同國家不同譯碼員間選擇原死因的信效度，但是僵硬的選擇過程卻常常忽略了開具者的意圖，這個問題在糖尿病尤其嚴重。本研究所提出的分類架構可以解決此問題，也比較可以合理估計糖尿病真正的死亡率改變型態
Background: Traditionally, mortality dataare tabulated according to single underlyingcause of death. Many cause of death relatedinformation filled on the death certificateswere not fully used.Objective: To assess the changes in diabetesmortality according by multiple cause ofdeath approach with special emphasize on thechanges in association between diabetes withcardiovascular diseases (CVD).Method: Two thousands death certificateswere systematic sampled of the year 1987,1992, and 1997 for analyses. All diagnoseson the death certificate were coded andACME was used to select the underlyingcause of death (UCD). Several multiplecauses of death analyses were used for thisstudy.Results: The number of diabetes deathsaccording to ACME was higher than thatselected by official coder in each study yearsamong all age groups. Proportion of diabetesin Part II of death certificate in maleincreased by year, nevertheless no suchincrease was observed in female. If we tookinto account both the severity and the intentof the death certifier to reevaluate the role ofdiabetes in the process of dying, the patternof changes in mortality was not the samewith those according to UCD. Number ofcardiovascular disease associated withdiabetes increased by year in male, but not infemale. The most frequent associated CVDwas renal diseases, followed by hypertensionand ischemic heart diseases.Conclusions: Though the use of ACMEcould standardize the selection of UCD among different coders in different countries. Nevertheless, the rigid selection rules sometimes overlooked the intent of the certifiers and this problem was especially prominent in diabetes. The classification frame work proposed by the author could solve this problem and can estimate a more accurate diabetes mortality changes.
