Objectives: To assess the risk of cardiovascular diseases (CVDs) associated with metformin, SU, MG, acarbose, pioglitazone, rosiglitazone and insulin in type 2 diabetic (T2D) patients considering effects of multiple drug switching patterns. Methods: This retrospective cohort study was conducted from April 2004 to February 2009 using the Taiwan Pay-for-Performance Diabetes Registry (November 2003 to February 2009). 71,552 adult newly-registered T2D patients were included. Individuals’ antidiabetic treatment was categorized to montherapy (metformin, sulphonylureas (SUs), meglitinides (MGs), acarbose, pioglitazone, rosiglitazone, and insulin), two drug combination (pioglitazone dual combination and rosiglitazone dual combination) and three drug combination (pioglitazone triple combination and rosiglitazone triple combination). The reference group was metformin monotherapy, pioglitazone dual combination and pioglitazone triple combination. The outcome was cardiovascular event (myocardial infarction, ischemic heart disease and congestive heart failure). Cox proportional hazard model was used to evaluate CVD risk associated with different interval groups and adjusted for adding variables in sequence. Results This study included totally 71,552 newly-registered T2D patients. 173,379 intervals and 1,797 events were identified. The mean follow-up duration was 27.34±6.73 months per patient and the mean interval period was 273±320 days. In model 1, when adjusting for age, gender and diabetic history, SU (P=0.0316), insulin (P

目的： 考慮換藥和組合用藥的情況，分析 metformin，磺脲類藥物（SU），苯甲酸 衍生物類促泌劑（MG），acarbose，pioglitzone，rosiglitazone 和胰島素對台灣 II 型糖尿病患者心血管疾病患病風險的影響。 方法： 本研究是回顧性隊列研究，基於“台灣全民健保糖尿病給付改善方案試辦計 畫”，2003 年 11 月 1 日至 2009 年 2 月 28 日的資料庫。納入 71,552 位糖尿病成 年新病例。將研究對象在觀察期內的降糖藥用藥分成不同的區間，包括單藥區 間：metformin，磺脲類藥物，苯甲酸衍生物類促泌劑，acarbose，pioglitazone， rosiglitazone，胰島素，以 metformin 單用為對照組；雙藥區間：含有 piolitazone 的雙藥用藥，含有 rosiglitazone 的雙藥用藥，以含有 pioglitazone 的雙藥用藥為對 照組；三藥區間：含有 pioglitazone 的三藥用藥，含有 rosiglitazone 的三藥用藥， 以含有 pioglitazone 的三藥用藥為對照組。相同的用藥區間歸為一組。效果事件 是第一次發生的心血管事件，包括心肌梗塞、充血性心力衰竭和缺血性心臟病。 採用 Cox 比例風險模型，依次增加纳入变量，計算不同用藥區間之間的風險比 及其 95%置信區間。 结果： 71,552 位研究對象共有 173,379 個用藥區間，平均每位病人追蹤 27.34±6.73 個月，平均每個用藥區間追蹤 273±320 天，共發生 1,797 件心血管事件。當第一 個模型納入處方開立年齡、性別、糖尿病病史等三個變量，磺脲類（P=0.0316）、 胰島素（P