目的：(1)分析全民健保資料口服降血糖用藥處方，調查可能導致用藥疏失的診斷與用藥不一致現象；(2)評估專業分工在診斷與用藥及發生錯誤處方的機會之關係；(3)探討藥物間發生此類不一致性可能性是否相同。
方法：本研究為回溯性的資料分析，資料來源為1998 年至2007 年台灣地區的全民健保資料庫，經篩選將無糖尿病診斷碼，但卻開立口服降血糖用藥的申報件數稱為不一致性處方，並進行統計分析。
結果與討論：本研究收集5,221,072 筆門診資料，其中開立口服降血糖用藥的資料共238,768 筆，而有開立口服降血糖用藥但無糖尿病診斷的不一致性處方共2,557 筆，其平均盛行率為1.07%。在不分科、內專科與內次專科分類上，其不一致性處方盛行率分別為2.45%、1.37%與0.52% (P<0.0001)，以內次專科最低；進一步分析內次專科中，心臟血管內科、腸胃內科、腎臟內科與內分泌科，其盛行率分別為1.27%、0.94%、0.71%與0.12% (P<0.0001)，其中以內分泌科最低。治療糖尿病的專門科別就是內分泌科，故可發現專業分工愈細，用藥及診斷符合比例愈高，進而推論可能發生潛在用藥疏失就愈低。在藥物分類上，盛行率分別為biguanides 0.73%、sulfonylureas 1.62%、alpha glucosidase inhibitors 0.45%與thiazolidinediones 0.16% (P<0.0001)，其中以sulfonylureas 盛行率最高，所以由結果發現低血糖風險越高的藥物，與診斷不一致比例也越高，表示可能被醫師誤開的機會也越高，所以藥師調劑口服降血糖用藥時，特別像是sulfonylureas，更應特別審視處方，確認診斷與用藥是否相符。
 

Purpose:The purposes of the study were: 1) to use Taiwan’s National Health Insurance Reseach Database (NHIRD) related to the presciptions of oral hypoglycemic agents (OHAs) for analyzing if medication therapy is consistent with diagnosis; 2) to explore the relationship between diagnosis and medication therapy in various professional levels; and 3) to investigate if inconsistency among medications is same.

Material and Methods:A long-term retrospective study was implemented to use NHIRD between January 1998 and December 2007. The unit of analysis was each OHA prescription without diagnosis codes of diabetes mellitus (DM). The study observation was described as inconsistent prevalence.
Results and Discussion:There were 5,221,072 outpatients’ prescriptions collected in this study, and 2,557 inconsistent prescriptions in 238,768 OHAs prescriptions were observed. The mean inconsistent prevalence is 1.07%.The prevalence of clinic in general, internal medicine and special internal medicine was respectively 2.45%, 1.37% and 0.52% (P<.0001). In special internal medicine, the prevalences of cardiology, gastroenterology, nephrology and endocrinology were 1.27%, 0.94%, 0.71% and 0.12% (P<.0001), respectively. Endocrinology is the professional of DM, therefore it could be found more professional is more consistent between diagnosis and medication therapy. It also implied that potential medication errors were less happened in professional clinics. In OHAs, the prevalence was individually for biguanides 0.73%, sulfonylureas 1.62%, alpha glucosidase inhibitors 0.45% and thiazolidinediones 0.16% (P<.0001). The sulfonylureas with higher risk of hypoglycemia provided higher inconsistent prescriptions among OHAs to assume more potential prescribing errors. For high-alert medications, the consistency between diagnosis and medication therapy should be carefully examined in the healthcare system.