利用公開資源於藥物流行病學研究建立藥品檔案

Drug list compilation for pharmacoepidemiology studies using available public resource in Taiwan

目標：提供利用我國健康資料進行藥物流行病學研究時系統性整理出完整藥品檔案的方法，並探討WHOATC碼於藥品檔案建立時之應用和注意事項。方法：利用衛生福利部食品藥物管理署（以下簡稱食藥署）和中央健康保險署（以下簡稱健保署）提供之公開網路資源，藉由藥物學名和WHOATC碼分別搜尋，以建立阻塞性呼吸道疾病藥品（ATC碼R03類）在台灣的藥品品項檔案。此外，本研究除了比較食藥署和健保署對於同一成份藥品之ATC碼編輯情形之差異外，亦比較我國藥品ATC碼和WHO所建立的ATC碼是否相互對應。結果：有關ATC碼為R03藥品，食藥署網站共計958張許可證資料，於健保署網站和藥品壓縮總檔則共計1,071筆健保碼。健保署提供之R03藥品皆已編輯ATC碼，食藥署之公開提供資訊則有143張藥品許可證尚未編輯ATC碼，並有117張藥品許可證與健保署編輯的ATC碼不同。另健保署有18筆健保碼（為6張藥品許可證）之ATC碼未與WHO同步，而發現我國健保署藥品ATC編碼出現五種R03ATC碼不存在於WHO分類中。結論：藥物流行病學研究中，以ATC碼可有效率的篩選藥品成分和劑型，然而食藥署和健保險署之編碼可能會有些許差異，且亦可能與WHO之編碼也有些許不同。因此建議除了以ATC碼取得特定成份和劑型藥品外，仍應以藥品學名再度檢查，以免造成藥品檔案之疏漏。

Objectives: This technical report intends to share a process of compiling a comprehensive drug list for pharmacoepidemiology studies in Taiwan that utilize the National Health Insurance (NHI) data. In addition, it also explains the application of World Health Organization Anatomical Therapeutic Chemical (WHO ATC) codes in research settings. Methods: Publicly available (webbased) resources provided by Taiwan Food and Drug Administration (TFDA) and National Health Insurance Administration (NHIA) were utilized in a study that evaluated drugs for obstructive airway diseases (R03 ATC category). We searched ATC codes and generic names in order to compile a comprehensive drug list for the R03 drugs. ATC codes for drugs from TFDA and NHIA information was compared against the official ATC categorization by World Health Organization. Results: Within the R03 ATC category, 958 TFDA-approved drug entities and 1071 unique NHI drug codes from NHIA were identified. Drugs in the information provided by NHIA were all coded with ATC codes whereas 143 TFDA-approved drug entities did not have ATC code from the TFDA file. In addition, information from the TFDA file for 117 TFDA-approved drug entities had ATC codes that were different from the ATC codes provided by NHIA. Six TFDA-approved drug entities (corresponding to 18 NHI drug items) had ATC code that were different from the WHO ATC coding system. Finally, 5 R03 ATC codes provided by NHIA did not have corresponding code in the WHO ATC system. Conclusions: In pharmacoepidemiology studies utilizing Taiwan NHI data, compilation of drug list of interest could start with ATC code since ATC code represents not only a molecule but also specifies its formulation. However, there may be minor discrepancy between the ATC information provided by TFDA and NHIA, and they may also differ from the WHO ATC coding. When applying ATC code to drug list construction, cross-referencing generic names, especially for combination products, and ATC codes from TFDA and NHIA is highly recommended.