2-B-P-076 〇鈴木 崇英¹、栢沼 玄¹、古田 晴香¹、山本 浩貴¹、永安 一樹¹、白川 久志¹、金子 周司^1,2,3 Takahide Suzuki¹, Gen Kayanuma¹, Haruka Furuta¹, Hiroki Yamamoto¹, Kazuki Nagayasu¹, Hisashi Shirakawa¹, Shuji Kaneko^{1,2,3 1}京都大・院薬・生体機能解析学、²京都大・院医・ビッグデータ医科学、³九州大・院薬・創薬産学官連携講座 ¹Dept. Mol. Pharmacol., Grad. Sch. Pharmaceut. Sci., Kyoto Univ., ²Dept. Biomed. Data Intelligence, Grad. Sch. Med., Kyoto Univ., ³Grad. Sch. Pharmaceut. Sci., Kyushu Univ. When conducting research using multiple claims databases, it should be noted that there may be bias among databases due to intrinsic preferences in drug prescription and disease diagnosis. For precise data analysis, it is important to understand the reality of drug prescriptions and disease diagnoses in each database; however, comparative analysis of claims databases in different countries have been rarely performed. In this study, we analyzed drug prescriptions and disease diagnoses recorded in JMDC Claims Databases (Japan) and IBM® MarketScan® Databases (USA) to explore the significance of using different databases with multiple examples. In each database, drug ingredient names were mapped to anatomical therapeutic chemical (ATC) codes and disease names were mapped to international classification of diseases, 10th edition (ICD-10) codes, so that 95.4% of prescriptions and all outpatient diagnoses were linked to nonredundant codes. In this analysis, 198 drugs including benzodiazepine hypnotics and 590 disease names such as interstitial lung disorder were significantly more frequent in Japan, while 220 drugs including TNF-alpha inhibitors and 785 disease names such as type 2 diabetes mellitus were significantly more frequent in the USA. These results suggest that a comprehensive comparative analysis of drug prescriptions and disease diagnoses in claims data may reveal previously unreported facts leading to a better understanding of national differences in pharmacotherapy.