PMDA, the Japanese regulatory agency, has worked for reinforcing and enhancing its post-marketing safety measures as stated in the second mid-term (FY 2009-2013) plan. MIHARI Project-Medical Information for Risk Assessment Initiative was started in FY 2009 to develop a new safety assessment system for post-marketing drugs using Japanese electronic healthcare data in PMDA. In the second mid-term, we examined characteristics of some electronic healthcare data available in Japan including, claims data, hospital information system data and Diagnosis Procedure Combination (DPC) data in order to make efficient use of those databases for the purpose of drug safety assessment. In addition, some pilot pharmacoepidemiological studies for risk assessment and drug utilization were conducted by use of these electronic healthcare data. Based on the accumulated findings, knowledge, and experiences from the pilot studies, we established the framework by FY 2013 to implement pharmacoepidemiology-based safety assessment in PMDA. In the third mid-term (FY 2014-2018), one of the important mission is to apply this framework into the current risk management process of drug safety. For that purpose, cooperation with other divisions of PMDA such as Office of Safety and Office of New Drug is a key. Simultaneously, we will work to establish an access to another database and novel pharmacoepidemiological methods using electronic healthcare data. A large-scale electronic health record database (MID-NET) as well as national claims database are expected to be an important database in the future activities of MIHARI Project. Furthermore, in April 2015, PMDA established new office focusing on Medical Informatics and Epidemiology for further promotion of electronic healthcare data utilization in Japan. In this article, we describe history and past activities of MIHARI Project followed by future challenges.

Objectives：MID-NET® has contributed to post-marketing drug safety assessment through various pharmacoepidemiological studies. To further promote an appropriate use of MID-NET®, this paper describes points to consider in conducting studies utilizing MID-NET® based on results from the two MID-NET® studies.Study design：Cohort study based on the secondary utilization of medical information databaseMethods：In the first study, the incidence of decreased liver function in patients prescribed drugs for pulmonary arterial hypertension was compared with the results from all-case surveillance (primary data collection). In the second study, real world utilization of biosimilar (hereinafter referred to as “BS”) was investigated to understand characteristics of BS prescription in clinical practice. Results：Although the incidence of decreased liver function varied depending on outcome definitions, the proportion on the definition identifying severe cases was similar to that in the all-case surveillance. It suggests that use of multiple outcomes including a definition taking into consideration a degree of severity is important to evaluate study results with robustness appropriately. In the second study, increased trends of BS prescriptions and various patterns of switching (from original biopharmaceutical to BS or from BS to original biopharmaceutical) depending on age and other factors were confirmed. It indicates that considerations about real world utilization of original biopharmaceutical and BS are necessary for future studies targeting BS.Conclusion：These studies further expanded an understanding and knowledge about data characteristics of MID-NET® and provide useful points to consider for promoting appropriate data utilization in future MID-NET® studies.