Kelly Huang1, Fang-Ju Lin1,2,3, Huang-Tz Ou4, Chien-Ning Hsu5,6, Ling-Ya Huang1, Chi-Chuan Wang1,2,3, Sengwee Toh7. 1. School of Pharmacy, College of Medicine, National Taiwan University, Taipei, Taiwan. 2. Graduate Institute of Clinical Pharmacy, College of Medicine, National Taiwan University, Taipei, Taiwan. 3. Department of Pharmacy, National Taiwan University Hospital, Taipei, Taiwan. 4. Institute of Clinical Pharmacy and Pharmaceutical Sciences, National Cheng Kung University, Tainan, Taiwan. 5. Department of Pharmacy, Kaohsiung Chang Gung Memorial Hospital, Kaohsiung, Taiwan. 6. College of Pharmacy, Kaohsiung Medical University, Kaohsiung, Taiwan. 7. Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, Massachusetts, USA.
Abstract
PURPOSE: Using real-world data to support regulatory decision has become a global movement. However, a robust platform for active surveillance of medical product safety has not been established in Taiwan. METHODS: Following the common data model structure of the U.S. Food and Drug Administration's Sentinel System, we built the Taiwan Sentinel Data Model (TSDM) using the National Health Insurance Research Database with longitudinal claims data from 23 million individuals, linked death and cause of death data from a national registry, and linked electronic health record data from a delivery system. We examined the conversion of the TSDM using the Sentinel Data Quality Review and Characterization Programs in a sample of sex- and age-stratified cohort of 3 million individuals. RESULTS: The TSDM fulfilled the requirements of data quality assurance. Only about 6% of sex and 0.0007% of birth year were missing, and <0.001% of date data had illogical values. CONCLUSIONS: The TSDM-converted database could be a valuable data resource for domestic pharmacovigilance analysis in Taiwan and cross-country evaluation.
PURPOSE: Using real-world data to support regulatory decision has become a global movement. However, a robust platform for active surveillance of medical product safety has not been established in Taiwan. METHODS: Following the common data model structure of the U.S. Food and Drug Administration's Sentinel System, we built the Taiwan Sentinel Data Model (TSDM) using the National Health Insurance Research Database with longitudinal claims data from 23 million individuals, linked death and cause of death data from a national registry, and linked electronic health record data from a delivery system. We examined the conversion of the TSDM using the Sentinel Data Quality Review and Characterization Programs in a sample of sex- and age-stratified cohort of 3 million individuals. RESULTS: The TSDM fulfilled the requirements of data quality assurance. Only about 6% of sex and 0.0007% of birth year were missing, and <0.001% of date data had illogical values. CONCLUSIONS: The TSDM-converted database could be a valuable data resource for domestic pharmacovigilance analysis in Taiwan and cross-country evaluation.