Statistical perspectives on the use of pharmacovigilance data and electronic health records to verify predicted chemical hazards of drugs from in-vitro toxicological data
The liver is considered as one of the organs that are highly susceptible to drug-induced toxicity, leading to a diverse set of responses such as acute liver injuries, black-box warnings, and possible market withdrawal of medications in spite of having first appeared non-toxic and effective in animal and clinical studies. Over the last few decades, drug safety assessment have been limited to animal studies and human clinical trials. These types of drug safety studies have suffered from low rates of predicting drug-induced toxicities with animal/human concordance less than 60% in major human organs. Screening electronic health records (EHR) and spontaneous reporting systems (SRS) for adverse drug reactions has gained a lot of attention recently. This project will explore a novel approach to use EHR from large population hospital data and SRS as principal human data to verify drug safety data from in-vitro toxicology studies. The main focus will be on assessing how much these laboratory data can truly predict drug induced hepatic toxicity in humans, while examining two case study drugs: Rosiglitazone and Troglitazone.