Скачать или смотреть 2011 04 05 NYU CHIBI Andrew Bate: Advances in Drug Safety Surveillance

Abstract: There is a need for ongoing monitoring of pharmaceutical drugs when on the market to ensure the detection of any safety issues that unavoidably went undetected prior to drug approval because of wider use in real world healthcare. Traditionally adverse drug reactions have been most commonly detected by case by case review of suspected adverse drug reaction reports (ADRs). There are increased efforts in quantitative screening of different real world data sets for knowledge discovery of drug safety issues. Spontaneous reporting of suspected adverse reactions has increased in volume and quantitative methods are increasingly essential to glean novel information on potential causal effects from the accumulating data. The existence and ease of access to both sources of Electronic Medical Records (EMRs) and health insurance claims data set has led to their routine use for formal hypothesis testing studies, and now emerging efforts to use such data for hypothesis generation. We describe the routine application of quantitative methods to spontaneous reports, and summarise the emerging use of quantitative methods for ADR detection in EMR and claims data sets. Examples will be chosen to demonstrate the opportunities and challenges in this emerging field, including a retrospective evaluation of angioedema recording with terbinafine use in a UK EMR data of approximately 2 million patient records that illustrates the potential for such methods to detect adverse effects of drug earlier than they are well-recognised. Challenges for the field include the development of methods with sufficiently strong performance characteristics, agreement on how to define and demonstrate appropriate performance levels, the need for the development of techniques for clustering of related diagnoses that often have rigid classification in hierarchical terminologies, and techniques for including other attributes on EMRs in semi-automated manner in quantitative analyses.