Evaluation of Reporting Bias in Postmarketing Risk Assessment Based on Spontaneous Reporting Systems
- PDF / 143,434 Bytes
- 6 Pages / 612.283 x 793.701 pts Page_size
- 68 Downloads / 169 Views
CURRENT OPINION
ยช 2009 Adis Data Information BV. All rights reserved.
Evaluation of Reporting Bias in Postmarketing Risk Assessment Based on Spontaneous Reporting Systems Demissie Alemayehu Statistics Department, Pfizer, Inc., New York, New York, USA
Abstract
Databases of spontaneously reported cases of adverse drug reactions play a significant role in the safety monitoring of marketed medicinal products. However, the inherent limitations of such data render the interpretation of results fairly difficult. One major issue is the bias introduced by under-reporting or differential reporting of events. Despite considerable progress in the development of analytical tools for spontaneously reported data, the issue of reporting bias has not been effectively addressed in the literature. This article discusses the problem of handling reporting bias, considers the plausibility of adapting techniques from other areas of application and suggests directions for future research. While the problems associated with spontaneous reporting data are complex and multifaceted, it is concluded that judicious and careful use of rigorous statistical techniques to assess bias can help enhance the reliability of research on the relative toxicity of marketed products.
Clinical trials conducted in the course of product development are inherently limited in their capacity to provide conclusive data about all safety concerns. Studies designed to assess specific characteristics of a drug may require the exclusion of subjects with co-morbid conditions or those receiving concomitant products. Furthermore, such studies do not have adequate power to detect signals of rare events even when aggregated in pooled analyses. Accordingly, spontaneously reported cases of adverse drug reactions (ADRs) play an important complementary role in evaluating and understanding the risk profile of a drug and for developing risk mitigation plans to protect public safety once the product is on the market. The primary source for pharmacovigilance evaluations are databases that rely on voluntary reporting of adverse events. Examples of such databases include the Adverse Event Reporting System database maintained by the US Food and Drug Administration, the EudraVigilance Data Analysis System developed by the European Medicines Agency, the World Health Organization (WHO) Adverse Drug Reaction database and databases maintained by healthcare providers and pharmaceutical companies. With the growing awareness of the significance of spontaneous reporting data, numerous methodological papers and guidance documents[1,2] have been published on the topic.
Although there is no universally accepted approach to handle all possible situations, there are many procedures in routine application. Poisson techniques are often used to quantify the chance of a given number of reports under the assumption of no relationship between the ADR and the suspected medication.[3] Because of the nature of the data, which lacks a natural denominator for computing rates and proportions, most commonl
Data Loading...
