T 34: Signal Analysis of Adverse Drug Reactions: Signal Detection/Evaluation Method Formulation Using Important Risk Visualizer™

Poster Presenter

Masahide Nakajima

Manager, Pharmacovigilance Cordination & Administration Department. Ikuyaku.
Mitsubishi Tanabe Pharma Corporation
Japan

Objectives

We have created an original signal detection/evaluation method for adverse drug reaction (ADR), based on multiple statistical parameters, because practical utilities of medical product ADR signals have not been fully examined in post-marketing.

Method

The data from the in-house safety database was installed into the "Important Risk Visualizer" ™, a registered trademark in Japan (IRV), the statistical parameters were calculated and ADRs were categorized by the analysis template based on the evidence-based and public health-based scores¹.

Results

The number of ADRs which may have any causality with each product spanned a few dozens – several hundreds in the safety database, despite the importance of prioritizing ADRs in order to enforce early and effective measures based on accurate the signal detection / evaluation. We calculated the 4 statistical parameters (Reporting Odds Ratio (ROR), Proportional Reporting Ratios (PRR), Gamma-Poisson Shrinker (GPS) and Bayesian Confidence Propagation Neural Network (BCPNN)) and examined significance of them. ROR and then GPS were relatively likely to be significant rather than PRR and BCPNN, regardless of numbers of cases. By the above methods, those ADRs were effectively categorized using the above two scores ranging one to 100, each with three input variables, in terms of the strength of evidence and the potential public health impact. A certain number of ADRs were categorized as “High priority”, since the both scores of them were beyond the thresholds. Most of these ADRs with “the High priority” were confirmed to be notified as the Safety Specifications in the Risk Management Plans and/or as the cautions in the package insert of each product.

Conclusion

These results demonstrate that ADR signals are practically available for pharmacovigilance activities. This original method for ADR signal detection/evaluation, using multiple statistical parameters, through IRV, may be potentially useful for the proactive and accurate ADR risk management. The authors would like to thank Kazutoshi Izawa and Yukio Kitajima (CAC EXICARE Corporation) for technical supports and helpful comments during the analysis of ADR information and preparation of the presentation. 1) Patrick Waller, Emma Heeley and Jane Moseley. Impact analysis of signals detected from spontaneous adverse drug reaction reporting data. Drug Safety, 2005.