Quantitative Methods in Pharmacovigilance
Pharmacovigilance serves to detect previously unrecognised adverse events associated with the use of medicines. The simplest method for detecting signals of such events is crude inspection of lists of spontaneously reported drug-event combinations. Quantitative and automated numerator-based methods such as Bayesian data mining can supplement or supplant these methods. The theoretical basis and limitations of these methods should be understood by drug safety professionals, and automated methods should not be automatically accepted. Published evaluations of these techniques are mainly limited to large regulatory databases, and performance characteristics may differ in smaller safety databases of drug developers. Head-to-head comparisons of the major techniques have not been published. Regardless of previous statistical training, pharmacovigilance practitioners should understand how these methods work. The mathematical basis of these techniques should not obscure the numerous confounders and biases inherent in the data. This article seeks to make automated signal detection methods transparent to drug safety professionals of various backgrounds. This is accomplished by first providing a brief overview of the evolution of signal detection followed by a series of sections devoted to the methods with the greatest utilisation and evidentiary support: proportional reporting rations, the Bayesian Confidence Propagation Neural Network and empirical Bayes screening. Sophisticated yet intuitive explanations are provided for each method, supported by figures in which the underlying statistical concepts are explored. Finally the strengths, limitations, pitfalls and outstanding unresolved issues are discussed. Pharmacovigilance specialists should not be intimidated by the mathematics. Understanding the theoretical basis of these methods should enhance the effective assessment and possible implementation of these techniques by drug safety professionals.
KeywordsSignal Score Information Component Spontaneous Reporting System Proportional Reporting Ratio Medicine Control Agency
Dedicated to the memory of my father Richard S. Hauben, MD.
We are grateful for the thoughtful insights and encouragement of Dr Ana Szarfman of the FDA and Andrew Bate of the WHO Collaborating Centre for International Drug Monitoring.
- 7.Evans SJW, Waller P, Davis S. Proportional reporting ratios: the use of epidemiological methods for signal detection [abstract]. Pharmacoepidemiol Drug Saf 1998; 7Suppl. 2: S102Google Scholar
- 9.Heeley EL, Wilton LV, Shakir SAW. The use of proportional reporting rations for signal generation in prescription event monitoring [abstract]. Pharmacoepidemiol Drug Saf 2002; 10(S1): S48Google Scholar
- 10.Kilgour-Christie J, Czarnecki A, Simmons V. Proportional reporting ratio calculations from a pharmaceutical company database [abstract]. Pharmacoepidemiol Drug Saf 2001; 10: S159Google Scholar
- 11.DuMouchel W. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system. Am Stat 1999; 53(3): 177–90Google Scholar
- 12.O’Neill RT, Szarfman A. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system [discussion]. Am Stat 1999; 53(3): 190–5Google Scholar
- 13.Louis TA, Shen W. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system [discussion]. Am Stat 1999; 53(3): 196–8Google Scholar
- 14.Madigan D. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system [discussion]. Am Stat 1999; 53(3): 198–200Google Scholar
- 15.DuMouchel W. Bayesian data mining in large frequency tables, with an application to the FDA spontaneous reporting system [reply]. Am Stat 1999; 53(3): 201–2Google Scholar
- 21.Schroeder DR. Detecting a rare adverse drug reaction using spontaneous reports [statistics]. Reg Anesth Pain Med 1998; 2: 183–9Google Scholar
- 24.Evans S. Statistical and epidemiological principles in detecting signals of adverse drug reactions. Drug Information Association 38th Annual Meeting; 2002 Jun 16-20: Chicago (IL)Google Scholar
- 25.Code of Federal Regulations (CFR). Parts 310, 314, and 600 [online]. Available from URL: http://www.access.gpo.gov/cgibin/cfrassemble.cgi?title=200221 [Accessed 2003 Jan 28]Google Scholar
- 28.Kwong K, Sammon J, Hornbuckle K, et al. A novel method for detecting potential drug-drug interactions using post-marketing adverse event database [abstract]. Pharmacoepidemiol Drug Saf 2001; 10(S1): S39Google Scholar
- 30.A new strategy for detection of signals in the WHO database 2002. Uppsala Reports 2001 Nov; 17: 8–9Google Scholar
- 32.Thakrar BT, Blesch KS, Sacks ST, et al. On technical methods of signal detection [abstract]. Pharmacoepidemiol Drug Saf 2001; 10(S1): S98Google Scholar
- 35.Bate A, Lindquist M, Edwards IR. Pattern recognition using a recurrent neural network and its application to the WHO database [abstract]. 17th International Society for Pharmacoepidemiology; 2001 Aug 23-26; Toronto, CanadaGoogle Scholar
- 36.DuMouchel W, Pregibon D. Empirical Bayes screening for multi-item associations. Proc. KDD. San Diego (CA): ACM Press, 2001Google Scholar