Why there is a need for pharmacovigilance

Abstract

Not everything is known about a medicine when it receives its licence for marketing. The merits of a new drug, balancing its beneficial and its untoward effects, become only established after sufficient experience has been gained from its use in real practice. Part of the reason for this is that our extensive phase III clinical trials fail to detect some side‐effects. Why is this so? Three groups of reasons may be envisaged, namely (1) our trials lack the power to detect rare side‐effects; (2) some side‐effects do not occur in the context of clinical trials; (3) some side‐effects, though common enough, fully or partly escape detection due to lack of suitable detection techniques. The following presents a closer look at these three mechanisms. Copyright © 1999 John Wiley & Sons, Ltd.

Flexible Empirical Bayesian Approaches to Pharmacovigilance for Simultaneous Signal Detection and Signal Strength Estimation in Spontaneous Reporting Systems Data

“…Instead, these priors are often represented as general mixtures of parametric distributions for model parameters and take a data-driven approach to estimating the prior from observed data. In the following, we describe a general structure for non-parametric empirical Bayes methods for the Poisson model (1).…”

“…This section introduces a non-parametric empirical Bayes framework to flexibly estimate {𝜆 𝑖𝑗 } within the Poisson model (1). As an extension of the parametric empirical Bayes, non-parametric empirical Bayes methods use priors that are not restricted to a specific parametric form.…”

“…Suppose the following regularity conditions hold:A. The underlying data-generating model(1) for the AE-drug report counts 𝑁 𝑖𝑗 stored in an 𝐼 × 𝐽 contingency table holds for some 1 ≤ 𝜆 𝑖𝑗 < ∞ for all 𝑖 = 1, . .…”

pvEBayes: An R Package for Empirical Bayes Methods in Pharmacovigilance

“…"#E69F00", "#56B4E9", "#009E73", "#F0E442", "#D55E00" )) + scale_shape_manual(values = c (4,15,16,16,16)) + scale_x_continuous(trans = "log10", breaks = c(1.2, 1.4, 1.6, 2, 2.5, 3, 4)) + labs(x = "signal strength", y = "Metric value", color = "model") + theme_bw() + guides(shape = "none") + theme( legend.position = "top" ) The above figure provides the signal estimating performance of the general-gamma model with hyperparameter α selected by various hyperparameter selection methods (color-coded). The row panels show the signal estimation metrics, Max-Scaled-RMSE and Average-Scaled-RMSE, respectively, along the vertical axes against different λ true ij values for the signal cells, across different zero-inflation levels.…”

Frequent Adverse Drug Reactions, and Medication Groups under Suspicion

“…Institute for Health-Economics & Health-Care Research, University of Applied Sciences, Cologne, 2 BfArM, Bonn,3 Institute for Medical Biometry, Informatics, and Epidemiology (IMBIE), University Hospital of Bonn, Germany,4 Clinic for Dermatology and Allergology, University Hospital (RWTH), Aachen, Germany,5 Department of Anesthesiology and Intensive Care Medicine, University of Bonn, Germany…”