Delays in the post-marketing withdrawal of drugs to which deaths have been attributed: a systematic investigation and analysis

Background As every healthcare intervention carries some risk of harm, clinical decision making needs to be supported by a systematic assessment of the balance of benefit to harm. A systematic review that considers only the favourable outcomes of an intervention, without also assessing the adverse effects, can mislead by introducing a bias favouring the intervention. Much of the current guidance on systematic reviews is directed towards the evaluation of effectiveness; but this differs in important ways from the methods used in assessing the safety and tolerability of an intervention. A detailed discussion of why, how and when to include adverse effects in a systematic review, is required. Methods This discussion paper, which presupposes a basic knowledge of systematic review methodology, was developed by consensus among experienced reviewers, members of the Adverse Effects Subgroup of The Cochrane Collaboration, and supplemented by a consultation of content experts in reviews methodology, as well as those working in drug safety. Results A logical framework for making decisions in reviews that incorporate adverse effects is provided. We explore situations where a comprehensive investigation of adverse effects is warranted and suggest strategies to identify practicable and clinically useful outcomes. The advantages and disadvantages of including observational and experimental study designs are reviewed. The consequences of including separate studies for intended and unintended effects are explained. Detailed advice is given on designing electronic searches for studies with adverse effects data. Reviewers of adverse effects are given general guidance on the assessment of study bias, data collection, analysis, presentation and the interpretation of harms in a systematic review. Conclusion Readers need to be able to recognize how strategic choices made in the review process determine what harms are found, and how the findings may affect clinical decisions. Researchers undertaking a systematic review that incorporates adverse effect data should understand the rationale for the suggested methods and be able to implement them in their review.

Numerous methods for causality assessment of adverse drug reactions (ADRs) have been published. The aim of this review is to provide an overview of these methods and discuss their strengths and weaknesses. We conducted electronic searches in MEDLINE (via PubMed), EMBASE and the Cochrane databases to find all assessment methods. Thirty-four different methods were found, falling into three broad categories: expert judgement/global introspection, algorithms and probabilistic methods (Bayesian approaches). Expert judgements are individual assessments based on previous knowledge and experience in the field using no standardized tool to arrive at conclusions regarding causality. Algorithms are sets of specific questions with associated scores for calculating the likelihood of a cause-effect relationship. Bayesian approaches use specific findings in a case to transform the prior estimate of probability into a posterior estimate of probability of drug causation. The prior probability is calculated from epidemiological information and the posterior probability combines this background information with the evidence in the individual case to come up with an estimate of causation. As a result of problems of reproducibility and validity, no single method is universally accepted. Different causality categories are adopted in each method, and the categories are assessed using different criteria. Because assessment methods are also not entirely devoid of individual judgements, inter-rater reliability can be low. In conclusion, there is still no method universally accepted for causality assessment of ADRs.

A principle concern of pharmacovigilance is the timely detection of adverse drug reactions that are novel by virtue of their clinical nature, severity and/or frequency. The cornerstone of this process is the scientific acumen of the pharmacovigilance domain expert. There is understandably an interest in developing database screening tools to assist human reviewers in identifying associations worthy of further investigation (i.e., signals) embedded within a database consisting largely of background 'noise' containing reports of no substantial public health significance. Data mining algorithms are, therefore, being developed, tested and/or used by health authorities, pharmaceutical companies and academic researchers. After a focused review of postapproval drug safety signal detection, the authors explain how the currently used algorithms work and address key questions related to their validation, comparative performance, deployment in naturalistic pharmacovigilance settings, limitations and potential for misuse. Suggestions for further research and development are offered.