We have systematically identified medicinal products withdrawn worldwide because of adverse drug reactions, assessed the level of evidence used for making the withdrawal decisions, and explored the patterns of withdrawals over time. We searched PubMed, the WHO database of withdrawn products, and selected texts. We included products that were withdrawn after launch from 1950 onwards, excluding non-human and over-the-counter medicines. We assessed the levels of evidence on which withdrawals were based using the Oxford Center for Evidence Based Medicine Levels of Evidence. Of 353 medicinal products withdrawn from any country, only 40 were withdrawn worldwide. Anecdotal reports were cited as evidence for withdrawal in 30 (75%) and deaths occurred in 27 (68%). Hepatic, cardiac, and nervous system toxicity accounted for over 60% of withdrawals. In 28 cases, the first withdrawal was initiated by the manufacturer. The median interval between the first report of an adverse drug reaction that led to withdrawal and the first withdrawal was 1 year (range 0-43 years). Worldwide withdrawals occurred within 1 year after the first withdrawal in any country. In conclusion, the time it takes for drugs to be withdrawn worldwide after reports of adverse drug reactions has shortened over time. However, there are inconsistencies in current withdrawal procedures when adverse drug reactions are suspected. A uniform method for establishing worldwide withdrawal of approved medicinal products when adverse drug reactions are suspected should be developed, to facilitate global withdrawals. Rapid synthesis of the evidence on harms should be a priority when serious adverse reactions are suspected.
Keywords: Adverse reaction; interval; side effect; systematic review; worldwide recall.