[Pharmacovigilance: a case of phantom ships and Russian roulette].

The medical literature is full of associations between drugs and adverse events which have not been proven by epidemiological studies, such as the 30-year-old discussion on levodopa use and the development of melanoma. The organisation of drug safety (pharmacovigilance) by marketing authorisation holders and regulatory authorities relies heavily on voluntary reporting schemes, despite the fact that such schemes suffer from substantial underreporting. Nevertheless, even though marketing authorisation holders are legally responsible for the product, regulatory authorities scarcely oblige them to perform hypothesis-testing epidemiological studies. In view of the importance of drug safety for healthcare and the need for checks and balances, pharmacovigilance should be coordinated by an organisation which is independent from the pharmaceutical industry and regulatory authorities, such as the Dutch Healthcare Inspectorate.