Several letters advocating the reinstatement of case reports in the Journal have been published recently (Reference WilliamsWilliams, 2004; Reference EnochEnoch, 2005). I believe that it would be useful to make a distinction between two substantially different kinds of such reports. The first group includes discussions of challenging cases with difficult clinical implications and interesting phenomenological descriptions, with the only aim to improve the readers' diagnostic and therapeutic skills. Typical examples are the 'Grand Rounds' that used to be published in the BMJ. I agree with Dr Enoch and Dr Williams' point of view and I would personally welcome the publication of these case reports in the Journal.

However, another group of reports have a substantially different objective. Their aim is to allow clinicians to share their anecdotal experience of unusual outcomes in clinical practice. These reports are a self-selected group of unlikely cases because only 'man bites dog' stories reach publication. The conclusions of sophisticated randomised trials with good statistical analyses are difficult enough to interpret because of biases such as unmasking, file drawer problems, etc. Anecdotal care reports can be confusing and misleading because the subjective data are often interpreted as objective, creating even more noise where the signal is already faint. The publication of a one-off case report of an adverse effect can profoundly influence clinical practice on the basis of a freak event. Infamous examples include the widely followed recommendation not to use haloperidol and lithium in combination (Reference Cohen and CohenCohen & Cohen, 1974) and the reluctance to use intravenous thiamine for the prevention of Korsakoff syndrome on the basis of a few reports of adverse reactions (Reference Thomson and CookThomson & Cook, 1997). The cases of the hundreds of thousands of people who have been safely and successfully treated with these medications are not published because no one wants to state the obvious. I believe that the past editor's decision to move on from publishing this latter group of case reports was extremely wise.