Spontaneous reporting of fatal adverse drug reactions.

179 Many countries run schemes which encourage physicians and other health care professionals to report suspected drug reactions (ADRs). The aims of these spontaneous reporting schemes (SRSs) are to identify previously unsuspected reactions, to elucidate the relevant risk factors, and to evaluate the comparative toxicity of drugs within the same therapeutic class [1,2]. By their nature these schemes have significant shortcomings. Only a fraction of all adverse drug reactions are reported, and only a minority of physicians submit reports [3]. Moreover, the causal link between previously unrecognized reactions and a suspect drug may be difficult to establish, particularly if there is a long latency between the prescription and the onset of the event. Reporting is also biased in favour of reactions that have been recently recognized but against reactions that have been known about for many years. Finally, reports are often difficult to interpret because reliable information on prescription volume, and on the demographic characteristics of the population receiving the drug, may be hard to obtain. Recently Kromann-Anderson et al. [4] have described 590 reports of fatal ADR's received by the Danish Committee on Adverse Drug Reactions between 1968 and 1988. The data are of special interest as they are the first review of fatal reactions since the Swedish experience (up to 1975) was published [5]. They also highlight the particular difficulties involved in evaluating serious and fatal adverse drug reactions. As with other national SRSs [6,7], fatal reactions constituted less than 3% of all reactions reported and the total number was very small, even though the Danish reporting rate of suspected ADRs is amongst the highest in the world. There are several reasons why so few reports of fatal ADRs are received. Fatal reactions directly and unequivocally caused by drug therapy using normal therapeutic doses are mercifully rare. The time interval between prescription and death may be long, for example with drug-induced cancers, and it may be more difficult