As early as 1965, L.J. Witts wrote that ''the final test of the safety of a drug is in fact its release for general use''. The recognition that not all hazards could be known before a drug was marketed and that spontaneous adverse drug reaction reporting systems may fail to identify all hazards, led to several proposals for schemes based on the identification of patients by means of prescription data. These schemes were largely intended to provide information on populations of known size so that the incidence of adverse reactions could be estimated with reasonable accuracy. The proposals included ''Recorded Release'', Registered Release'', ''Retrospective Assessment of Drug Safety'' and a number of variants (Inman, 1978a).
One of the limitations of spontaneous reporting is that doctors may fail to identify and report illnesses which they do not suspect to be due to a drug. This realisation led to the development of systems based upon ''event'' reporting in which the doctor did not need to diagnose or suspect the true cause but was asked merely to record events. To this thinking the distinguished statistician, D.J. Finney, made a fundamental contribution in a paper in 1965 in which an event was defined as ''a particular untoward happening experienced by a patient, undesirable either generally or in the context of his disease'' (Finney, 1965).
These ideas—published only four years after the original announcements of Lenz regarding thalidomide and congenital abnormalities (Lenz, 1961, 1962)—came together in the founding by W.H.W. Inman of prescription-event monitoring (PEM). The establishment of PEM at the University of Southampton in 1980 and Inman's early experience with this technique have been recorded in publications (Inman, 1981, 1981b; Inman et al., 1981) which established that the key objective was to recruit the first 10 000 patients who received a new drug of interest so that any adverse event that occurred in more than one in 1000 patients would be reliably identified.
Was this article helpful?