Is the incidence of upper respiratory tract infection independent of drug treatment in large cohort studies of longer term use drugs?

In observational cohort studies which monitor drug safety, the patterns of reported events are likely to be influenced by a number of factors. We hypothesized that the distribution of events which are unlikely to be adverse events associated with drug use, differ from that for events which may be adverse drug reactions. In 39 individual Prescription-Event Monitoring (PEM) studies, the incidence rates for upper respiratory tract infections (URTI) and back pain, in the first month of treatment and in the subsequent 5 months, were compared to those for nausea/vomiting and malaise/lassitude. For URTI and back pain there was no statistically significant difference in the event rates between these time periods. This pattern may be characteristic of events which are independent of the disease being treated and are unlikely to be adverse events associated with drug use. However, for nausea/vomiting and malaise/lassitude, which can be adverse events associated with drug treatments, the event rates in the first month of treatment were significantly greater than in subsequent months. These observations confirm that doctors are reporting events irrespective of whether or not they suspect the event to be an adverse event associated with the drug. This provides a simple validation of the PEM methodology. Copyright 1998 John Wiley & Sons, Ltd.