Design and analysis of drug safety studies, with special reference to sporadic drug use and acute adverse reactions

Abstract

A methodological approach for analysis of drug safety data which permits the calculation of risk measures for intermittent exposures and acute adverse reactions is presented. This method allows for the observation of exposures that are temporary and which may be of varying lengths, as in the application of analgesics. Risk measures are determined using the relative risk and the excess risk to assess drug safety. The latter has important public health policy significance for decision making about the safety of a drug. The model depends on a simple counting process and allows for the consideration of concomitant variables which can likewise exert an influence on the absolute risk. Hazard estimates are derived by the maximum likelihood method. Three possible approaches to the planning and analysis of studies of drug safety are discussed: cohort studies, case-cohort studies, and case studies. The case study represents an approach which requires only data from the individuals in which an adverse reaction has occurred. The different sampling designs are illustrated using data from a study investigating the etiology of wound healing disorders.