Epidemiologic assessment of risks of adverse reactions associated with intermittent exposure.

Measures of risk of adverse reactions, such as the absolute risk, the relative risk, and the excess risk, are analyzed for individuals who are exposed occasionally for periods of varying duration. This type of exposure is of particular importance for the epidemiologic assessment of drug safety. A Poisson point process model that allows for the simultaneous consideration of concomitant factors is derived for the estimation of the hazard rates. The model can judge acute adverse reactions and is applicable not only for rare risks since an undesired effect may occur repeatedly for an individual. Hazard estimation is achieved with the aid of the maximum likelihood method for several sampling designs, such as cohort studies, case-cohort studies, and case studies. In the latter design only those individuals who have had an adverse reaction during the time of observation are considered for hazard estimation. An application to the analysis of wound-healing impairments is given.