Self-Controlled Case Series Analysis

The case series method was developed by Farrington 1 to investigate the strength of association between a time-varying exposure and a potentially recurrent adverse event, using cases only. The method has been widely used in pharmacoepidemiology, particularly in the study of vaccine safety. The method is derived from an age-dependent Poisson model by conditioning on the number of events and on exposure histories. The conditioning leads to estimation within individuals, and thus the effects of fixed covariates cancel out. In some circumstances, the case series method has high efficiency, relative to the cohort method. A semiparametric version of the case series method, in which the underlying age effect is left unspecified, is also available.