Simple Approaches to Analyzing Self-Controlled Case Series (SCCS) Data

ABSTRACT

Adverse events that occur at moderate rates can be identified during the development of vaccines. When large populations are vaccinated, new rare adverse events can be observed. Determining whether an observed adverse event is related to a vaccine poses statistical issues. Cohort studies can be biased when comparing vaccinated and unvaccinated cohorts. In self-controlled case series designs, each case acts as his or her own control, eliminating the potential bias between vaccinated and unvaccinated participants in a cohort study. The comparison is whether there is an excess of events close to the time of vaccination versus later in time from the vaccination. In this design, one must specify a time point where the excess risk from the vaccine is thought to be over. We explore the effect of misspecifying this time point and develop analysis methods that do not require the specification of the time point. Power and Type I error rates of different analysis methods are simulated.

KEYWORDS:

• Case control
• Guillain–Barré Syndrome
• Scan statistics
• Vaccine studies