The self-controlled case series method: performance and design in studies of vaccine safety

Musonda, Patrick (2007). The self-controlled case series method: performance and design in studies of vaccine safety. PhD thesis The Open University.

DOI: https://doi.org/10.21954/ou.ro.0000faeb

Abstract

The self-controlled case series method (SCCSM) is a novel study design to investigate associations between acute responses with transient point exposures (for example vaccination). The method provides an attractive alternative to cohort and case-control designs. The method is unusual in that it requires data only on individuals who experience a response (the ‘cases’). The method works as follows. Prior to the study a post-exposure risk period is defined, which corresponds to the period in which responses causally related to exposure are likely to occur. An observation period is also defined, and individuals with responses arising within this observation period are sampled. The data are then analysed using a Poisson model, conditional on the total number of events occurring for each individual. This conditioning ensures that including only cases does not bias the relative risk estimator.

The self-controlled case series method has been used to good effect in many settings, particularly in investigating putative associations between adverse events and paediatric vaccines. However, so far only limited research has been undertaken on the statistical properties of the method in finite samples, and virtually no work has been undertaken on design issues. The method also needs to be extended in various directions, for example application in surveillance methods.

This thesis provides detailed investigations of these topics. To this end, expressions for the asymptotic bias, variance and mean square error of the log-relative incidence are derived. Simulation studies taking account of age are carried out to study small and medium sample performance. Sample size formulae are obtained and validated via simulations, thus improving the design of self-controlled case series studies. The method is extended to applications in surveillance and simulation studies are conducted to evaluate this use of the method. The methods are illustrated using data on intussusception and oral polio vaccine.