Simulation Studies of Self-Controlled Case Series Methods in Vaccine Safety Research

• Activity Number: 140
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Guoying Sun*+ and Wei Hua and Nick Andrews and Caitlin N. Dodd and Silvana A. Romio and Hector Izurieta and Heather J. Whitaker
• Companies: U.S. Food and Drug Administration and U.S. Food and Drug Administration and Health Protection Agency and Cincinnati Children's Hospital Medical Center and Erasmus University and U.S. Food and Drug Administration and Open University
• Keywords: Self-controlled case series ; Pseudo-likelihood ; post-vaccination ; Contraindication

Abstract

Self-controlled case series (SCCS) method was developed to investigate the association between vaccine and adverse event (AE). When the AE is a contraindication to the vaccine, the SCCS assumption that the event must not alter the exposure process is violated. To investigate this problem, we ran a series of simulations to determine the magnitude of bias and to evaluate different methodologies developed for dealing with this. Three analysis approaches were used to assess the power and accuracy of estimates: 1) the standard SCCS method, 2) post-vaccination follow-up time only with the standard method and 3) the pseudo-likelihood method. The simulations showed that when there was no contraindication to vaccination, the standard method made the best use of all exposure information appropriately and provided higher power; when the contraindication did exist, the pseudo-likelihood method was more appropriate, providing more accurate point estimates. The post-vaccination cases only method worked well when individuals received a single exposure, with a lower power relative to other methods. These results could provide insight into choosing the most appropriate method in real data analysis.