A three-part model for the self-controlled case series design to estimate and characterize adverse event risk in an overlapping risk period after multiple vaccines: application to ischemic stroke following Pfizer-BioNTech bivalent COVID-19 vaccine and influenza vaccine.

Abstract

This study proposes a three-part model to assess and characterize the risk of serious adverse events (SAEs) when two vaccines are administered on the same day or in close proximity within a self-controlled case series framework. Simulations showed that the three-part model yielded unbiased relative incidences (RIs) after each vaccination and during the overlapping risk period, while censoring follow-up at dose 2 reduced estimation precision but produced unbiased point estimates. Assuming positive multiplicative and positive additive effects, including the overlapping risk period in the first risk interval overestimated the RI after the first dose by 6.0%-26.0%, while including it in the second overestimated the second RI by 7.3%-34.0%. Overall analysis using the three-part model found no increased ischemic stroke risk 42 days after Pfizer-BioNTech bivalent COVID-19 vaccination or after influenza vaccination or during the overlapping risk period among Kaiser Permanente Southern California members <65 years. Among those with a prior-year history of SARS-CoV-2 infection, the overlapping period showed a significantly increased risk (RI=2.74 [95% confidence intervals, 1.07-7.07]), indicating both positive multiplicative and additive effects. Further research is needed to validate these ischemic stroke findings with chart review confirmation and to apply the model to other vaccination scenarios.