An efficient statistical algorithm for a temporal scan statistic applied to vaccine safety analyses

Abstract

In the US, the Vaccine Safety Datalink (VSD) project, sponsored by the Centers for Disease Control and Prevention, conducts near–real-time, population-based, active surveillance for vaccine safety. One of the steps in analyzing signals, if there are enough cases, is to apply temporal scan statistics. The purpose is to determine if the cases clustered in time within an overall a priori defined post-vaccination observation interval. We presented a relatively efficient and accurate algorithm for the purely temporal scan statistic as applied to vaccine safety investigations. It only needs SAS/BASE® software, and the algorithm is simple enough to be programmed in another software languages. Our present work is focused on incorporating the temporal scan statistic algorithm within our previous approach for finding an optimal risk window for studies of vaccine safety.