Data mining in the US Vaccine Adverse Event Reporting System (VAERS): early detection of intussusception and other events after rotavirus vaccination

Abstract

The Vaccine Adverse Event Reporting System (VAERS) is the US passive surveillance system monitoring vaccine safety. A major limitation of VAERS is the lack of denominator data (number of doses of administered vaccine), an element necessary for calculating reporting rates. Empirical Bayesian data mining, a data analysis method, utilizes the number of events reported for each vaccine and statistically screens the database for higher than expected vaccine-event combinations signaling a potential vaccine-associated event. This is the first study of data mining in VAERS designed to test the utility of this method to detect retrospectively a known side effect of vaccination–intussusception following rotavirus (RV) vaccine. From October 1998 to December 1999, 112 cases of intussusception were reported. The data mining method was able to detect a signal for RV-intussusception in February 1999 when only four cases were reported. These results demonstrate the utility of data mining to detect significant vaccine-associated events at early date. Data mining appears to be an efficient and effective computer-based program that may enhance early detection of adverse events in passive surveillance systems.