The utility of telephone helpline data for real-time syndromic surveillance of adverse events following immunization: Retrospective evaluation

Abstract

Background: Vaccines are generally regarded as safe and adverse events following immunizations (AEFI) are mostly mild and self-limiting. However, serious AEFI may occur coincidentally or associated with vaccination. Possible AEFIs need to be identified and managed in a timely way to maintain public confidence in vaccines and minimize vaccine hesitancy. In most countries, passive surveillance systems predominantly employed for AEFI monitoring. However, these systems have well-known limitations, especially underreporting that have posed challenges to detect possible AEFI early. Objective: To evaluate the utility of telephone helpline data for syndromic surveillance of AEFI to augment existing systems in Victoria, Australia. Methods & Materials: Using data collected retrospectively from 1 February 2009 to 31 December 2017, weekly telephone helpline calls related to an AEFI were compared against the expected level estimated from historical data. In addition, we used the Victorian enhanced passive surveillance data as a reference to examine the timeliness of the telephone helpline data. The temporal pattern cross-correlation coefficient at different time lags was estimated as a measure of timeliness evaluation. Statistical temporal AEFI signal was examined using a statistical algorithm for monitoring count time series, specifically using the Farrington method. Results: During the study period, a total of 2,005,226 telephone calls were made to the telephone helpline service. Of these, 13,719 (0.68%) were AEFI related calls. In the same period, the Victorian enhanced passive surveillance received 10,367 AEFI reports. Cross-correlation analysis showed that the maximum positive correlation (r = 0.4) between the two datasets occurred at a negative lag of 1 week. For individual years, the cross-correlation coefficient was highest (r = 0.66) in 2010 at a negative lag of 2 weeks. Using a statistical algorithm, we were able to detect the 2010 incident of increased fever and febrile convulsions following seasonal influenza vaccination data 4 weeks earlier than it was detected at that time. Conclusion: Telephone helpline data was found to detect unusual AEFI occurrences earlier than the enhanced passive AEFI surveillance system. This study demonstrates that telephone helpline data could have promise as part of an integrated AEFI early signal detection system.