Optimization of a quantitative signal detection algorithm for spontaneous reports of adverse events post immunization

Abstract

Within the scope of this study, a unique MGPS algorithm across vaccines with the original full stratification but a lowered cut-off provided major performance improvement.

Good Signal Detection Practices: Evidence from IMI PROTECT

“…This unexpected finding that has not been reported elsewhere has important implications because a number of organisations routinely use stratification in their analyses, but very few routinely employ subgrouping. Other previous studies [ 36 , 38 ] have observed modest improvements for stratified analyses consistent with the results from the stratified analyses in the PROTECT study and therefore these findings from the other studies may also be artefacts from the stratification process rather than a true effect.…”

Application of biclustering algorithm in adverse drug reaction monitoring system of China

“…Additionally, the disproportional signals were classified as true positives if they corresponded to known ADRs and otherwise classified as false positives. This approach has been used in other studies . Therefore, the product label database collected by the National Center for ADR Monitoring of China was used as a good proxy for a gold standard.…”

“…This approach has been used in other studies. 1,15,[18][19][20] Therefore, the product label database collected by the National Center for ADR Monitoring of China was used as a good proxy for a gold standard. This database listed 45 393 drug-ADR pairs that were considered as true signals.…”

Adjuvanted (AS03) A/H1N1 2009 Pandemic Influenza Vaccines and Solid Organ Transplant Rejection: Systematic Signal Evaluation and Lessons Learnt

“…Signal detection activities use all available data sources including spontaneous reports, review of the literature, clinical trials, and epidemiological studies specifically designed to evaluate the signal. Disproportionality analyses assess the frequency of a vaccine–event pair relative to background reporting in a pharmacovigilance database [42, 43] and aim to detect greater-than-expected reporting of vaccine–event combinations. Time-to-onset methods may be used to detect whether the distribution of the time-to-onset of a vaccine–event pair within a pre-defined period, days after vaccination, is different than what is expected for the reported time-to-onset distribution of the event of interest after other vaccines, and the reported time-to-onset distribution of other events following the vaccine of interest [43–45].…”