OBJECTIVES : Spontaneous reporting systems (SRS) of adverse drug events are important sources for postmarketing evaluation and monitoring of adverse drug events. However, these sources have limitations. Common sources of bias in signal detection and clarification utilizing SRS are described to guide researchers on appropriate application of methods and interpretation of findings from SRS.

METHODS : A hybrid of literature review and application of disproportionality analysis to the FDA Adverse Event Reporting System (FAERS) was used.

RESULTS : Pharmacovigilance analysis of SRS data is encountered with a myriad of challenges that stem from the characteristics of data source which bias the interpretation of identified signals. The following sources of bias are described with real-world examples: confounding; masking effect; Weber effect; ripple effect; multiplicity effect; polytherapy bias; notoriety bias; stimulated reporting bias; reporting bias; misclassification and diagnostic bias; and limitations of disproportionality analysis. Examples are provided from published literature and from empirical data mining testing of the FAERS.

CONCLUSIONS : Despite the challenges of passive safety surveillance, spontaneous reporting of adverse drug events is a crucial source for drug safety management. Identified signals should be interpreted in light of the inherent limitations of SRS, and sources of bias that affect signal interpretation should be considered. A checklist of potential sources of bias and confounding is recommended.