Safety Signal

A safety signal is information suggesting a new or changed causal association between a medicine and an event — enough to justify investigation, but not yet established. The distinction matters: a signal is a hypothesis, not a conclusion. Treating every signal as a confirmed risk would make labels unusable; ignoring signals is how avoidable harm continues.

Signal management is a defined sequence. Detection can be qualitative — a clinician’s report of something unexpected — or quantitative, using disproportionality statistics across a spontaneous-report database to find combinations reported more often than expected. Validation asks whether there is enough evidence to warrant analysis, weeding out duplicates, confounding, and already-known effects. Assessment then weighs all available evidence, and if the association is confirmed, action follows and is communicated.

Two failure modes recur in inspection. The first is detection without decision: signals are logged, tracked, and never resolved. The second is undocumented dismissal — a signal is closed as not validated with no recorded reasoning, which is indistinguishable from not having looked. GVP Module IX sets expectations for timeframes and documentation precisely because the decision trail is the evidence that the system works.

• —A signal is a hypothesis warranting investigation, not an established risk.
• —Detection is qualitative (clinical reports) and quantitative (disproportionality analysis).
• —Validation filters duplicates, confounding, and already-labelled effects.
• —Confirmed signals drive label changes, risk-minimisation measures, or restrictions.
• —Document the reasoning for closing a signal — an undocumented dismissal looks like no review.
• —Detection without timely decision is the most common inspection failure.

EMA GVP Module IX (Signal Management) with its Addendum I on methodological aspects; Commission Implementing Regulation (EU) No 520/2012; the EudraVigilance data analysis system; US expectations under 21 CFR 314.80 and FDA post-marketing surveillance programmes.