Signal Detection & Validation

Practice of validating signals using internal methodology and documentation without using the formal EMA signal validation form; acceptable when equivalent rigor is applied.

Pharmacovigilance Risk Assessment Committee (EMA); assesses signals, recommends regulatory action, and oversees RMPs and referrals.

Statistical method to compare reporting frequency of a drug–event pair vs expected (e.g., PRR, ROR, EBGM). Used in signal detection; not proof of causation.

Measure of disproportionality: (a/(a+b)) / (c/(c+d)) where a=drug+event, b=drug+other, c=other+event, d=other. Thresholds often ≥2 used as screening.

WHO global database of reported potential side effects of medicines; used for signal detection and reference by regulators and MAHs.

Information suggesting a potentially causal association between a drug and an event, which was previously unknown or incompletely documented.

End-to-end process: detection, validation, prioritization, assessment, and recommendation (e.g., label update, no action).

Documented output of signal detection (e.g., list of drug–event pairs, metrics, and recommendation for validation).

Structured evaluation of a signal (data, methodology, clinical relevance, conclusion) per GVP IX; may use EMA form or equivalent.

Ranking of signals for resource allocation and assessment order (e.g., by strength, novelty, public health impact).

Signal involving multiple cases in a short period or specific subgroup (e.g., batch, region); may indicate quality or subgroup issue.

Potential signal arising from published literature (case report, study); requires verification and assessment like other signals.

Group of MedDRA terms for searching or analysis (e.g., hepatic SMQ); used in signal detection and reporting.

Conclusion after validation that a signal represents a confirmed or potential risk; may lead to RMP/label update.

Conclusion after validation that available data do not support a causal association.

Validation does not confirm or refute; further data or monitoring needed.

Empirical Bayes Geometric Mean; disproportionality measure used in signal detection (e.g., FDA).

Reporting Odds Ratio; disproportionality metric for signal detection.

WHO method for signal detection using neural network; produces IC (information component).

How often signal detection is performed (e.g., quarterly); should be documented in PSMF.

EMA procedure for arbitration (e.g., safety concern); PRAC and CHMP involved.

Referral under Directive 2001/83 for safety (or efficacy) concern.

Referral under Regulation 726/2004 for centrally authorized products.

Referral to scientific committee (e.g., PRAC) for safety assessment and recommendation.

Recommendation from PRAC (e.g., SmPC change, PASS, referral); CHMP adopts for CAP.

Systematic distortion in AE reporting (e.g., under-reporting, media effect); considered in signal interpretation.

Increased reporting after public awareness (e.g., media); can inflate disproportionality.

Interval indicating uncertainty of estimate (e.g., 95% CI for rate ratio); used in signal and RWE.

Result unlikely under null hypothesis (e.g., p<0.05); one consideration in signal/RWE, not sole criterion.

Collection of cases with common feature (e.g., same drug–event); used in signal assessment.

Review of individual cases (narrative, causality, quality) as part of signal validation.

Expected rate of event in population without drug; used to interpret reporting.

Comparison of observed cases to expected (e.g., from background rate); signal evaluation.

Ability to reproduce result (e.g., signal detection run) with same inputs and version.

Execution of signal detection (e.g., quarterly) producing list of drug–event pairs for review.

Log of signals (detected, validated, closed) with dates and outcomes.

Signal that has been assessed and closed (e.g., refuted, or confirmed and action taken).

Signal under evaluation (not yet closed).

Meeting to review and decide on signals (prioritization, validation, action).

Signal validated using internal equivalent to EMA validation form; acceptable if equally rigorous.

Ongoing or more frequent signal detection (e.g., monthly) for high-risk products.

Signal that may apply to a class of drugs (e.g., same mechanism); assessment may be class-wide.

Signal specific to one product (not class).