Pharmacovigilance from first awareness to a valid case, a report for each destination and a governed signal.

1Similar reports are not yet a signal.

Several reports arrive through different countries and channels with different verbatim descriptions. They may be duplicates, background disease, a known adverse reaction, a product-quality issue, a reporting artefact or new safety information. A count alone does not settle which.

Seal keeps each report, its source and every later judgement on connected records, so reviewers can see what the evidence supports and what it does not. The same records carry the case from first awareness to a valid case, a report for each destination, an aggregate analysis and, where the evidence warrants it, a governed signal and a benefit-risk action.

1.1Why teams choose Seal for pharmacovigilance

A standalone safety database holds the case while clinical events, complaints, deviations, product registrations and approved documents live in other systems, so every aggregate report and signal assessment starts by collecting evidence again. Seal puts the safety case, reportability decisions, ICSR lifecycle, aggregate analysis and signal governance on the same record model as that evidence, and a reporting rule changes as versioned configuration. A due date can be explained from the rule that produced it, and a signal review starts from connected evidence rather than a fresh collection exercise.

EDC and CTMS supply clinical context, complaints and deviations supply product-quality evidence, and regulatory records supply products, markets and commitments.

2Build a valid case from the source.

Intake keeps the channel, sender, country, language, source document, verbatim narrative and attachments as received. Transcription and translation sit beside the original rather than replacing it, and the record shows when the organisation first had enough information to begin evaluation, because reporting clocks depend on that date.

Minimum criteria (an identifiable patient and reporter, a suspect product and an adverse event) are assessed according to the applicable process, and each outcome keeps its rationale: valid, potential, invalid, product complaint, medical information or follow-up needed. Seriousness, pregnancy, medication error, lack of effect or a suspected quality defect can drive priority and routing.

Duplicate management protects case counts. Candidate matches come from patient characteristics, dates, products, events, narrative similarity and external identifiers. A decision to merge, link, keep separate or nullify preserves the source reports, submitted identifiers and reviewer rationale.

Follow-up is specific to the question and the source: what information is missing, who was asked, by which route and what came back. New information creates a case version, which may change seriousness, expectedness, causality, reportability, due dates or prior submissions.

3Keep coding and medical judgement separate.

Coding keeps the reported term beside the MedDRA lowest-level and preferred terms, the dictionary version and any recoding. When the dictionary is upgraded, the change is assessed, and historical submissions retain the terms and version actually used.

Medical assessment records seriousness, expectedness against the applicable reference, reporter causality and company causality as separate judgements, each with its evidence: temporal relationship, dechallenge and rechallenge, alternative causes, medical history and concomitant therapy. An algorithm may support consistency, but the judgement stays with qualified reviewers.

Case quality is reviewed against the same record: completeness, coding, chronology, narrative, assessment, reportability and due-date logic. Sampling and error trends can target training or process change without obscuring individual corrections.

4Report by destination and track every message.

Reportability depends on context: clinical or postmarketing status, authority and country, seriousness, expectedness, causality, the clock-start rule, the reporting party, partner agreements and the version of the rule applied. Seal calculates each destination’s obligation from configured requirements and shows the rationale behind the due date, rather than applying one global seven- or fifteen-day rule.

Each ICSR message has its own lifecycle: identifiers and version, message type, sender and receiver, E2B data elements, validation errors and warnings, transmission, acknowledgement, rejection and resubmission. An accepted acknowledgement closes the technical loop. It does not replace medical review of the case or outstanding follow-up.

5Analyse a frozen population, then govern the signal.

An aggregate report is defined by its product and authorisation scope, reporting interval, data-lock point, case inclusion rules and exposure source. Line listings and summary tabulations are reproducible queries against that frozen population. Safety scientists remain accountable for the medical analysis, limitations, benefit-risk interpretation and conclusions.

Swipe to see the full diagram →

Signal management begins after detection. A disproportionality score is one detection input, not proof of a signal.¹ The signal record keeps the data source, search strategy, MedDRA scope, case series, literature and confounding considered, and moves through potential, validated, refuted, closed and reopened states with the evidence and governance decision behind each.

When assessment leads to action (a new study, a labelling change, a risk-minimisation measure, a healthcare communication, a product-quality action or a deliberate decision to change nothing), implementation and effectiveness remain linked to the signal and to the markets it affects.

6Keep partners on the same case history.

A pharmacovigilance agreement defines who does what: product and territory scope, day-zero exchange, case and follow-up transfer, submissions, literature, signals and aggregate reports. Exchange records and reconciliation show whether both parties hold the intended cases and versions, so a missing case is found during reconciliation rather than at an inspection.

7Start with one case through its full lifecycle.

Follow a representative case from first awareness through validity, duplicate search, follow-up, coding, medical assessment, reportability for each destination, ICSR generation and acknowledgement. Include a follow-up that changes seriousness or expectedness, and a duplicate that must be linked rather than counted twice.

Then extend the same records into aggregate reporting and signal management. Replacing an existing safety database depends on your products, countries, volumes, gateway arrangements, dictionaries, migrations and partner processes, so scope those before committing to a cutover.