Signal management as a Swiss authorisation holder
A signal is information that makes a new or changed association between a product and an event look possible. Signal management is the documented cycle from detection through validation, prioritisation and assessment to a decision and its closure. With small Swiss case numbers the deliverable is the reasoning, not the statistics.
- Last reviewed:
- Reading time: 4 min
- Reviewed by Swiss Pharmacovigilance
What is a signal?
A signal is a hypothesis, not a finding. It can arise from a single well documented case if the reaction is characteristic, and it can fail to arise from fifty vague ones. What turns information into a signal is that it would change how the product is used if it held up.
Where signals come from
Restricting detection to your own case database is the most common design error in a small pharmacovigilance system, because the Swiss case count is usually too low to surface anything. Most signals relevant to a Swiss authorisation arrive from outside it.
- Your own spontaneous cases, including the qualitative detail in the narratives
- Literature, from the documented interval search
- Actions by other regulators: label changes, referrals, safety communications
- The global safety database of the licensor or affiliate
- VigiBase and other public data sources
- Class effects reported for related products
- Quality trends and complaint patterns
- Clinical trial and post-authorisation study findings
The signal management cycle
Each step has an owner and produces a record. The cycle is not complete when a decision is made: it is complete when the action has been taken and the signal has been formally closed with its outcome.
- Detection: review new information from every source on a defined cadence.
- Validation: is there enough substance to justify further work, or is this noise?
- Prioritisation: does this need action now, or at the next periodic review?
- Assessment: gather cases, literature and exposure and weigh the evidence.
- Recommendation: state what should change, or that nothing should.
- Decision: the RPPV decides and the reasoning is recorded.
- Action: label variation, communication, RMP measure or continued monitoring.
- Closure: confirm the action is done, record the outcome and close the signal.
Step, owner, evidence, timing
The table below is what an inspector reconstructs from your signal log. If a step has no owner or no date, the log is a narrative rather than a control.
| Step | Owner | Evidence | Typical timing |
|---|---|---|---|
| Detection | Pharmacovigilance | Source review log | Continuous, reviewed monthly |
| Validation | Safety physician | Validation note with the rationale | Within 2 weeks of detection |
| Prioritisation | RPPV | Priority assigned in the signal log | With validation |
| Assessment | Safety physician, PV | Assessment report with the data used | 4 to 8 weeks, sooner if urgent |
| Decision | RPPV | Minuted decision with reasoning | Within 2 weeks of assessment |
| Action | Regulatory, PV, medical | Variation, communication or RMP update | Per the action type |
| Closure | RPPV | Closure entry with the outcome | When the action is complete |
Small case numbers: what helps instead of statistics
Disproportionality analysis needs volume that a Swiss portfolio rarely has. Running it anyway produces noise and a false sense of rigour. What works instead is qualitative: read the narratives, look at time to onset and the course after withdrawal, compare against what the class does, and align deliberately with the global signal process so that a signal found elsewhere reaches the Swiss assessment.
Document the method you actually use
An SOP that promises quantitative signal detection on a database of forty cases a year will fail its own audit. Describe the qualitative review you really perform, on the cadence you really keep, and evidence it. An honest small method beats an aspirational large one.
Documenting a decision so an inspector can follow it
The test is whether someone who was not in the room can read the file and understand why the decision went the way it did. That needs four things in the record: what information triggered the review, what data was considered, what alternatives were weighed, and who decided what on which date. A signal log with statuses but no reasoning fails that test even when every decision was right.
Frequently asked questions
How often do we have to look for signals?
Detection is continuous, with a documented periodic review. Monthly is common for an active portfolio and quarterly can be defensible for a small, stable one, as long as the cadence is written down and kept.
Can a single case be a signal?
Yes. A single well documented case can constitute a signal where the reaction is characteristic, the time course fits and there is no plausible alternative explanation. Rare, serious events are usually found this way rather than statistically.
Do we have to report a signal to Swissmedic?
A validated signal that changes the benefit-risk assessment, and any action you take on it, belongs in your communication with Swissmedic and in the PSUR. Urgent safety issues are raised without waiting for the periodic report.
Can we rely on our licensor's signal management?
You can use their work, but you remain responsible for the Swiss assessment. Set the exchange out in the SDEA, receive their signal outputs, and record your own Swiss decision, including where you agree with theirs.
What do we do with a signal we decide not to act on?
Record it, with the reasoning and the data considered, and close it or keep it under observation with a review date. An explicitly closed signal with a stated rationale is a strength in an inspection; a signal that simply disappears from the log is not.
Do you need an RPPV in Switzerland?
We act as the qualified person responsible for pharmacovigilance, run case processing and write the PSURs for your Swiss authorisations.