Written and maintained by CASRAI Editorial Board
Last updated
What reference safety information actually is
Reference safety information (RSI) is the specific, version-controlled statement of a product’s known and expected adverse reactions that a sponsor designates as the benchmark for deciding whether a new adverse event is expected or unexpected. It is not the whole safety profile of a drug and not the whole Investigator’s Brochure (IB) — it is a defined subset of that document, deliberately separated out so it can be identified, cited, and updated on its own timeline rather than requiring a full IB revision every time expectedness needs re-checking.
RSI matters because expectedness is one of the three elements that turns a serious, causally-plausible adverse event into a SUSAR, alongside seriousness and suspected causality. Per ICH E2A, an event is unexpected when its nature or severity is not consistent with the applicable product information — and RSI is the document that operationalizes “applicable product information” into something a safety reviewer can actually check a case against.
Where the RSI lives — IB section vs. approved labeling
For an investigational product still under an IB, the RSI is drawn from the IB’s Summary of Data and Guidance for the Investigator section — the section that exists specifically to give investigators a working understanding of a product’s risks so they can recognize and manage them. Because this section is the one that has to function as an expectedness reference for safety reporting, sponsors commonly present it (or a clearly labeled subset of it) as its own version-controlled unit within the IB — sometimes a dedicated subsection, sometimes a standalone appendix — so a new RSI version can be issued without triggering a full IB rewrite every time.
The moment a product has an approved marketing authorization and is being used as a comparator or background therapy, the reference point switches: RSI is then drawn from that product’s approved labeling — the Summary of Product Characteristics (SmPC) in the EU, or equivalent approved labeling in other jurisdictions — not the investigational IB. This distinction matters for multi-arm trials using a licensed comparator: the investigational arm’s expectedness is assessed against the IB’s RSI, while adverse events on the comparator arm are assessed against that product’s own approved label, and the two documents are not interchangeable.
Whichever document supplies it, the operative discipline is the same: expectedness is decided against a named, versioned reference, not against a reviewer’s clinical judgment of how alarming an event sounds. A sponsor that can’t say which RSI version was in effect when a given case was assessed has a traceability gap that shows up in an inspection.
Keeping the RSI current
The IB — and therefore the RSI it contains — is expected to be reviewed at least annually and updated whenever significant new safety information becomes available, consistent with ICH E6‘s ongoing-review expectations for investigator brochures. In practice this happens on two tracks: a scheduled annual refresh that rolls in accumulated trial and post-marketing experience, and an unscheduled update triggered by a specific finding — a new class-effect signal, an emerging safety trend across sites, or a regulator-requested labeling change for an approved comparator.
Ownership sits with the sponsor’s safety/medical function, not with individual investigators — a site cannot locally decide an event is “already known” and downgrade its own reporting obligation. Distribution has to keep pace with the update: every site running the protocol, plus the relevant ethics committees and regulators, needs the current RSI version, because assessing a case against a stale RSI produces the wrong expectedness call regardless of how carefully the rest of the assessment was done.
What actually changes when the RSI is updated mid-trial
An RSI update does not rewrite history. The core mechanic to hold onto: expectedness is assessed against the RSI version that was in effect at the time the case was evaluated, not against whatever version happens to be current when someone later reviews the file. A reaction correctly classified as unexpected — and reported as a SUSAR — under RSI version 3 does not get retroactively reclassified as expected just because RSI version 4, issued two months later, now lists that reaction. The original classification and the report that went to regulators stand as made.
What the update does change is everything going forward from its effective date:
- New cases occurring after the new RSI’s effective date are assessed against it. The same event that was unexpected under the old version may now be expected under the new one — or vice versa, if a signal review actually removes a previously-listed reaction.
- Periodic aggregate reporting — the DSUR — has to reconcile cases that straddle an RSI change within a single reporting interval. A DSUR covering a period in which the RSI changed mid-way needs to make clear which version applied to which cases, rather than silently reassessing the whole period’s case list against whichever version happens to be current when the DSUR is written.
- Site and IRB/EC notification becomes its own compliance action. A materially updated RSI is safety-relevant information in its own right — sites need to be told what changed and by when the new version takes effect, and study staff applying an outdated RSI to a live safety assessment is a distinct, commonly-cited finding in GCP inspections.
- Informed consent and protocol documents that summarize known risks may need review if the RSI change is substantive enough to affect what a participant is told — a new expected-but-previously-unlisted reaction, or the removal of one, is the kind of change that can trigger a consent form update independent of the safety-reporting mechanics.
The practical takeaway for anyone running a trial’s safety operations: track RSI version alongside case data as a first-class field, not an afterthought. “Which RSI was this case assessed against” needs to be an answerable question for every SUSAR on file, indefinitely — not just at the moment the case was closed.
Frequently asked questions
Is reference safety information the same as the Investigator’s Brochure?
No. The RSI is a specific, defined subset of the IB — drawn from its Summary of Data and Guidance for the Investigator section — used specifically for expectedness assessment. The IB as a whole covers pharmacology, nonclinical findings, and clinical experience far more broadly than the RSI does.
What is the RSI for a trial using an approved drug as a comparator?
The comparator’s own approved labeling — the SmPC in the EU, or equivalent approved labeling elsewhere — not the investigational product’s IB. Events on the comparator arm are assessed against that label, not against the IB governing the investigational arm.
Does updating the RSI reclassify SUSARs that were already reported?
No. Expectedness is assessed against the RSI version in effect when the case was evaluated. A later RSI update governs cases going forward from its effective date; it does not retroactively change the classification, or the report already filed, for cases assessed under the prior version.
Who decides when the RSI needs updating?
The sponsor’s safety/medical function, not individual investigators or sites — driven by the IB’s expected annual review plus any unscheduled update triggered by a significant new safety finding. Sites apply the current RSI they’re issued; they don’t locally reinterpret it.
For the classification mechanics RSI feeds into, see SUSAR and serious adverse event (SAE). For the case-level workflow a SUSAR moves through, see ICSR Processing Workflow and, for the EU-specific reporting timelines, GVP Module VI. For how RSI-governed expectedness rolls up into periodic safety reporting, see DSUR and PSUR/PBRER Format.
