Table of contents
At a glance
What needs to keep working during a safety database outage?
Provide a controlled way to receive and preserve new information, assess reporting deadlines, process cases and make required submissions. The fallback procedure should distinguish a local outage from a confirmed EudraVigilance disruption. After restoration, reconcile open cases and transmission statuses. A reachable application or transmission log does not demonstrate that the backlog has been cleared.
- Acceptance, evaluation, notification generation, and transmission can occur independently or fail separately.
- A system failure does not automatically suspend reporting deadlines.
- Rerun is only considered complete upon controlled post-processing and status reconciliation.
Identify the affected PV process
This article concerns pharmacovigilance for authorised human medicines and the relevant EU submission routes. GVP Module I addresses critical PV processes and their continuity. The EU ICSR Implementation Guide distinguishes technical transmission problems. A general IT continuity plan and GMP requirements need their own scope assessment.
First, determine what is actually affected. Can personnel receive alerts but still be unable to open cases? Is only the XML generation impaired? Has the message been sent successfully, but the processing at the recipient end remains unclear? These distinctions determine which information remains available and whether any alternative action is meaningful.
Four disturbance scenarios with different subsequent steps
Our own work matrix links the failure profile with an open technical question. It does not replace any current technical procedure directive from EMA or your service provider. Specifically, prior to using alternative channels, verify that these are designated and available for the specific situation.
| Disruption | Process question | Suitable evidence |
|---|---|---|
| Safety database not accessible | Which inputs and amendments remain outside the system? | Protected Alternate Register with Assignment and Receipt |
| News creation disturbed | Which evaluated cases can still not be dispatched? | Open submission list with prioritization |
| Transport disrupted | What messages have reached the intended recipient? | Transport status and documented clarification |
| EMA processing or unclear feedback | Which cases are accepted, rejected, or still open? | Associated Acknowledgements and current outage reference |
Ensure availability of alternative information during backup operation
A pre-prepared emergency register should clearly display case reference, information source, entry time, and responsible processing. Additional information required depends on the specific process. Access control and appropriate safeguards apply to temporary aids as well; a publicly accessible spreadsheet is not suitable merely because it is labeled as an emergency register.
Retain attachments with their assigned reference and actual receipt date. Known case numbers facilitate future transfer, but lack of database access may temporarily prevent assignment. Explicitly indicate this uncertainty. A preliminary entry must not be considered a final case evaluation or an already reported authority notification.
Assess deadlines during a disturbance from a technical standpoint
However, the EU ICSR Implementation Guide also regulates this case: Prevents the failure to generate a safety message and does not comply with endangered 7-, 15- or 90-day deadlines, if the receiving organisations are to be informed about the problem and expected time for the correction. After remedial action, open cases are transmitted as soon as possible (I.C.2.1.6.1).
If the own gateway fails while valid messages are being generated, I.C.2.1.6.2, as notified by the EMA IT Service Desk, specifies transmission via CD-ROM or DVD, or alternatively through an existing Eudralink access. For physical media, the delivery proof is considered a compliance document. Prior to use, verify the current service and security instructions; the alternative pathways mentioned in the guideline must not be equated with arbitrary channels.
Validated EMA disruptions must be distinguished from such cases. Section I.C.2.1.6.3 includes a specific compliance calculation under its prerequisites when remediation is completed within two EMA working days following the reavailability of the gateway. Neither a standalone system failure nor a support ticket alone triggers this requirement. Additionally, the original awareness timestamp of an incident must be maintained separately from its subsequent database entry.
Incident scenario: five submissions with three different statuses
Example: Following a system failure, five expected case submissions require processing. Two have been confirmed with a corresponding positive case acknowledgment. One has been rejected; two others remain unresolved. All five are visible in the dispatch list, but only two meet the defined closure status of a confirmed acceptance.
Two confirmed, one rejected, and two unresolved submissions result in five expected submissions. Initially open are three processing or clarification tasks; no three new safety incidents have been identified. A partially or fully rejected safety message containing valid cases requires, upon feedback, immediate correction and resubmission. In the absence of feedback, clarification of the actual receipt and processing status is required first.
Would the team retransmit all five messages without review, it could lead to additional ambiguities. Therefore, the intended workflow evaluates existing feedback, message assignment, and current recipient information. The timing for a retransmission depends on the specific error and receipt status; the table does not assert a universal retransmission rule.
| Procedures | Status | Count | Next step |
|---|---|---|---|
| A and B | Positive fall assignment assigned | 2 | Preserve the confirmed state and avoid an unjustified retransmission |
| C | Rejected | 1 | Clear errors and required correction |
| D and E | Processing unclarified | 2 | Determine status; justify next step |
| Total | Five expected submissions | 5 | Close three open tasks strategically |
Separate out and complete data transfer and backpressure handling
Following technical recovery, information secured during the disruption must be transferred into the intended workflow. Verify alignment with existing cases, appendices, entry times, and previously performed actions. Otherwise, manually edited follow-ups may appear as new cases, or previously transmitted statuses may be reprocessed as pending.
The technical restore and professional residue clearance each require a clearly defined status. A restore can reinstate application availability without retaining information generated after the backup timestamp. Therefore, compare the processes conducted during the failover operation with the restored dataset and the transmission feedback. The required scope is determined by the actual data pathway and the extent of the outage.
Checklist for a Viable Failure Response Procedure
For small teams, a concise and accessible procedure with readily available contacts is often more practical than an extensive document that no one can access during an outage. Assess the alternate procedure based on realistic scenarios, including a previously known case with a new addition.
The checklist is a standalone planning aid. It does not specify a fixed general recovery time and does not require a complete functional retest of all functions following each incident.
- Is the disturbance technically and functionally isolated?
- Are the entry point, source, and dependencies securely and permanently available?
- Are the Responsible Party and Representative reachable?
- Who assesses expiry effect and necessary communication?
- Are acceptable alternative routes known and accessible?
- Are acceptance, rejection, and unclear status handled separately?
- Have backup registers, databases, and feedback been reconciled after restart?
- Are the impacts and required actions demonstrably evaluated?
Draw targeted actions from the incident
After remediation, assess whether information was lost, incorrectly assigned, or delayed in processing or reporting. Distinguish between confirmed impacts and remaining open questions. A delayed technical response does not automatically indicate that the report was not received; conversely, a successful transmission does not constitute confirmation of technical acceptance.
Take actions rooted in the actual root cause: lack of access to the backup register, unclear representation, misinterpreted feedback, or incomplete reconciliation. The key factor is whether the corrected procedure functions effectively in the next occurrence of a similar incident. The effort required is more significantly influenced by data pathways and open responsibilities than by the number of pages in the emergency response plan.
Project experience and collaboration
Test the fallback procedure with outstanding submissions
With Pharmacovigilance and IT, we simulate a disruption using prepared valid, rejected and unresolved messages. We test which information about receipt, transmission and acknowledgement remains available. A gateway fault needs a different fallback from a database that cannot generate a valid message. We plan communications and reconciliation so outstanding reports can be processed after restart without creating duplicates through uncoordinated retries.
Frequently asked questions about Safety database outages
Does daily backup suffice as a PV outage concept?
No. While a backup supports recovery, it does not address how new information is received, deadlines are assessed, and required notifications are processed during a disruption. Similarly, it remains unclear how interim data is resumed upon system restart. These responsibilities require a continuity procedure specifically tailored to the particular PV process.
Sources: [1]
Does a failure of our Safety Database extend the reporting deadline?
No. A system failure within the organization does not extend the reporting deadline. If 7-, 15-, or 90-day deadlines are at risk due to a disturbance that cannot be promptly resolved, I.C.2.1.6.1 requires the recipient organization to be informed of the expected repair timeframe and the prompt transmission of open cases following repair. Special compliance considerations for confirmed EMA-reported incidents must be separately evaluated and are not transferable to internal system outages.
Is a Message Disposition Notification sufficient to constitute case acceptance?
No. The Guide distinguishes the technical receipt confirmation at the transport level from further processing and acknowledgements. Assign each response to the correct message and corresponding case. A positive transport status must not be treated as general evidence of accepted case transmission.
Sources: [3]
What does lack of feedback mean?
First, an undefined status. A Gateway acknowledgment and a technical processing response are distinct forms of evidence. In the case of the EudraVigilance processing disruption described in I.C.2.1.6.4, notifications may have already been received while the acknowledgment is still pending. If the acknowledgment is missing for two working days, you may contact the EMA IT Service Desk for clarification. A confirmed rejection, however, requires immediate resubmission with correction; blindly retransmitting due to lack of a processing response may result in duplicate records.
Sources: [3]
Can we simply select an alternative distribution route in the event of a disruption?
Only the predefined pathway for the specific failure scenario. In the event of failure of the own gateway, the EU ICSR Implementation Guide I.C.2.1.6.2 specifies to inform the EMA IT Service Desk via CD-ROM, DVD, or an existing Eudralink access for valid safety messages. The submission of CIOMS-I forms by fax or post to the EMA is explicitly excluded. Please review the current EMA service instructions and data protection measures before utilizing the agreed alternative procedure.
What information should a preliminary failure register contain?
As a core support tool, the following information must be clearly visible for assignment and further processing: source, entry point, known case reference, new content, responsible party, and open status. The specific scope shall be determined by your data flow. Access protection and traceability must not be compromised due to temporary usage.
Must every message be resent following restart?
No, not unverified. Verified assumptions, genuine rejections, and unresolved statuses require distinct handling steps. First, clarify message assignments and existing feedback. A repeat should correct the specific error and must not be achieved through a generic restart, which could generate additional duplicates or contradictory processing states.
Sources: [3]
When is the failure technically considered closed?
When the relevant information and processes are controlled, ongoing transmission statuses are clarified, and potential impacts are assessed, the application may technically be reactivated. For closure, remaining tasks and responsibilities should be clearly identifiable; a green availability monitor does not replace this evaluation.
Must the entire software be revalidated after each failure?
No, not after every failure. The need for additional validation evidence depends on the root cause, changes made, and actual impact. A recurring failure without modification, a corrected configuration, and a new software version represent distinct scenarios. Investigate the affected functions and data and justify which additional checks support safe continued PV operation.
Sources: [1]
How can a small team effectively validate an alternate procedure?
Use a small number of meaningful scenarios: a new entry, a follow-up to an existing case, and an unexplained submission. Assess responsibility, information security, and reconciliation. This is a standalone testing recommendation; the scope and frequency should be based on your process and risk profile, not on a universal number of exercises.
Sources and review date
Source date: 7 October 2026. The EU ICSR Implementation Guide is based on Revision 2 issued in March 2021. Specific current alternate channels and technical operational information must be verified on a case-by-case basis. The numerical example is a standalone reference document.
- EMA: GVP Module I – Pharmacovigilance Systems and Quality Management SystemsEMA/541760/2011; applicable from 02.07.2012; I.B.10–11.3, p. 9–12
- EMA: GVP Module VI – Collection, Processing, and Reporting of Suspected CasesRev. 2, dated 28.07.2017; applicable from 22.11.2017; VI.B.2–B.5 and B.7, pages 14–18 and 21–22
- EMA: EU ICSR Implementation GuideEMA/51938/2013 Rev. 2, March 2021; I.C.2.1.6–.8, p. 17–21; Differences between MDN, Notification and Case Acknowledgements
Sources checked: .
Practically prepare for the PV-standby operation
We help you translate failure scenarios into concrete responsibilities, suitable tools and verifiable restart criteria. We combine professional PV processes with actual IT dependencies.
- Computer system validation →
Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

