Regulations and compliance · Practical guide

GxP go-live with open deviations: decision criteria

Findings remain open shortly before the planned start. Some have little effect on use, while others may conceal a critical data error. Assess the actual consequences and establish effective controls before deciding on any limited use.

Table of contents

At a glance

Can a GxP system go into use with open test deviations?

Yes, in some circumstances. Demonstrate suitability for the intended use, with defined restrictions where needed. The authorised function assesses the impact, effective controls and residual risks and records its decision. Annex 11 calls for reports on observed validation deviations (4.2) and a justified, documented risk assessment (1). This is not general permission to overlook unresolved critical failures. Time or cost pressure does not demonstrate suitability.

Scope and Applicability

The contribution does not constitute general approval for a go-live with defects. The responsible organization shall decide within the existing quality system; necessary requirements must not be circumvented by reclassifying a deviation.

Sources: [1] [2]

Describe the finding and the affected usage in detail.

An open test deviation initially represents a difference between expected and observed response. For the decision on intended use, it must be clarified which functions, data, and processes are affected. A minor screen display error has a different impact than an incorrect batch assignment. The severity does not solely follow from the technical designation or repair effort.

Describe the expected result, actual result, reproducibility, and potential scope. Distinguish known facts from assumptions. An unknown root cause may significantly limit the decision. The note "low priority" without technical justification is not sufficient evidence of an appropriate assessment.

Annex 11 supports risk-based evaluation and traceable validation documentation. This does not imply general authorization to initiate studies with arbitrary open findings. Similarly, a universal "zero-defect" rule applicable to all uses is not asserted. The specific decision requires appropriate evidence and authorized responsibility.

Annex 11 section 4.2 calls for reports covering all deviations observed during validation. Under section 1, the extent of validation and data integrity controls follows a justified, documented risk assessment. A decision matrix supports this work but provides no regulatory basis for release on its own.

Sources: [1] [2]

Use a decision matrix for open items

Link findings, potential consequences, scope of impact, and existing controls. Assess whether the use of a function can be clearly restricted. A non-utilized function may be managed differently than a central function whose failure is not reliably detectable. Unknown data impact must not be deemed negligible solely due to a temporal impression.

Own work aid: Use a decision matrix for open items
QuestionOwn contentSignificance for the decision
What is wrong? Actual difference from the expected result Clear assessment result instead of generic error category
What could be affected? Functions, Data, Time Period and Users Scope that can be clearly bounded
How is the error detected? Verified Control Not just a theoretical discovery
Can usage be restricted? Specific allowed area No unclear global release
How is it closed? Responsible party, correction and retesting Traceable further path

Sources: [1] [2]

Example: Report layout versus incorrect result reference

Finding A concerns a truncated heading that does not affect a decision. Finding B concerns a result that can be assigned to the wrong sample. The technical fixes may take similar effort, but the potential consequences differ.

For A, it is assessed whether the content, unit, and required context remain unambiguous. If a heading nonetheless predisposes the risk of misidentifying two samples, the evaluation is adjusted. For B, the cause, affected scope, and potential for detection are investigated. Additional manual verification may only be warranted if it actually detects the specific sample misidentification and is practically executable.

The decision may establish different usage boundaries. A clearly independent pathway may be continued for use, while the affected result pathway remains locked. This boundary requires appropriate technical or organizational implementation. A signature on a general release does not substitute for such implementation.

Own work aid – Example: Report layout versus incorrect result reference
Test caseOwn actionExpected finding
Layout finding Read the representative report No relevant ambiguity
Assignment error Use known swapped reference values Control identifies the specific error
Limited usage Use only authorized areas The boundary can be enforced in practice
Compensation Manually verify the realistic process Efficacy and effort known
After correction Affected cases recur Root cause addressed and correct function verified

Sources: [1] [2]

Define transition checks and usage limits

If a transitional measure is envisaged, assess its effectiveness and practical feasibility. A manual reconciliation may be viable for a small number of transactions but will fail under high load. Therefore, responsibility, supporting evidence, and a clear end date must be defined.

Define precisely what may be released. An unaffected part of a process may remain in use while a critical function stays blocked, provided dependencies are understood and the restriction works. Record that specific decision and its limits.

Sources: [1] [2]

Practically validate compensation controls

An intermediate control must include scope, method, responsible party, and duration. "QA monitors additionally" is too vague. Specify the data, reference basis, and decision criteria when a difference is detected. Review a known failure scenario to ensure the control is not merely theoretical.

Assess capacity and frequency. A manual check of every case may be suitable in theory but impracticable in operation. Open cases must not silently acquire a reviewed status. Depending on the risk and process, define when to stop or escalate.

Define what happens in the event of a failure of control. Vacation, missing source or too high amount of process can limit its effect. The decision to use should know these conditions. A control whose conditions are not met cannot be used unchanged as a justification.

Sources: [1] [2]

Release, residual points, and subsequent closure are binding

The decision should identify system status, permitted use, open findings and controls. Responsible roles decide according to their own procedure. The technical implementation partner can provide information, but does not automatically take over the use in the process decision of the regulated company.

Open issues require a clear and traceable closure pathway. Identify correction actions, appropriate follow-up testing, and the decision to conclude the intermediate control. A corrected code without verification of the affected process may be insufficient. Assess the data generated during the intermediate phase if the findings require it.

Avoid a permanently ambiguous status. A deadline may support traceability but should not be treated as a mere procedural date replacing substantive assessment. If correction cannot be achieved in a timely manner, further evaluation of usage is required. The original decision remains valid only within its defined conditions.

Sources: [1] [2]

Learn structurally from open findings

Recurring deviations may point to unclear requirements, inappropriate tests or a weak change model. Treat this cause instead of just closing each individual case. The scope of a further measure depends on knowledge and risks.

Provide assessment logic and practical examples to the team. A system owner should be able to explain why a result restricts certain usage and what evidence is required to close the evaluation. A color code without justification provides little support for the next decision.

Sources: [1] [2]

Draft: document a limited usage decision

Complete the table for the specific finding and verify the intended control using a known failure. Consider data access, competence, and realistic effort. If a prerequisite is no longer met or if usage changes, the responsible party must re-evaluate the decision. A blanket release with hidden conditions is less effective than a clearly documented boundary.

Sample filling example – internal work aid
Decision areaExample entryAcceptance question
Finding Result attribution may be swapped. Is the potential consequence understood?
Scope Designated function and data scenarios. Are additional areas justified and defined?
Control Independent comparison of sample and result. Does she identify the specific defect?
Use Only a clearly defined area may commence. Is the border actually feasible?
Responsibility Authorized Decision and Responsible Party. Are responsibilities realistically assigned?
Closure Correction and appropriate retest. When does the intermediate control end?

Sources: [1] [2]

Next steps

  • Associate deviations with requirements and consequences of errors.
  • Assess risks and potential transition controls.
  • Define usage limits and decision-making authority for responsible parties.
  • Schedule and track the correction and evidence of effectiveness.
Further reading:

Project experience and collaboration

Prepare the decision on outstanding items for the responsible owners

With QA, IT and the process owners, we prioritise open findings by their actual effect on intended use. For limited outstanding issues, we examine possible controls and whether they will work in practice. If the consequences of a critical failure remain unclear, we recommend deferring the affected use. Your authorised decision-makers make that call. We organise the evidence, limits of use and follow-up of outstanding issues so they can make an informed decision.

Frequently asked questions about Release with outstanding findings

Are open documentation points always non-critical?

No. Missing evidence may preclude the assessment of suitability. The content of the gap is decisive.

Sources: [1] [2]

Can a workaround remain permanent?

Only if it is evaluated and continuously monitored for use as a permanent solution. A temporary exception should not become standard practice without notice.

Sources: [1] [2]

Is it generally possible to perform a go-live with open deviations?

Yes, under clear conditions. The impact and extent of the deviation must be known, and the intended use must still be appropriately managed despite the open point. Any usage limitation or compensatory control must be genuinely effective and validated. The authorized party makes the decision and continues to manage the remaining issue. Annex 11, Section 4.2 and Section 1 require deviation reports and a justified risk assessment; they do not grant a blanket go-live authorization. An unclarified critical data error cannot be mitigated by an additional signature.

Sources: [1] [2]

Does an additional signature serve as a compensating control?

Not solely sufficient. The individual must be able to identify a specific error and prevent inappropriate use. For this, information, criteria, time, and authority are required. Test known error scenarios that are relevant. A signature indicates a documented action, but does not automatically prove its effectiveness.

How do we differentiate a cosmetic from a professional laboratory investigation report?

Assess the actual effect. A truncated heading may be misleading or lead to a significant misinterpretation. The cosmetic designation is insufficient. Consider content, unit, record reference, and decision. The evaluation should be based on specific, representative expenditures.

Must open items be tracked further after go-live?

Yes, provided they remain part of the limited decision and have not yet been finalized professionally. Identify corrections, follow-up testing, responsible parties, and appropriate timelines. The initiation of a process does not automatically constitute completion of a finding. The conclusion of an intermediate control requires the corresponding supporting evidence.

Sources and review date

Annex 11 and Chapter 1 form the GMP framework for validation, deviations, and quality responsibility. The decision matrix is a standalone reference tool. It does not grant general authorization for use of a system with open critical defects.

  1. European Commission: EU GMP Annex 11, Computerised Systems (2011)Current version as per EudraLex inventory as of 14.10.2026. Application, risks, operation, and data lifecycle; specific testing proposals are derived independently.
  2. European Commission: EU GMP Chapter 1, Pharmaceutical Quality SystemQuality system, root cause analysis, CAPA and continuous improvement.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Prepare justification points for a legitimate usage decision.

We support you in assessing the current situation, clarifying responsibilities, and implementing appropriate evidence within operations. The type of support that is most effective depends on your system, your team, and the specific objective.

  • 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.

Send a project enquiry Customer testimonials →