Writing a Useful Deviation Record for Research Work
Anchor the record to the intended work
Identify the study, procedure and version that applied when the event occurred. Quote or reference the relevant requirement, then describe the observed difference. A deviation cannot be understood if a reviewer cannot establish what was meant to happen in the first place.
An example is: “Study S-08 required instrument check IC-04 before analysis; check was documented after the first sequence; affected sequence identifiers listed in attachment A.” This gives a concrete boundary. “Procedure not followed” is too vague to show which work may need assessment.
Preserve the event as first recorded
Capture who noticed the issue, when it was discovered and what is known about its timing. Distinguish direct observations, instrument evidence and retrospective accounts. Where timing remains uncertain, write the uncertainty explicitly rather than assigning a plausible time to complete the form.
The OECD test-item management guidance offers a risk-based lifecycle context for evaluating material-related issues. A local research deviation should use evidence proportionate to the actual question. Preserve the original affected records and connect them to the deviation rather than editing them to make the discrepancy disappear.
Assess impact before selecting data
List affected samples, results and comparisons. Consider whether the deviation could introduce bias, invalidate a prerequisite or make a result harder to interpret. Record the reasoning and its limitations, including what further evidence could resolve uncertainty. A conclusion of “no impact” needs a rationale that a reviewer can follow.
The NC3Rs DRIVER recommendations on exclusions encourage explicit criteria for anticipated exclusions and transparent reporting. Their preclinical context is useful when planning research decisions, but does not replace the requirements of a particular study. Avoid deciding which data to retain solely because the result agrees with expectations.
Close the loop with accountable actions
Separate immediate containment, the study decision and any action to reduce recurrence. For each action, identify the owner, due date and evidence of completion. “Staff reminded” may describe a communication, but it does not by itself establish whether the underlying source of the problem was addressed.
A final note might say: “Affected results retained with limitation stated; check evidence reviewed by project lead; procedure revision scheduled under action A-17.” The research hub provides wider reading on evidence assessment. Keep the approved disposition, rationale and action history accessible from the study file so later reporting reflects what actually happened.
Related research reading
- Planning Research-Sample Disposal Through Laboratory Procedures
- What Appearance Can Tell You About a Reagent—and What It Cannot
Explore the research blog and original batch-document library for further context.