13.2 Potential Fraud, Misconduct and Evidence Preservation

Potential Fraud, Misconduct and Evidence Preservation

Fabrication creates data or activities that did not occur; falsification manipulates records or processes so they misrepresent what happened. Examples can include invented visits, altered eligibility results, forged signatures or intentionally omitted safety information. The ECO requires recognising and reporting potential fraud and misconduct. A CRA's duty is to protect participants, preserve reliable evidence and escalate appropriately, not to conduct an unauthorised criminal investigation.

Many anomalies have innocent explanations. Repeated rounded values can reflect the instrument display or recording convention. A late electronic entry may be a delayed transcription from contemporaneous source. An identical visit time may be an imported default. These explanations still need verification when the data are important. Avoid both dismissing a pattern because staff are friendly and declaring fraud solely from a dashboard signal.

Signals deserving investigation

• Records show assessments at a time when the facility or instrument was unavailable.
• An entry was changed after randomisation to satisfy an eligibility threshold without supporting source.
• Signatures appear inconsistent with attribution or staff acknowledge using another person's account.
• A site has unexpectedly uniform endpoint values or very low safety reporting after relevant exposure.
• Product, laboratory or participant tracking records contradict a claimed visit.
• Original records repeatedly cannot be produced while clean summaries are offered instead.

Each signal points to a question, not an automatic diagnosis. Establish the relevant protocol requirement, source location, timestamps, responsible person and available corroborating evidence. Preserve the distinction between what you observed, what someone reported and what remains uncertain. An objective description such as an unsupported post-randomisation value change is more useful than a speculative accusation.

Immediate priorities

If participants may face an immediate risk, involve the investigator and sponsor's urgent clinical/quality contacts without waiting to finish a routine visit report. The investigator handles medical care; the sponsor determines wider trial measures and appropriate notifications. Preserve the relevant records and system evidence under authorised procedures, maintaining privacy and blinding. Do not remove original charts, seize devices or copy identifying information to a personal account.

Inform the designated clinical management and quality functions according to the escalation plan. A serious data-integrity concern can require independent QA or another specialised review. The monitor should not promise confidentiality that prevents required reporting, bargain to suppress a concern or warn people to delete embarrassing records before an audit.

Reconstructing a questionable visit

Suppose a source note claims a timed assessment on Saturday morning, but the imaging system shows no corresponding study and the site was apparently closed. Review the actual source locations, appointment record, instrument output, acquisition metadata and any authorised alternate facility. Ask neutral questions about the discrepancy. The assessment could have occurred elsewhere or the date could be a transcription error; neither should be assumed without evidence.

If a supported correction is needed, it should preserve the original entry, actual correction date, attribution and reason. If the assessment cannot be verified, document that limitation and escalate the impact for appropriate evaluation. Recreating a clean worksheet from memory does not establish that a critical endpoint was measured at the required time.

Evidence preservation and controlled access

Electronic records include relevant metadata, access logs, version history and audit trails. A screenshot may fail to preserve context or become an uncontrolled disclosure. Use the responsible system and quality functions to retain appropriate evidence. Record what was reviewed and how it can be retrieved. Do not disable a user account yourself unless that is within your actual authority; facilitate prompt containment through the designated administrator.

For paper records, maintain originals and traceable copies as appropriate. Never discard a page because a corrected version is easier to read. Late documentation should clearly distinguish the historical event from the current entry. A properly labelled late entry can be legitimate; backdating to create a false history is not a correction.

Sponsor and investigator response

E6(R3) expects significant noncompliance to prompt appropriate action, root-cause evaluation and remediation, with relevant notifications under applicable requirements. If significant noncompliance persists despite efforts to remediate it, the sponsor should consider terminating the site or service provider's participation and minimise the impact on participants and reliable results. A fixed five-step escalation ladder must not delay an urgent response.

A final decision should evaluate the affected records, participants, processes and possibly other studies or sites. Corrective actions may address supervision, access controls, staffing, tools or incentives, but retraining alone cannot establish that already questionable data are reliable. Follow-up should verify both implementation and effectiveness using objective evidence.

Applying the lesson to exam choices

Prefer an answer that documents facts, protects participants, preserves records and promptly involves the appropriate sponsor/investigator functions. Reject backdating, quiet deletion, automatic manufacture of missing data and unsupported accusations. The professional boundary is active oversight with evidence and escalation, supported by transparency even when the finding is uncomfortable or threatens a recruitment milestone.

Official sources

Checked October 18, 2026: ICH E6(R3), Principles and Annex 1, ICH E8(R1), ACRP CCRA Exam Content Outline.