27.2 Audit Trails, Electronic Outcomes and Device Choice

Audit Trails, Electronic Outcomes and Device Choice

Relevant metadata explain data context: who or what generated a record, dates/times, workflow, changes and transfers. Plan review and retention based on risk and criticality. A timestamp should be unambiguous, for example using UTC or a clear time-zone convention; UTC is not the only possible implementation.

Systems should retain initial entries and subsequent changes/deletions and capture reasons where appropriate. Audit trails and logs should not be disabled. E6(R3) allows rare justified modification, such as removal of inadvertently entered personal information, only when the action and justification are logged. Therefore, an absolute claim that no administrator can ever modify any audit record is inaccurate. Controlled traceability and protection against inappropriate alteration are the underlying requirements.

Review higher-criticality data and metadata according to the plan. An unusual entry time can reflect a legitimate shift or time-zone difference, not necessarily an unauthorised foreign contractor. Investigate patterns in context and preserve evidence before drawing conclusions.

Anatomy of an EDC Audit Trail Record

Every individual entry in an EDC audit trail contains five traceable data elements:

CRA Audit Trail Review Strategies (E6(R3))

E6(R3) calls for planned risk-based review of relevant data and metadata. The actual monitoring responsibilities and scope should be documented in the trial’s approach. CRAs examine audit trail logs for specific data integrity red flags:

• Late Eligibility Adjustments: Alterations to inclusion or exclusion criteria (e.g., baseline lab values or diagnostic scores) executed days after randomization, suggesting that a non-eligible participant was retroactively made eligible.
• Post-Hoc Safety Modifications: Downgrading an Adverse Event from "Severe" to "Mild" or altering investigator causality from "Related" to "Unrelated" shortly after sponsor safety communications.
• Bulk Data Alterations: Large tranches of data entered or modified minutes before a scheduled monitoring visit or regulatory inspection.
• Unusual Working Hours: Extensive data entry or signature execution occurring at 03:00 AM on weekends, warranting contextual review of work patterns, time zones and authorised access.

Direct Electronic Source (eSource), ePRO, eCOA & Wearables

The traditional paradigm where clinical data are recorded on paper medical records and subsequently transcribed into an EDC system introduces human transcription error, operational lag, and duplicate documentation burdens. Direct electronic capture can reduce transcription while still requiring suitable controls over source, metadata, access and transfer.

ePRO and eCOA Modalities

• Electronic Clinical Outcome Assessment (eCOA): An umbrella term encompassing clinical assessments captured electronically, including clinician-reported outcomes (ClinRO), observer-reported outcomes (ObsRO), performance outcomes (PerfO), and patient-reported outcomes (ePRO).
• Electronic Patient-Reported Outcomes (ePRO): Questionnaires, visual analog scales (VAS), pain scores, and medication adherence logs completed directly by the trial participant using a digital tablet or smartphone.
• Elimination of "Parking Lot Syndrome": A notorious deficiency of paper subject diaries is the "parking lot phenomenon," wherein participants neglect daily diary logging for an entire month, sit in the clinic parking lot ten minutes prior to their appointment, and rapidly fabricate thirty days of retrospective symptom entries. Validated ePRO devices eliminate this hazard by utilizing tamper-evident automated timestamps and programmed completion windows (e.g., daily morning symptom questionnaires unlock at 07:00 AM and lock permanently at 10:00 AM, controlling retrospective entry according to the designed process).

Wearable Sensors and Digital Health Technologies (DHTs)

Modern trials increasingly incorporate continuous biosensors, such as continuous glucose monitors (CGMs), wearable electrocardiogram patches, and digital actigraphy watches measuring sleep-wake patterns. These devices stream millions of data points directly to cloud analytics platforms.

• Data Management Challenges: Ingesting high-frequency continuous data streams requires specialized data pipelines, algorithmic noise filtering, and predefined statistical triggers to alert clinical teams to safety anomalies.

Device choice and accessible participation

A personal-device approach and sponsor-provided hardware have different compatibility, privacy, accessibility, support and supply implications. Neither guarantees higher adherence or perfect security. Confirm the intended users can access the tool, the instrument remains suitable across supported configurations, staff know how to report failures, and continuity arrangements preserve reliable data. A device's convenience should not exclude otherwise appropriate participants without evaluation.

A time-locked questionnaire can reduce retrospective back-filling but cannot prove who answered, whether a participant understood the questions or whether the device clock was correct. Distinguish observation time, entry time and transmission time. Direct capture still needs metadata and appropriate review, not an assumption that every digital value is automatically authentic.

Official sources

Checked October 18, 2026: ICH E6(R3), Principles and Annex 1, ICH E9(R1).