26.2 Good Documentation, Completeness and Corrections

Good Documentation, Completeness and Corrections

Record source data contemporaneously with the assessment, intervention or observation. Delaying data entry introduces recall bias, memory lapses, and factual distortion.

• Late Entries Protocol: When clinical realities prevent instantaneous charting (e.g., emergency resuscitation or high-volume surgical procedures), any subsequent documentation must be explicitly labeled as a "Late Entry." The author must record: (1) the current calendar date and time the note is written, (2) the known historical event date/time, with uncertainty identified rather than invented precision, and (3) their signature/initials. Back-dating a record (writing yesterday's date to make an entry appear timely) is a severe regulatory non-compliance.

Original (O)

Source data must be captured in the primary recording medium first touched by the observer. Transcribing values from loose scraps of paper, sticky notes, or temporary whiteboards into the official chart—and then discarding the scrap—violates the "Original" requirement. The loose paper was the true original.

• Certified True Copies (ICH E6(R3)): When original paper records must be digitized or copied (e.g., off-site scanning or archival conversion), the resulting copy must qualify as a Certified True Copy. Under ICH E6(R3), a certified copy is an exact copy (photocopy, scan, microfiche) that has been verified—via a dated signature or a validated electronic verification process—to have the exact same information, including data describing the context, content, and structure of the original.
• Certification Mechanics: The individual verifying the copy may document certification with wording such as: "Certified True Copy of the original record," followed by their signature, printed name, and date of verification. Partial scans missing double-sided pages, cropped margins, or unreadable lab attachments fail true copy standards.

Accurate (A)

Source data must represent a truthful, uncorrupted, and mathematically correct representation of clinical facts. Accuracy implies that data are free from transcription errors, represent plausible physiological ranges, and are supported by diagnostic evidence.

• Medical Plausibility: A recorded pulse rate of 18 bpm in an alert outpatient or a markedly abnormal laboratory result without documented clinical review calls data accuracy into question and requires prompt investigation.
• Protocol Adherence: The CRA verifies that numerical scores on validated psychometric instruments (such as HAM-D, ADAS-Cog, or RECIST tumor measurements) were calculated accurately using protocol-specified scoring algorithms.

Complete (+)

Source records must account for every assessment, visit, laboratory panel, and diagnostic test mandated by the clinical study protocol.

• Documenting Omissions: In clinical research, a blank data field is ambiguous—it could indicate that the assessment was omitted, that the test was negative, or that the result was lost. If a protocol-mandated procedure was not completed (e.g., subject refused blood draw, ECG machine broken), the source document must explicitly state: "Test not done: [clinical reason]," attributed to the appropriate responsible person.
• All Pages Accounted For: Multi-page survey instruments, multi-lead ECGs, and multi-sheet lab summaries must be complete without missing intermediary pages.

Consistent (+)

Clinical data across multiple distinct source documents within the same investigative center must demonstrate mutual concordance without contradiction.

• Cross-Source Triangulation: If the physician progress note states that the patient experienced "mild nausea lasting two hours post-infusion," but the nursing flowsheet records "severe vomiting and IV ondansetron administration," an internal inconsistency exists that must be queried and reconciled.
• Chronological Coherence: Medication stop and event resolution dates can differ legitimately; query an apparent inconsistency rather than assuming the dates must match.

Enduring (+)

Data must be recorded upon physical or digital media engineered to survive without degradation throughout statutory retention periods (as determined by applicable requirements and the sponsor’s retention needs).

• Thermal Paper Degradation: A classic CRA monitoring finding involves thermal paper printouts from automated ECG machines, laboratory balance scales, or pharmacy temperature dataloggers. Thermal output may fade; evaluate the medium and preserve a durable readable record.
• Mandatory CRA Action: The CRA must instruct site staff to photocopy or scan vulnerable records in a timely way under a suitable verified process, certify the copy as a Certified True Copy, and retain records under the approved copying and retention process in the subject source binder.

Available (+)

Clinical records must be promptly retrievable and accessible for review by authorized personnel—including CRAs, sponsor quality assurance auditors, ethics committee representatives, and regulatory inspectors—throughout the active trial and the statutory post-trial archive period.

• Physical & Digital Readiness: Secure or off-site storage can be appropriate if authorised retrieval and readability are maintained. A missing retrieval route or deactivated access that prevents necessary review is a concern to resolve.

Official sources

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