Create a Protocol-Deviation Log That Supports Analysis

Capture the event before judging its impact

Start with a stable deviation identifier, participant or site identifier where appropriate, occurrence date, discovery date, protocol version, and the specific requirement involved. Describe what happened in factual terms. Keep the original record and any later correction traceable.

A useful log separates the event from its assessment. “Visit occurred on day 38” is an observation; “outside the permitted window” requires checking the applicable window; “important for interpretation” requires a further assessment. Combining those statements into one free-text label makes later review difficult.

Use the applicable study process

Record who assesses the deviation, its potential effect on participant protection and data reliability, corrective action, and closure status. Route urgent issues through the study's established procedures. This practical log does not replace sponsor, ethics, institutional, or jurisdiction-specific obligations.

FDA's December 2024 protocol-deviation guidance page discusses consistent identification, classification, and reporting. At this source check it remains draft guidance, explicitly not for implementation; it should not be presented as a final binding reporting rule.

Work through a fictional timing issue

Imagine a protocol permitting a day-28 assessment within seven days either side. The stated window is days 21 through 35. A fictional assessment on day 38 falls three days after the upper boundary. Record the actual day and window before assigning the study's deviation category.

Then check whether the measurement is usable under the prespecified outcome and analysis rules. A visit-window deviation does not automatically mean the participant must be removed from every analysis. Keep the operational classification and analysis decision in separate fields, with the rationale for each.

Review patterns without rewriting history

Summarize deviations by type, site, time, and relevant study group using defined denominators. Distinguish multiple events affecting one participant from the number of participants affected. Repeated late visits may indicate a scheduling problem; a change in detection practice may also change the apparent rate.

Retain the initial assessment, subsequent revisions, reviewer, and dates. If a rule is clarified, document whether prior records require reassessment. Avoid changing classifications after inspecting favorable or unfavorable outcome results.

Before analysis, reconcile the log with data queries and the analysis-population specification. Produce an unresolved-items list and a decision trail for exclusions or sensitivity analyses. The useful output is an accountable record connecting source facts to study decisions, not a label that silently removes inconvenient data.

Source checked on 31 October 2026; fictional window calculation independently checked.