The Unreported Deviation and the Site's Pattern
A CRC at a multi-site oncology trial reviews a monitoring report and finds that across the last 14 enrolled participants, 6 had at least one cycle of study drug administered outside the protocol-specified window (cycle length 21 days ± 2; actual intervals ranged 17–30 days). The deviations were never logged in the site's deviation tracker, and none were reported to the sponsor or IRB. The principal investigator says: 'These are minor - nobody was harmed, and the drug's half-life makes a few days irrelevant. Logging them will just trigger a site audit right before our next enrollment push.' The CRC also learns that two of the six participants had dose delays because the pharmacy was closed for a holiday - a scheduling cause, not a participant cause. The protocol's statistical analysis plan defines a per-protocol population that excludes participants with major deviations.
1. Assess whether these deviations are 'minor' and whether the PI's reasoning is defensible.
Read the worked answer
The PI's conclusion may turn out correct, but the reasoning is not defensible as stated, and 'minor' cannot be asserted unilaterally by the site.
Why the reasoning fails:
- 'Nobody was harmed' is not the criterion. Deviation classification depends on the protocol's definitions and on whether the deviation affects participant safety, rights, welfare, or data integrity. A deviation can be harm-free and still be major if it compromises the interpretability of the data.
- The PI's pharmacology argument is an assertion, not a finding. 'Half-life makes a few days irrelevant' may be true, but it is a scientific judgment the sponsor and IRB - not the site alone - are positioned to weigh, especially where the SAP defines a per-protocol exclusion.
- Data integrity is squarely implicated. Deviations that could exclude participants from the per-protocol population affect the analysis. Concealing them from the sponsor distorts the sponsor's dataset.
- 'Logging will trigger an audit' is not a valid reason. That is the opposite of the rationale - the tracker exists to surface and correct problems. Suppressing records to avoid scrutiny is a serious integrity issue independent of the deviations themselves.
Classification: Each event needs individual assessment (magnitude, cause, whether it affected safety or the primary endpoint). Some may be minor; the 30-day interval and the holiday-caused delays warrant closer review. The site should propose a classification and let the sponsor/IRB confirm.
2. What are the CRC's obligations given the PI's instruction not to log the deviations?
Read the worked answer
The CRC has obligations that do not dissolve because the PI disagrees:
- Document the deviations in the source and the tracker. Accurate records are a core duty. The CRC should log them with dates, magnitudes, and causes.
- Escalate within the institution. If the PI blocks logging, the CRC should raise it through the site's research compliance or IRB channels - most institutions have a mechanism short of a formal whistleblower report. The CRC should not simply comply with an instruction to falsify or omit records.
- Notify the sponsor per the protocol's deviation-reporting requirements. Sponsors generally require prompt reporting of deviations affecting safety or data integrity; the site does not get to decide the sponsor doesn't need to know.
- Do not alter or backdate records to make the intervals appear compliant.
- Protect themselves with contemporaneous documentation of what was found, what was raised, and what was decided.
The CRC is not required to resolve the disagreement - only to ensure the issue reaches people empowered to resolve it and that the record is truthful in the meantime.
3. The two holiday-related delays were caused by site scheduling. How does cause affect the analysis and the corrective action?
Read the worked answer
Cause matters for two reasons: classification and prevention.
Classification: A deviation caused by site logistics is a site/system failure, not a participant behavior. That distinction affects how it is characterized and reported, and it points to a fixable process rather than an unavoidable event. It also means the site cannot attribute the pattern to 'difficult patients' - the pattern is internal.
Prevention: Holiday pharmacy closures are predictable. A corrective action could include:
- A dispensing calendar that flags closures and pre-supplies or reschedules visits.
- A visit-window scheduling buffer that builds the ±2 days in before the holiday, not after.
- A backup dispensing pathway.
- A pre-visit check that confirms drug availability before the participant travels.
Analysis: Because the cause is systematic rather than random, the pattern of 6/14 is more informative than isolated events - it suggests the site's scheduling process, not chance, is producing deviations. That strengthens the case for a process fix and for reporting the pattern to the sponsor rather than treating each event as an isolated 'minor' blip. The CRC should present the cause breakdown (4 participant-related, 2 site-scheduling) so the corrective action targets the right thing.