The Enrollment Curve That Doesn't Match the Site's Capacity
You are the study lead for a Phase II rheumatoid arthritis trial targeting 180 subjects across 20 sites in 9 months. At month 4, you have 61 subjects enrolled. Site 03 (a large academic center) has enrolled 22 - far ahead of plan - but its screening failure rate is 58% versus a study average of 24%. Reviewing Site 03's screening log, you notice that 14 of its 22 enrolled subjects have baseline DAS28 scores just above the 3.2 inclusion threshold, clustered at 3.3-3.5. The site's sub-investigator is a rheumatologist who joined the site two months ago and is paid per enrolled subject through a site sub-contract. The PI at Site 03 has not signed the delegation log entry for this sub-investigator. Two other sites have flagged that Site 03 referred them patients who "didn't quite meet criteria."
1. What are you seeing, and what do you do before the next monitoring visit?
Read the worked answer
What you're seeing - three overlapping signals:
- Score clustering at the threshold. Fourteen subjects at 3.3-3.5 against a 3.2 cutoff is not natural distribution; it's a pattern consistent with either measurement bias or criteria-stretching. DAS28 involves a tender/swollen joint count and a patient global - all subject to assessor judgment.
- A financial incentive aligned with enrollment. Per-subject payment to a sub-investigator who is also the high-enrolling assessor is a conflict-of-interest structure that regulators scrutinize.
- A delegation gap. If the sub-investigator isn't on the delegation log, procedures he performed may not be attributable to a qualified, delegated person - a GCP finding independent of the enrollment question.
Before the next visit:
- Do not tip off the site. Instruct the CRA to conduct an unannounced or short-notice source review focused on the 14 clustered subjects: source worksheets, joint-count documentation, any re-scoring, and who performed each assessment.
- Pull the delegation log and training records for the sub-investigator - confirm dates and whether he was trained on the protocol and DAS28.
- Review the sub-contract to confirm the payment structure and whether it was disclosed to the IRB and in the consent/conflict-of-interest disclosures.
- Escalate to the Medical Monitor now, not after the visit - if scores were manipulated, subjects may have been enrolled without meeting criteria, which affects both safety and the integrity of the primary endpoint.
2. If the monitoring visit confirms the clustering, what happens to the data and the site?
Read the worked answer
Data:
- The 14 subjects are not automatically excluded - that is a statistical and medical decision, not an operations one. Convene the Medical Monitor and the study statistician to determine whether the assessments are unreliable (in which case a sensitivity analysis excluding Site 03 may be pre-specified) or whether the subjects genuinely met criteria and the pattern is coincidental.
- Do not silently drop them. Any exclusion must be documented, justified, and reflected in the analysis plan before unblinding. Dropping subjects after seeing the data is a worse integrity problem than the one you're investigating.
- Preserve all source documents - do not let the site "correct" worksheets.
Site:
- Suspend enrollment at Site 03 pending investigation. This is protective, not punitive, and should be communicated as such.
- Report to the sponsor and, if noncompliance is confirmed, to the IRB. If subjects were enrolled without meeting criteria, assess whether they were exposed to risk (e.g., an ineffective or unsafe drug) and whether any require notification.
- The sub-investigator's delegation status is a separate finding: if he performed study procedures undelegated, that is a reportable GCP deviation regardless of the enrollment question.
- Financial conflict: review whether the per-subject arrangement created an undue inducement and whether it was disclosed. This may require a site-level corrective action plan and, depending on your SOPs, referral to the compliance office.
Judgment point: The referral of "didn't quite meet criteria" patients to other sites is the most damaging detail - it suggests the site knew the patients were ineligible and tried to place them elsewhere. That elevates this from a data-quality concern to a potential fraud inquiry.