Recruitment That Does Not Distort the Trial
Part of Principal Investigator (PI) Training: Recruitment, Enrollment and Retention.
About 4 minutes. Free, in your language, no account needed.
Why this matters
A trial answers a question about a population. Which people you enroll decides which population it actually answers for — and a site can quietly change that without breaking a single rule, simply by who it finds easiest to approach.
This is the part of recruitment that is scientific rather than ethical, and physicians are unusually well placed to see it going wrong, because you know your own clinic.
The estimate, revisited
Module 1 covered §2.2.1: demonstrate a potential for recruiting the required number within the agreed period, using retrospective or currently available data. Here is what to do when the estimate turns out to be wrong, which it usually does.
Tell the sponsor early and with numbers. "We are screening four a month, not twelve; of the last thirty clinic patients, nineteen failed on criterion 6." That sentence is useful. "Recruitment is slow" is not, and it invites pressure rather than a solution.
Never close the gap by loosening. The four common ways sites do it are all deviations: stretching a criterion, screening people you know will fail to inflate the funnel, re-screening outside what the protocol permits, and enrolling on unconfirmed history. Each is covered by §2.5.3, and each is worse than a slow site.
Screen failures are information, not waste
Sites treat screen failures as a cost. They are the most useful recruitment data you have, and they are free.
Record, for each one, the single criterion that stopped them. After twenty you will know which of these you have:
- One criterion doing all the work. Take it to the sponsor. If it blocks suitable people at every site, it is an amendment conversation and you have the evidence.
- Failures spread evenly. The population estimate was wrong. Adjust the commitment.
- Failures at the assessment stage rather than the criterion stage. An operational problem — a turnaround time, a scheduling window — that is yours to fix.
Who you are missing
Ask periodically who isn't in your trial who should be. The usual absentees are consistent across studies, and each has a fixable cause:
| Missing group | Usual cause | Usually fixable by |
|---|---|---|
| People who work | Weekday-only visit slots | An early or late clinic |
| People far from the site | Travel cost and time | Reimbursement, or remote visits the protocol allows |
| People with limited literacy in your language | Materials and discussion in one language | Approved translations, an interpreter |
| Older people with comorbidity | Cautious criteria, cautious clinicians | Checking whether the criterion actually excludes them |
| Carers | Visit burden | Scheduling around them |
None of that is a regulatory requirement. It is the difference between a trial result that applies to your patients and one that applies to the subset who found it convenient.
Feasibility for the next study
The recruitment record you build on this trial is what makes the next feasibility honest. Keep four numbers per study: how many were approached, how many consented, how many enrolled, how many completed. Those four ratios are worth more than any impression, and a sponsor asking about your site's capability is asking for exactly them.
A site that can say "we approach forty to enroll twelve and eleven complete" is a site that gets chosen — and one that will never again over-promise, because the arithmetic is on the page.
Key takeaways
- Who you enroll decides who the answer applies to; a site can distort a trial without breaking a rule.
- Report shortfalls early, with numbers. Never close a recruitment gap by loosening criteria — every common method is a §2.5.3 deviation.
- Log the blocking criterion for every screen failure; the pattern tells you whether it is the protocol, the population or your own operations.
- Ask who is missing and why, and keep approached / consented / enrolled / completed for the next feasibility.