Recruiting for Protocols With a Keyhole Eligibility Window

Some protocols are hard because the disease is rare. Others are hard because the sponsor stacked twelve exclusions onto a common condition and turned a broad population into a keyhole. Either way, the recruitment approach that works for a straightforward Phase III hypertension study will bury you here. Volume-based outreach on a narrow protocol just generates screen failures and coordinator burnout.

Find the one criterion that does the filtering

Every narrow protocol has a dominant limiter — the single criterion that eliminates most of the population. Sometimes it is a lab value in a tight range. Sometimes it is prior therapy exposure, or a washout that rules out anyone currently stable on treatment. Sometimes it is something mundane like a BMI ceiling or a contraception requirement.

Before you build any outreach, read the I/E list and rank the criteria by how many people each one removes. You do not need precision; you need order of magnitude. Then design your entire funnel around the top one or two. If prior biologic exposure is the limiter, your search should start with pharmacy and infusion records, not with a call campaign. If it is a narrow lab window, you need recent labs in hand before anyone gets a phone call.

The common failure is treating all thirty criteria as equally important during pre-screening. You end up with a fifteen-minute call that annoys the patient and still misses the thing that disqualifies them.

Go to records before you go to advertising

For tight protocols, chart mining and EMR queries almost always outperform paid media on cost per enrolled subject. Ads select for people who are interested. Records select for people who are eligible. On a narrow study, eligibility is the scarce resource.

Practical sequence that works in most site datasets:

1. Query your own database first with the dominant limiter as the primary filter, not the diagnosis code.
2. Cross-check against your active studies so you are not pulling patients already committed elsewhere.
3. Pull referring-physician lists for the specific sub-population, and ask for the specialty clinic that manages the limiter — the rheumatology infusion suite, the sleep lab, the endocrine practice.
4. Only then layer outreach to fill gaps, targeted to the sub-population rather than the broad indication.

Pre-screen deeper, contact less

With narrow criteria, the correct move is to spend more time per candidate and touch fewer candidates. Build a short-form pre-screen that resolves the dominant limiter in the first two questions, then a second-stage review that a coordinator or clinician does with chart access before scheduling.

Schedule fewer screening visits, but expect a much higher conversion rate from visit to randomization. That is the whole point. A site that runs eight screening visits and enrolls six has a healthier operation than one that runs forty and enrolls seven, even at the same enrollment number — less coordinator time, less patient goodwill spent, fewer screen-fail costs the sponsor may not fully reimburse.

Manage the sponsor conversation early

If your feasibility numbers assumed a broader population, say so before the site initiation visit, not at month four. Ask specifically:

• Which criteria has the sponsor amended at other studies in this program?
• Is there flexibility on window timing, washout length, or lab retest?
• Will they reimburse screen failures, and at what stage?

Sponsors running narrow protocols usually know enrollment will be slow. What damages your standing is silence followed by zeros. A site that reports "we screened four, enrolled three, and here is the criterion killing us" is a site sponsors come back to.

Protect your team's expectations

Set a realistic cadence internally. One enrollment a month on a keyhole protocol is a win, and coordinators need to hear that framed as success rather than shortfall. Track pre-screen-to-screen and screen-to-enroll ratios separately so the team can see the funnel working even when the raw count is small.