How sponsors choose sites, and what they check

How does a sponsor build its site list?

Selection usually starts before a site hears about the study. The sponsor, or the CRO running feasibility, pulls candidates from its own performance records, investigator databases, its site network, and sites that ran similar protocols. Regulation sets the floor: for drug studies under an IND, 21 CFR 312.53(a) says a sponsor must "select only investigators qualified by training and experience as appropriate experts to investigate the drug."^⁠[12] Above that floor, the choice is a judgment about who will enroll.

1. 1

   Long list from data

   The sponsor or CRO screens its database and prior-study records for indication experience, enrollment history and investigator status. Sites that under-delivered recently are often dropped here.

2. 2

   Confidentiality agreement and synopsis

   Shortlisted sites sign a confidentiality agreement and receive a protocol synopsis.

3. 3

   Feasibility questionnaire

   The site answers questions on its patients, competing studies, staff, equipment, start-up timelines and expected enrollment. Sites are typically not paid for this work.^⁠[8]

4. 4

   Pre-study or qualification visit

   A CRA visits, in person or remotely, to confirm facilities, investigator oversight, pharmacy and lab capability, and the recruitment plan.

5. 5

   Selection and start-up

   Selected sites move into budget, contract and IRB work. A 2025 paper notes that industry still describes more than two months to select sites, then eight more months to activate them.^⁠[5] Before an investigator starts, 21 CFR 312.53(c) requires the sponsor to collect a signed Form FDA 1572 and the investigator's CV or equivalent.^⁠[12]

Our feasibility questionnaire template lists the questions most sponsors ask. The glossary defines site feasibility and site activation.

Why are sponsors wary of a site's enrollment estimate?

Because the record says estimates are unreliable. In a 2013 Tufts CSDD analysis of more than 150 studies and nearly 16,000 sites, 11% of sites in a typical trial enrolled no one, 37% under-enrolled, 39% met their target and 13% exceeded it. Most trials (89%) still reached their enrollment goals, but usually by nearly doubling the original timeline.^⁠[1]

Weak sites are expensive. Phesi estimated that a cancer-trial site enrolling a single patient costs about $130,000 per patient, almost ten times the $14,167 at a better-performing site.^⁠[3] In the STICH heart-failure trial, 44 sites that had been approved never enrolled a patient, at an estimated $10,000 per deactivated site.^⁠[7]

Activation is a separate risk. At the 2024 SCOPE Summit, Ken Getz of Tufts CSDD said that about 40% of North American sites a sponsor looks to engage will not activate: some never get through contracting and budgeting, and some never get IRB approval.^⁠[2]

Which site metrics do sponsors weigh most?

In WCG's 2024 survey of 852 sites, 69% named budgets and contracts among the largest contributors to start-up time, more than any other factor. The same report cites a 2024 benchmarking study of North American cancer centers in which only 9% met a 90-day activation timeline for industry-sponsored studies, against nearly 60% for NCI-sponsored studies.^⁠[4]

Why does "we have many patients" not win the study?

Most sites say they have the population. A sponsor cannot tell a site that has checked from one that is guessing, so it discounts both. A diagnosis count from a billing query skips the criteria that do most of the excluding: lab thresholds, prior therapies, washout windows, comorbidities, and whether the patient still gets care at the site.

Track record fills the gap. A 2018 review of why trials fail found that sites with a history of meeting enrollment targets are more likely to meet them again, and that a site that has run 6 to 10 trials has a greater probability of enrolling on time than one with fewer.^⁠[7] A site with less history needs stronger evidence, usually a patient count with its method attached.

How should a site present EHR-derived patient counts?

Present a count like a lab result: with the method, the date and the limits, so the sponsor can judge how much it will shrink at screening.

1. 1Start from the protocol, not the diagnosis. List the inclusion and exclusion criteria and mark which the EHR can answer from structured fields, which need notes or reports, and which cannot be checked in the chart at all.
2. 2Show the funnel. Report the count after each group of criteria: diagnosis, age and sex, a recent visit at your site, key labs, prior or current therapy, major exclusions. The drop between steps shows the sponsor you applied the criteria.
3. 3State the query date and lookback window. Say when the query ran and how recently a patient must have been seen to count as active.
4. 4Separate computed from reviewed. If coordinators checked a sample of charts from the funnel, say how many and what share held up. That ratio is often the most persuasive number you can offer.
5. 5Break out demographics. Give the final count by race, ethnicity, sex and age group where the data allows. Sponsors writing diversity plans will ask (see the next section).
6. 6Convert to an enrollment estimate last. Apply your own historical screen failure and consent rates, and show that step, rather than presenting the eligible pool as your commitment.

Keep counts aggregate and free of identifiers. The chart review behind them usually falls under the HIPAA provision for reviews preparatory to research, 45 CFR 164.512(i)(1)(ii), which bars removing protected health information from the covered entity during the review.^⁠[13] Follow your privacy office's process.

Identify, Bond Health's EHR screening step, reads structured and unstructured records against a study's inclusion and exclusion criteria, ranks candidates and shows the chart evidence behind each criterion decision. A coordinator can use that evidence to check a sample before reporting a count. Full EHR integration typically takes 48 hours, depending on the EHR and IT review.^⁠[14]

How do diversity requirements affect site selection?

The Food and Drug Omnibus Reform Act of 2022 (FDORA) amended the FD&C Act to require a Diversity Action Plan for Phase 3 and other pivotal drug studies and for certain device studies, under sections 505(z) and 520(g)(9). The plan must set enrollment goals disaggregated by the race, ethnicity, sex and age group of the clinically relevant population.^⁠[10]

The requirement applies to certain studies that begin enrollment more than 180 days after FDA publishes final guidance.^⁠[10] As of September 2026, FDA's guidance page still lists the June 2024 document as a draft and notes that it was restored under a court order, so the 180-day clock has not started.^⁠[11] FDA received voluntary plans for some studies in fiscal years 2023 and 2024, and a sponsor that writes one needs sites that can help meet its goals.^⁠[10]

For a site, demographic data becomes part of feasibility. Be ready to report your population by race, ethnicity, sex and age group, the languages your staff can consent in, and your community outreach. Say where race and ethnicity fields are incomplete. The Diversity Action Plan entry covers the plan itself.

How should a site track the metrics sponsors check?

Keep a one-page performance summary and update it every quarter. Most of it comes from your CTMS and the EDC reports sponsors already send you.

• Randomized patients against commitment for every study that closed in the past three years.
• Median days from contract receipt to activation, split into budget, contract and IRB.
• Median days from activation to first patient screened and first patient randomized.
• Screen failure rate by study, with the top reasons.
• Open queries and median days to resolve them.
• Protocol deviations per enrolled subject, major and minor.
• Active studies per coordinator and current open positions.

Report weak numbers with a reason next to them. A sponsor trusts a site that explains a slow study more than one that lists only its best. The guide to winning more studies covers the relationship side.

What should I do this week?

• Pull enrollment against commitment for your last five closed studies.
• Take one open feasibility request and rebuild the patient count as a funnel, with the query date and the criteria you could not check.
• Ask a coordinator to review a small sample of charts from the bottom of the funnel and record how many hold up.
• Time your last budget and contract cycle from receipt to signature, and find the step that took longest.
• Check how complete the race, ethnicity and preferred language fields are in your EHR.
• Save the answers in the feasibility questionnaire template so the next response takes less time.