Clinical Trial Patient Recruitment: The Pre-Screening Call Nobody Staffs

Clinical trial patient recruitment is the work of finding, contacting, and qualifying people who might be eligible for a study, then getting the ones who are to a screening visit at a site.

Sponsors spend heavily on the finding. Advertising, patient registries, referral networks, and matching platforms all generate volume. What happens next is usually a coordinator with a phone, a protocol, and more referrals than hours.

That call is where most of the funnel disappears. This covers what the pre-screening call has to establish, why it goes unstaffed, and which parts of it can be automated without touching consent or clinical judgment.

TL;DR

• Recruitment spend concentrates on generating referrals. The bottleneck is usually the call that qualifies them.
• Pre-screening confirms interest, basic eligibility against protocol criteria, location and travel feasibility, and books the screening visit.
• Referrals go cold fast, and coordinators have clinical duties that outrank the phone.
• Pre-screening is not consent and not enrollment. Informed consent is a separate, regulated process that belongs to qualified site staff.
• An automated call can gather stated answers to protocol questions. It should not assess eligibility, give medical advice, or persuade.
• Before any call goes out, settle the authority to make it: contacting a prospective participant using protected health information needs authorization or an IRB waiver, and the recruitment materials themselves usually need IRB approval.

What clinical trial patient recruitment involves

Four stages, with very different cost structures.

Stage one is a media budget. Stage three is clinical capacity. Stage two is a phone call, and it is the one nobody funds, which is why it is the one that fails.

The economics make that failure expensive. Every unqualified referral that reaches a screening visit consumes clinical time, and every qualified referral that never gets called was paid for and wasted.

Why the pre-screening call goes unstaffed

Four reasons, none of which is about the quality of the coordinator.

• Coordinators have clinical duties. Study visits, source documentation, monitoring visits and queries all have deadlines. Returning calls does not, so it slips.
• Referrals arrive unevenly. A campaign or a media mention produces a week of volume in two days, and the queue forms immediately.
• Contact windows are wrong. People with jobs and conditions are reachable in the evening, which is precisely when the site is closed.
• Many referrals do not answer the first call. Whether they ever enter the funnel depends on the retry cadence, which is the first thing to lapse when the coordinator is in a visit. Retell does not cap how often an agent calls a number, so the redial cadence lives in your own list and workflow.

The result is a queue of referrals going stale. Interest in participating decays quickly, and a callback a week later reaches someone who has moved on or forgotten applying.

Multi-site studies compound it, because the same problem runs independently at every site with no visibility into which sites are actually calling their referrals.

What a pre-screening call should establish

Stated facts against protocol criteria, and nothing that requires clinical judgment.

1. Whether the person is still interested, and that they understand this is a research study rather than treatment.
2. Basic protocol criteria that a person can answer about themselves: age range, diagnosis, treatment history, current medications where the protocol asks, and common exclusions.
3. Location and travel feasibility against the site and the visit schedule, which is a frequent and under-asked disqualifier.
4. Availability for the visit schedule, including any overnight or long-visit requirements.
5. Who to contact, when, and on which number.
6. The next step, booked: a screening visit or a call with site staff.

Items 2 and 4 are where this call starts collecting health information over the phone, so the minimum-necessary test applies to the question list itself. Ask what the protocol actually needs at this stage, and nothing beyond it.

What it must not do is decide. A pre-screening call collects answers; a qualified investigator or delegated staff member determines eligibility. Those are different acts and the distinction is not cosmetic.

It also must not persuade. Recruitment communication that oversells benefit or downplays risk is a regulatory problem as well as an ethical one, and the script should read as neutral information rather than a sales call.

Where automated calls fit, and where they stop

They fit the contact problem: reaching every referral quickly, at the hours people answer, without taking a coordinator off a study visit.

Retell AI is an AI voice agent platform for contact centers. For recruitment that means an outbound call as soon as a referral arrives, batch calls to work a backlog rather than the most recent arrivals, an inbound line that answers people calling about an advertisement, and booking the next step on the call when your scheduling runs on Cal.com or Calendly. The clinical trial prescreening page shows the workflow as a product: calls to people who asked about a study, protocol questions one at a time, and a warm handoff to your study team.

Four honest boundaries, and they matter more here than in most industries.

• No informed consent. Consent is a regulated process conducted by qualified site staff, and nothing about an automated call should resemble it or precede it in a way that implies enrollment.
• No eligibility determination. The agent records answers. A delegated clinician decides. Build the workflow so the determination step is visibly separate.
• No medical advice, ever. Questions about a person's condition, their medication, or whether they should join go to site staff. This is the transfer rule to test hardest.
• No persuasion. If someone is hesitant, the agent's job is to offer a conversation with the site, not to overcome an objection.

The transfer path therefore carries more weight than the automation rate. A caller who asks a clinical question should reach a person with the context already captured, not be told to call back.

Can a clinical trial call center be automated?

Part of it can: the contact work. A clinical trial call center answers referral calls and phones sign-ups to run the pre-screening script, and an AI voice agent can take over those calls while people keep the parts that need judgment.

The split follows the diagram above. An agent can place the first call to each referral, answer the study line, ask the approved questions in order, record what the person says, and pass the call to site staff on request. It cannot decide eligibility, obtain consent or answer medical questions, so a call center that does those today still needs trained people for them. Where the study line has more calls than staff, call transfer sends the callers who need a person to the right team with the answers already captured.

If you are comparing options, AI prescreening calls for clinical trial recruitment describes how one study is set up, and the pilot advice there is the same as here: start with one study and read every transcript.

Compliance: HIPAA, IRB, and the recruitment materials

Three separate approvals, and teams routinely secure one and forget the others.

The first HIPAA question is not about the vendor. It is whether you have authority to make the call at all. Using protected health information to identify and contact a prospective participant requires either the individual's authorization or a waiver, or partial waiver, of authorization granted by an IRB or Privacy Board under 45 CFR 164.512(i). A review preparatory to research lets you identify potentially eligible people from records; on its own it does not let you telephone them. Settle which basis you are relying on before any call is placed, and keep the waiver documentation with the protocol.

For the payer side of healthcare calls, prior authorization automation covers where voice fits in that call loop.

The vendor questions come second and are the familiar ones: a business associate agreement where applicable, access controls on recordings and transcripts, a retention policy, and a defensible minimum-necessary position on what the call captures at all. No vendor claim substitutes for your own assessment. Retell offers a self-service business associate agreement before any PHI is transmitted, and its data storage settings run from full retention down to basic attributes only, with PII scrubbing available on top, so the minimum-necessary question has a configuration to point at rather than a promise.

FDA's guidance says advertising for study subjects should be reviewed and approved by the IRB, and that the IRB should look at first-contact scripts. An automated call script belongs in that package. It is not a piece of marketing copy that can be edited on a Tuesday afternoon, which changes how change control has to work.

FDA's guidance on recruiting study subjects has a section on receptionist scripts, which covers the first contact where a script is used to determine basic eligibility. It lists the questions an IRB may ask about that first call. They map to settings you control, so answer them before the first call goes out.

The right-hand column is where a setting exists, not an IRB determination. The IRB decides whether your procedures are adequate.

That has a practical consequence worth planning for. Your script iteration loop runs at the speed of your IRB, so build the script to be right rather than to be revised, and keep a clear record of which version was in use on which date.

Two more surfaces apply on top. Outbound calling to a referral's phone sits under the TCPA and state calling rules. The FCC has ruled that the TCPA's limits on artificial or prerecorded voice cover AI-generated voices, so calls like these generally need the prior express consent of the person called. Confirm what consent the referral route actually captured and how a revocation is honoured. Retell has do-not-contact handling that marks a contact so later calls skip the number, but batch calls do not check it, so check your list before you upload it. Separately, state call recording consent rules vary, and a multi-site study means multiple states at once.

None of this is legal or regulatory advice; it is the list of people who need to be in the room before the first call.

What to measure

Recruitment reporting usually stops at referrals generated, which is the least useful number in the chain.

• Time from referral to first contact attempt. In hours. This is the number that predicts whether the referral converts.
• Contact rate. The share of referrals ever reached, and how many attempts it took. Usually the largest single leak.
• Pre-screen completion rate. Of those reached, how many completed the question set.
• Pre-screen to screening visit booked, and to attended. The gap between booked and attended is a reminder problem and is fixable.
• Screen failure rate. If it is high, the pre-screening questions are not catching what they should, which wastes clinical time.
• Cost per randomized participant. The only figure the sponsor is actually buying.

For multi-site studies, report contact rate by site. It usually varies more than anything else in the chain and it is invisible in referral-level reporting. Post-call analysis gives the per-call record that makes that comparison possible.

Frequently asked questions

What is clinical trial patient recruitment?

The process of identifying, contacting and qualifying potential participants for a study, and getting eligible people to a screening visit. It spans outreach, pre-screening, screening and enrollment, and it is one of the most common causes of study delay.

What is pre-screening in a clinical trial?

An initial check, usually by phone or online form, of whether someone plausibly meets protocol criteria before clinical time is spent on a screening visit. It gathers stated answers; eligibility is determined by qualified site staff.

Can AI make clinical trial recruitment calls?

It can handle contact and structured pre-screening questions, book the next step, and answer logistical questions from approved wording (see AI prescreening calls for clinical trial recruitment). It should not obtain informed consent, determine eligibility, give medical advice, or persuade a hesitant person, and, following FDA's recruiting guidance, the call script belongs in the package your IRB reviews.

Is automated pre-screening HIPAA compliant?

Compliance depends on the deployment, and the first question is authority rather than tooling. Contacting a prospective participant using protected health information needs the individual's authorization or an IRB or Privacy Board waiver under 45 CFR 164.512(i). After that comes the vendor layer: the appropriate agreements, access controls, retention limits, and a minimum-necessary position on what is collected. Ask any vendor for current attestations and assess them against your own requirements.

Why do clinical trials fail to recruit?

Frequently not for lack of interested people. Referrals go uncontacted or are contacted too late, eligibility criteria are narrow, travel and visit schedules are impractical for the participant, and site capacity is consumed by screen failures that better pre-screening would have caught.

What do IRBs ask about pre-screening call scripts?

FDA's recruiting guidance says the IRB should be assured that first-contact procedures protect prospective subjects. Its examples include what happens to personal information if the caller hangs up, who gathers the data, whether names of non-eligible people are kept, and how records are disposed of. Have written answers for each before the first call.

Does the pre-screening script need IRB approval?

FDA expects IRBs to review advertising used to recruit study subjects, and its guidance also covers the scripts used for first contact. Treat an automated call script the same way, and plan change control around your IRB's timelines rather than around your content workflow.

Every referral called the day it arrives, on an IRB-approved script.

Build the agent against your IRB-approved script, run it on one site's referral queue, and read every transcript before you take it anywhere else. Consent, eligibility and anything clinical stay with site staff. See how the clinical trial prescreening workflow is set up, or run a pilot on your own calls.