Screen Failure in Clinical Trials: How to Calculate the Rate, Document It Under GCP, and Read It as a Critical-to-Quality Signal

At a glance

• A screen failure is a person who was screened against the protocol’s inclusion and exclusion criteria and did not qualify to enroll. There is no single FDA or ICH definition of the term, so your protocol has to draw the line.
• Treat screen failure as a documentation-and-consent discipline, not just a recruitment statistic. Every screened subject, consenting and not, belongs in the screening log with eligibility evidence.
• Under GCP, no trial-related procedure may happen before informed consent is signed and dated. A consent that post-dates a screening procedure is a finding, not a paperwork nit.
• The screen-failure rate is a quality signal. A high rate usually says your inclusion and exclusion criteria are mis-calibrated, which ICH E8 treats as a critical-to-quality factor to fix at the protocol level.
• Re-screening is a discrete consent and source-data event with its own record, not a quiet retry on the same subject number.

A high screen-failure rate gets framed as a recruitment cost to “reduce.” That framing misses the two things an inspector actually cares about: whether every screened subject is documented, and whether consent preceded every procedure. This guide defines the term honestly, gives you a rate formula you can apply consistently, and maps the GCP mechanics that make a screening log defensible. Sibling articles on informed-consent timing, source data and ALCOA, and protocol eligibility criteria go deeper on the pieces named here.

What a screen failure is, and the consent line most teams blur

A screen failure is a candidate who entered screening and did not meet the protocol’s eligibility criteria, so they were never enrolled or randomized. The study population is defined through its inclusion and exclusion criteria, and ICH E8(R1) §5.1 states that the population to be studied is defined through the inclusion and exclusion criteria for the study, and that the degree to which a study succeeds in enrolling the desired population will impact its ability to meet its objectives. A screen failure is simply the boundary of that population made visible: the protocol said no.

The line teams blur is not the eligibility line. It is the consent line. People assume “screening” is a low-stakes pre-step you can run before consent. It is not. ICH E6(R3) §2.8.7 requires that, prior to trial participation, the informed consent form be signed and dated by the participant (or their legally acceptable representative) and by the investigator or delegated site staff who conducted the consent discussion. The principle is stated even more plainly in the GCP principles: ICH E6(R3) §1.4 (Principle 4) requires that freely given informed consent be obtained and documented from every participant prior to clinical trial participation. So a subject can be a perfectly legitimate screen failure and still represent a serious GCP problem if a screening blood draw, imaging scan, or washout happened before that signature.

Is there a standardized FDA or ICH definition? The honest answer

No. Neither ICH E6, ICH E8, nor 21 CFR Part 312 defines “screen failure” as a regulated term with a fixed boundary. That is why recruitment vendors and biostatistics blogs each use their own definition, and why “the rate” is hard to compare across studies.

The absence of a definition is not a gap you ignore. It is one you close in your own protocol. Anchor the term to the regulations that do exist: the eligibility boundary comes from your inclusion and exclusion criteria, the documentation comes from the GCP recordkeeping requirements, and the consent timing comes from the consent rules above. ICH E8(R1) Section 7 lists eligibility criteria among the design aspects to get right, noting that the eligibility criteria should be reflective of the study objectives and be well documented in the clinical study protocol. If your protocol defines, in writing, who counts as screened, when the screening window opens, and what makes someone a screen failure versus a withdrawal, you have a defensible boundary even without an external definition.

How to calculate screen failure rate

The formula is simple. The inclusion rules are where consistency breaks down.

Screen failure rate = (number of screen failures / number of subjects screened) x 100

The denominator is everyone who was screened. The numerator is everyone screened who did not enroll because they did not meet eligibility. The judgment calls are in what counts as “screened” and what counts as a “failure” versus some other disposition.

Case	In numerator (failure)?	In denominator (screened)?	Note
Met all criteria, enrolled	No	Yes	The success case.
Failed an inclusion or exclusion criterion	Yes	Yes	The core screen failure.
Withdrew consent during screening, before an eligibility decision	Usually no (separate disposition)	Yes if a trial procedure occurred after consent	Code as withdrawal, not eligibility failure. Document the consent and withdrawal.
Pre-consent prospect (phone pre-screen, chart review only, no consent, no procedure)	No	No	Not yet a subject. No trial procedure occurred, so they sit outside the screened population. Track separately if your protocol uses a pre-screening mechanism.
Re-screened subject who fails again	Count the new screening event	Count as a new screening event	Each screening attempt is its own event; define in protocol whether re-screens are counted once or per attempt, and apply it consistently.
Eligible but never dosed for a non-eligibility reason (site closure, logistics)	No (not an eligibility failure)	Yes	Disposition is “not randomized,” reason recorded.

Two consistency rules keep the rate meaningful. First, decide once, in the protocol or statistical analysis plan, whether re-screens are counted as new screened subjects or folded into the original. Second, separate eligibility failures from withdrawals and administrative non-enrollments; lumping them inflates the rate and hides the real signal.

A worked example: 220 candidates consent and begin screening. 80 fail an inclusion or exclusion criterion. 12 withdraw consent mid-screening. 8 are eligible but never randomized due to a site logistics issue. The eligibility-based screen failure rate is 80 / 220 = 36 percent. Note how the 12 withdrawals and 8 logistics drops change the picture if you (wrongly) move them into the numerator: 100 / 220 = 45 percent. Same study, very different story. This is why widely quoted benchmarks like “about 36 percent” mislead. They average across studies that count differently. Your own consistently computed trend over time is far more useful than any cross-study benchmark.

Why screens fail: protocol design versus site execution

Reasons cluster into two buckets, and the bucket matters more than the raw count because it tells you where the fix lives.

Protocol-design reasons: inclusion and exclusion criteria that are too narrow, lab thresholds that exclude most of the real-world population, washout requirements that few candidates can meet, and a target population that does not exist in the numbers the feasibility assumptions promised. ICH E8(R1) Section 7 notes that clinical investigators and potential study participants have valuable insights into the feasibility of enrolling participants who meet proposed eligibility criteria. When most failures are eligibility-criterion failures, the protocol, not the site, is the lever.

Site-execution reasons: incomplete pre-screening, scheduling that lets conditions change before the eligibility visit, missed source documentation that forces a conservative exclusion, and inconsistent application of criteria across coordinators. These show up as the same exclusion codes appearing at one site but not its peers.

Mapping each failure to a reason code, and each reason code to “design” or “execution,” turns a vague high rate into an actionable diagnosis.

The GCP documentation a screen failure triggers

Every screened subject generates records, whether they enroll or fail. The screening log is an essential document, not an optional convenience. ICH E6(R3) Section 8 (essential records) lists, among the records that document trial conduct, the completed participant identification code list and enrolment log and a completed participants screening log, alongside the completed signed and dated informed consent forms. A screening log that captures only the people who enrolled is incomplete by definition.

Eligibility is a determination that needs evidence behind it. ICH E6(R3) §2.13.2 requires the investigator/institution to maintain adequate source records that include pertinent observations on each trial participant, and states that source records should be attributable, legible, contemporaneous, original, accurate and complete, with the investigator defining what is considered a source record before the trial starts. An “ineligible: hemoglobin too low” entry with no lab report behind it is an assertion, not a record.

21 CFR Part 312 reinforces this from the investigator side. The case-history requirement in §312.62(b) states that the case history for each individual shall document that informed consent was obtained prior to participation in the study, and that case histories include the case report forms and supporting data including signed and dated consent forms and medical records. The investigator’s general responsibility in §312.60 is to ensure the investigation is conducted according to the signed investigator statement, the investigational plan, and applicable regulations, and to obtain the informed consent of each human subject in accordance with Part 50.

Screening log and documentation checklist, for every screened subject (consenting and not):

• Subject screening identifier and screening date.
• Whether informed consent was obtained, with the consent date, and confirmation that consent preceded any trial-related procedure.
• Eligibility outcome: enrolled, screen failure, or other disposition.
• For a screen failure, the specific inclusion or exclusion criterion not met, with a reason code.
• Source documentation supporting the eligibility determination (lab report, medical record, assessment).
• For withdrawals, the withdrawal record kept distinct from eligibility failures.

Re-screening: when it is allowed, and the record it demands

Re-screening is not a quiet retry. It is a new screening event for a subject who previously failed, and the regulations treat it as such, even though none of the in-scope regulations name “re-screening” as a defined term. Permissibility comes from the protocol: re-screening is allowed when the protocol provides for it and defines the conditions (for example, a transient lab value expected to normalize). If the protocol is silent, you do not improvise; you amend it.

Because every screening event must be preceded by valid consent, a re-screen generally requires a fresh, signed and dated consent before any new trial-related procedure, under the same ICH E6(R3) §2.8.7 and §1.4 (Principle 4) rule that governed the first attempt. A consent signed for the first screening attempt does not automatically cover a screening procedure performed weeks later, and ICH E6(R3) §2.13.2 source-record expectations mean the re-screening assessments need their own contemporaneous source documentation, traceable to the right event.

Re-screening decision checklist:

• Does the protocol explicitly permit re-screening for this reason? If not, stop and amend.
• Is a new informed consent signed and dated before any new screening procedure? (Confirm the consent date precedes the procedure date.)
• New subject number, or the same one? Follow the protocol’s convention and apply it consistently across the study so the screening log and rate calculation stay coherent.
• Are the re-screening assessments captured as their own contemporaneous source records, linked to the re-screen event rather than overwriting the original?
• Is the re-screen reflected in the screening log and counted consistently in the rate?

Reading the screen-failure rate as a critical-to-quality signal

This is where the recruitment-statistic framing fails the practitioner. A persistently high screen-failure rate is rarely a “try harder at recruitment” problem. It is usually telling you the eligibility criteria are mis-calibrated against the population that actually exists. ICH E8(R1) §3.2 frames critical-to-quality factors as attributes of a study whose integrity is fundamental to the protection of study participants and to the reliability and interpretability of the results, and the Quality-by-Design approach in ICH E8(R1) §3.1 sets out to ensure that quality is driven proactively by designing quality into the study protocol and processes.

Eligibility criteria are squarely one of those factors. ICH E8(R1) Section 7 states that the eligibility criteria should be reflective of the study objectives and be well documented in the clinical study protocol, and that feasibility should be assessed so the study is operationally viable. Read together, these say the quiet part out loud: if you are failing most of the people you screen, the design, not the site, is most likely the thing out of tolerance. The screen-failure rate, trended and decomposed by reason code, is one of the cleanest early read-outs you have on whether your protocol was calibrated to reality.

None of this certifies anything. A clean screening log and a sensible rate do not make a study compliant; they are evidence that supports the sponsor’s and investigator’s responsibilities, which under 21 CFR §312.60 and ICH E6(R3) remain with the people running the trial. The tooling and the log enable a defensible position. The accountability stays human.

Where teams get it wrong

The two findings that recur in BIMO-style inspections are the same two this article opened with. First, an incomplete screening log: only enrolled subjects recorded, screen failures and non-consenting prospects missing, so the denominator of your own rate is unverifiable and the essential-document expectation in ICH E6(R3) Section 8 is unmet. Second, a consent that post-dates a screening procedure: a lab or scan time-stamped before the signature on the consent form, which collides directly with ICH E6(R3) §1.4 (Principle 4), §2.8.7, and the 21 CFR §312.62(b) case-history requirement that consent be documented as obtained prior to participation. A third, quieter failure is treating re-screening as a retry on the same record, which destroys the traceability that §2.13.2 source records demand. Fix the boundary in your protocol, log every screened subject, keep consent ahead of every procedure, and the rate becomes a signal you can actually use.

Sources

• ICH E6(R3) Good Clinical Practice, version r3 (ICH, 2025) — https://www.ich.org/page/efficacy-guidelines
• ICH E8(R1) General Considerations for Clinical Studies, version r1 (ICH, 2021)
• 21 CFR Part 312 Investigational New Drug Application, version 2026-04 (FDA)