Eligibility Criteria

Overview

Eligibility criteria define who may enter a study and under what conditions. They convert the population and research question into operational decisions that can be applied consistently during screening.

Good criteria do more than produce a list of acceptable participants. They protect safety, support valid measurement, preserve a meaningful study population and make recruitment feasible. Poor criteria may exclude people unnecessarily, introduce subjective screening, reduce applicability or permit post hoc removal of participants whose data appear inconvenient.

This lecture distinguishes inclusion, exclusion, withdrawal, discontinuation and analysis criteria and provides a structured approach to writing and appraising them.

Why this matters

Eligibility decisions can change the clinical meaning of a study. Consider two studies of postpartum glucose testing:

• Study A includes all women diagnosed with gestational diabetes who delivered at participating hospitals and remained contactable for 12 weeks.
• Study B includes only women aged 25–35 years, fluent in English, with no comorbidity, attending every antenatal visit and owning a smartphone.

Study B may be easier to administer, but its criteria exclude many women for reasons unrelated to the central question. Its findings may underestimate service barriers and have limited applicability.

Eligibility criteria influence:

• participant safety and ethical acceptability;
• clarity and reproducibility of screening;
• expected outcome frequency and variability;
• recruitment rate and study duration;
• selection bias and effect modification;
• burden on patients and staff; and
• applicability to routine practice.

Key concepts

Inclusion criteria

Inclusion criteria identify the characteristics required for a person or unit to be eligible. They commonly specify:

• clinical condition or exposure;
• age or life stage;
• setting and geographical area;
• calendar period;
• timing in relation to diagnosis, treatment or delivery;
• capacity or pathway for informed consent; and
• availability of information necessary to establish eligibility.

Each criterion should be necessary and operationally testable.

Exclusion criteria

Exclusion criteria identify otherwise eligible people who should not enter because inclusion would create unacceptable risk, make the central measurement uninterpretable, conflict with the study purpose or make the protocol impossible to implement.

Exclusion criteria should not merely repeat the inverse of inclusion criteria. For example, if inclusion requires age 18 years or older, “age below 18 years” need not be repeated unless the protocol template requires it.

Withdrawal, intervention discontinuation and analytical exclusion

These events occur after enrolment and should not be disguised as eligibility decisions.

• Withdrawal occurs when a participant revokes consent for some or all further participation.
• Intervention discontinuation means the assigned intervention stops, but follow-up may continue if consent permits.
• Loss to follow-up means outcome information cannot be obtained despite planned attempts.
• Analytical exclusion means enrolled data are omitted from a particular analysis according to a prespecified rule.

Building criteria from the research question

Eligibility should be developed in the following order:

1. define the target and source populations;
2. identify the primary objective and estimand;
3. specify the minimum clinical features needed to answer the question;
4. identify genuine safety or ethical restrictions;
5. determine what can be measured reliably;
6. evaluate operational feasibility;
7. assess how each restriction affects applicability; and
8. pilot the screening procedure.

DECISION GUIDE

Test every proposed criterion

For each inclusion or exclusion criterion, ask:

• Which objective, safety requirement or measurement need does it serve?
• Can two trained screeners apply it consistently?
• Is the required information available before consent or enrolment?
• Could the same issue be managed by adaptation rather than exclusion?
• Which real patients would be removed, and would that alter the intended inference?
• Is the criterion scientifically justified or merely convenient?

Operational wording

“Severe disease”, “poor compliance”, “unable to cooperate” and “unreliable participant” are not adequate criteria without operational definitions. They invite subjective judgement and may encode stigma or clinician preference.

A criterion should name the observable rule and source. For example:

• “Delivery recorded at a participating hospital from 1 January to 31 December 2025.”
• “Gestational diabetes documented according to the hospital diagnostic protocol.”
• “Resident within the state through 12 weeks postpartum.”
• “Unable to provide informed consent and no approved proxy-consent pathway is available.”

Eligibility table for the illustrative cohort

illustrative

Table 1. Proposed eligibility criteria and rationale

Criterion	Type	Operational rule	Rationale
Gestational diabetes	Inclusion	Diagnosis documented in the maternity record	Defines the clinical population
Delivery at participating hospital	Inclusion	Live birth or stillbirth recorded during the study period	Defines the source population
Follow-up period	Inclusion	Expected to remain in the state for 12 weeks	Enables outcome ascertainment
Critical illness at approach	Temporary deferral	Approach postponed until clinically appropriate	Protects welfare and voluntariness
Smartphone ownership	Not used	Alternative telephone or record follow-up available	Avoids unnecessary digital exclusion
Chronic disease	Not used as exclusion	Recorded as a covariate where relevant	Preserves routine-care applicability

Note: Criteria and rationales are illustrative.

Balancing safety, validity and applicability

Restrictive criteria can reduce heterogeneity, but homogeneity is not automatically desirable. If the research question concerns routine health-service performance, excluding complex patients can remove precisely those most likely to experience access barriers.

In intervention studies, restrictions may be needed for contraindications, drug interactions or inability to complete essential safety assessments. Even then, researchers should distinguish:

• scientifically necessary restrictions;
• temporary deferrals;
• modifiable barriers, such as language or disability accommodations; and
• convenience exclusions that mainly simplify the research team’s work.

Eligibility criteria may also modify the expected treatment effect or event rate. Sample-size assumptions and interpretation should therefore use data relevant to the eligible population, not a broader population with different risk.

Screening and documentation

PROTOCOL CHECKLIST

Eligibility and screening procedures to specify

• complete inclusion and exclusion criteria;
• operational definition and source for each criterion;
• who performs screening and what training is required;
• order of screening, approach and consent;
• management of uncertain or temporarily unmet criteria;
• process for resolving disagreements;
• screening-log variables and reasons for non-enrolment;
• handling of ineligible participants enrolled in error; and
• distinction between withdrawal, loss to follow-up and analysis exclusions.

A screening log should use standard reason categories. It should not collect more identifiable information than governance approvals permit. Aggregate reporting of the numbers assessed, excluded and reasons helps readers judge recruitment and applicability.

Ethical and governance considerations

Interface with statistical analysis

Eligibility determines the population for which the primary estimate is intended. It can alter baseline risk, variability, event frequency and the distribution of effect modifiers.

The protocol should separately define:

• the enrolled population;
• the safety population;
• the intention-to-treat or treatment-policy population where relevant;
• per-protocol or adherent populations;
• outcome-specific denominators; and
• rules for missing or invalid measurements.

Post-randomisation exclusions can undermine the protection of random allocation. In observational research, excluding participants based on variables affected by the exposure may create selection bias.

Eligibility criteria and the estimandMMed extension

An estimand identifies the treatment condition, population, outcome, summary measure and handling of intercurrent events. Eligibility defines the population attribute of that estimand. A trial can therefore be internally coherent yet answer a question about a population narrower than routine practice. Explicit estimands help separate a deliberate population restriction from an accidental loss of applicability.

Worked example

WORKED EXAMPLE

Revising an unnecessarily restrictive eligibility list

Original exclusions: any chronic disease, non-English speaker, caesarean delivery, residence more than 20 kilometres away, no smartphone.

Primary objective: estimate postpartum glucose-test completion under routine public-hospital care and identify barriers.

Review:

• Chronic disease may affect follow-up and should usually be measured, not excluded.
• Language can be accommodated with approved translated materials or interpreters.
• Mode of delivery does not invalidate the outcome.
• Distance is a potential explanatory factor and should not be removed.
• Smartphone ownership is unnecessary when telephone and record-based options exist.

Revised criteria: documented gestational diabetes, delivery at a participating hospital during the study period, residence within the state for the outcome window, and an ethically approved consent pathway for the prospective component.

Result: The revised sample better reflects the service population and preserves clinically important variation.

Suggested protocol or reporting wording

Evidence basis and critical appraisal

The World Health Organization protocol structure requires the study population and eligibility criteria to be stated in a way that supports ethical and scientific review [1]. Criteria should therefore be evaluated as part of the research question, not as an administrative appendix.

Restrictions can substantially narrow the population to which findings apply [2]. Empirical assessment of major clinical trials has demonstrated that apparently routine criteria may exclude many patients seen in practice [3]. Critical appraisal should distinguish restrictions needed for participant safety or interpretable measurement from those introduced mainly for convenience.

Current good clinical practice guidance emphasises participant protection, proportionate procedures, protocol clarity and reliable trial conduct [4]. Reporting guidance for observational studies likewise expects eligibility, selection and participant flow to be transparent [5].

For each criterion, reviewers should ask whether:

• the rationale is explicit and prospectively defined;
• the wording can be applied consistently across screeners and sites;
• an accommodation could replace exclusion;
• the criterion changes baseline risk or effect-modifier distributions;
• screening failures and reasons are documented; and
• post-enrolment events are reported separately.

Common mistakes

Knowledge check

Should a participant who stops the assigned intervention be declared ineligible?

Answer and reasoning

No. Eligibility was determined before enrolment. Intervention discontinuation is a post-enrolment event. Follow-up may continue if the participant consents and the protocol permits it.

When is an exclusion for language potentially justified?

Answer and reasoning

Only when the study cannot ethically and validly obtain consent or essential data despite proportionate accommodations, and the restriction is approved and transparently justified. Convenience alone is insufficient.

Why is 'likely to be non-compliant' a poor criterion?

Answer and reasoning

It is subjective, difficult to reproduce and may reflect clinician bias. It may also remove participants at high risk of the outcome, thereby distorting the study population.

What should happen when an eligibility criterion is uncertain at screening?

Answer and reasoning

The protocol should specify verification, temporary deferral or adjudication. Staff should not improvise different rules across participants or sites.

Key take-home messages

SUMMARY

Key take-home messages

• Derive eligibility criteria from the population, objective, safety needs and measurement requirements.
• Make every criterion operational and testable before enrolment.
• Separate baseline eligibility from withdrawal, discontinuation, loss to follow-up and analysis exclusions.
• Avoid restrictions introduced mainly for researcher convenience.
• Record screening outcomes and reasons for non-enrolment.
• Judge every restriction by its effects on safety, validity, feasibility and applicability.

References

1. World Health Organization. Recommended format for a research protocol. Geneva: World Health Organization. https://www.who.int/groups/research-ethics-review-committee/recommended-format-for-a-research-protocol
2. Rothwell PM. External validity of randomised controlled trials: “to whom do the results of this trial apply?” Lancet. 2005;365(9453):82–93. https://doi.org/10.1016/S0140-6736(04)17670-8
3. Van Spall HGC, Toren A, Kiss A, Fowler RA. Eligibility criteria of randomized controlled trials published in high-impact general medical journals: a systematic sampling review. JAMA. 2007;297(11):1233–1240. https://doi.org/10.1001/jama.297.11.1233
4. International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. ICH harmonised guideline E6(R3): guideline for good clinical practice. Geneva: ICH; 2025. https://www.ich.org/page/efficacy-guidelines
5. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453–1457. https://doi.org/10.1016/S0140-6736(07)61602-X