A clinical trial can reach its enrollment target and still produce evidence that is too narrow for the population it is intended to serve.
The headline number may look reassuring, but it does not reveal whether the cohort reflects the relevant differences in age, sex, geography, language, health status, or socioeconomic circumstances. It also says nothing about whether those differences remain visible at the end of the study.
This is the overlooked connection between recruitment and retention. Representation can be weak from the outset, but it can also erode during a trial. If participants facing the greatest practical barriers are more likely to withdraw, the group that completes the study may be less representative than the group initially enrolled.
Enrollment is therefore not the finish line. The real scientific and ethical test is whether the people who need to be represented can enter the study, remain in it voluntarily, and contribute meaningfully to the evidence.
Representation must be defined by the intended use of the evidence
A representative cohort is not simply one that mirrors a national census. Its composition should reflect the population affected by the condition and likely to use the intervention.
That distinction matters. The relevant population for a malaria study will not be the same as that for a diabetes, cancer, or maternal health trial. Representation must be designed around disease burden, patterns of care, the intervention’s intended use, and the characteristics that may influence safety or treatment response.
These characteristics can include age, sex, genetic variation, nutritional status, concurrent illness, and the medicines participants are already taking. African populations contain substantial genetic and clinical diversity, yet African data remain underrepresented in many areas of biomedical research (1, 2). Merely locating a trial on the continent does not correct that gap. Diversity must exist within the evidence, not only in the trial address.
The same principle applies to coexisting conditions. Depending on the research question and setting, conditions such as HIV, tuberculosis, malaria, sickle cell disease, hepatitis B, or malnutrition may be part of the clinical reality in which an intervention will be used. Excluding affected participants without a clear scientific or safety rationale can produce findings that are difficult to apply in routine care.
Some exclusions are necessary, particularly in early safety studies. But eligibility criteria should evolve as evidence accumulates. Restrictions appropriate in an early phase should not automatically be carried into later studies without review. Evidence on trial representativeness has repeatedly shown how restrictive criteria and selective participation can limit the applicability of findings (3).
Recruitment determines who enters. Retention determines who remains visible.
Recruitment receives intense attention because it is immediate and easy to measure. Screening rates, enrollment velocity, and milestone dates are monitored closely. Retention is often treated as a later operational concern, addressed when missed visits or withdrawals begin to threaten timelines.
That is a mistake. Retention is not merely about preserving sample size. It is about preserving the integrity of the cohort.
Participants do not leave studies at random. Travel distance, transport costs, lost income, caregiving responsibilities, inflexible work, language barriers, long waiting times, repeated procedures, side effects, relocation, and poor interactions with study staff can all affect continued participation. Attrition can therefore change the composition of the trial population, even when the study began with a reasonably representative cohort.
An acceptable overall retention rate may conceal this shift. A study could retain most participants while losing a disproportionate share from rural communities, older age groups, people with disabilities, or those balancing precarious employment and family responsibilities.
Trial teams should consequently ask more than, “How many participants have we retained?” They should ask, “Who is leaving, when are they leaving, and what does their departure mean for the evidence?”
This is where retention becomes a scientific quality measure. Aggregate figures alone are insufficient. Retention data should be examined across the characteristics that matter to the research question, with appropriate protections for privacy and against misleading analysis of very small groups.
Many apparent participant failures are design failures
The language of “non-compliance” can obscure the source of a problem. A missed visit may be recorded as an individual failure even when the protocol assumes access to reliable transport, flexible employment, childcare, and several uninterrupted hours at a research site.
Those assumptions shape participation long before the first person is recruited.
Site selection is one example. Urban research centers may offer experienced teams and established infrastructure, but an exclusively urban site strategy can place participation beyond the practical reach of rural and peri-urban populations. Feasibility assessments should therefore evaluate population coverage and accessibility alongside enrollment projections and operational capacity.
Consent is another. Translating a technical document does not necessarily make consent meaningful. Participants need information in language and formats they can understand, enough time to consider their decision, and a genuine opportunity to ask questions without pressure. Consent should remain a process throughout the trial, not a signature collected at the beginning.
Visit schedules also create filters. Frequent appointments, narrow visit windows, lengthy procedures, and centralized follow-up can systematically favor people with time, money, proximity, and social support. These requirements should be justified by scientific need, not inherited because they are familiar.
None of this means every withdrawal can or should be prevented. Participants must remain free to leave a study at any time. The obligation is to distinguish voluntary withdrawal from avoidable attrition caused by burdens the study could reasonably reduce.
Community engagement should shape the trial, not decorate it
Community engagement is often concentrated around recruitment: meetings are held, information is shared, and community representatives are asked to help build awareness. Once enrollment closes, engagement may become less visible.
That approach treats communities as a route to participants rather than contributors to better research.
Evidence from infectious disease trials in sub-Saharan Africa shows that community engagement is important to both the ethical and scientific conduct of research, while also revealing that it is not always embedded consistently across the research process (4). Its value begins before recruitment, when community insight can identify impractical eligibility criteria, inaccessible sites, confusing consent materials, and visit schedules that conflict with work, caregiving, or seasonal livelihoods.
It continues during implementation through participant advisory mechanisms, responsive communication, and credible channels for raising concerns. It should also extend beyond the final visit, including appropriate communication about study progress and results.
Engagement does not mean promising that every request will be adopted. It means treating community knowledge as evidence relevant to trial design and delivery, documenting decisions transparently, and explaining constraints honestly.
From enrollment targets to accountable evidence
Improving representation and retention requires action at several decision points.
Protocol developers should define the population to which the findings are intended to apply, justify exclusions, and test whether study procedures create unequal burdens. Flexible visit windows, remote or decentralized follow-up where appropriate, accessible consent, and proportionate data collection can reduce barriers without weakening scientific rigor.
Sites should monitor the participant experience as closely as recruitment performance. This includes examining waiting times, missed visits, complaints, withdrawal patterns, and whether support measures are reaching the people who need them.
Research networks can strengthen accountability by incorporating representation into feasibility assessments and treating retention patterns as quality indicators. Sponsors can support this by avoiding incentives that reward enrollment speed while ignoring cohort composition and avoidable attrition.
Regulators and ethics committees also have an important role. WHO’s guidance for best practices for clinical trials emphasizes inclusivity, participant-centered research, and the generation of evidence that can benefit relevant populations (5). Applying those principles requires scrutiny not only of recruitment plans, but also of who may be excluded by the protocol and who may struggle to remain.
For the Africa Clinical Research Network, community-centered research means viewing recruitment, retention, and representation as one connected responsibility. The goal is not diversity as a headline or enrollment as a performance metric. It is evidence that remains clinically relevant because the people whose experiences matter were able to participate and remain visible.
A trial should not be judged only by how quickly it fills. It should be judged by whether its cohort reflects the population the research is meant to inform, whether participation is realistically accessible, and whether attrition changes whose experience counts.
Enrollment opens the door. The credibility of the evidence depends on who can enter, who can stay, and whose experience is still present when conclusions are drawn.
References
- Pereira L, Mutesa L, Tindana P, Ramsay M. African genetic diversity and adaptation inform a precision medicine agenda. Nat Rev Genet. 2021;22(5):284–306.
- Veale C, Edkins A, Winks S, Njoroge M, Chibale K. Including African data in drug discovery and development. Nature Reviews Drug Discovery. 2023;22:521–2.
- Kennedy-Martin T, Curtis S, Faries D, Robinson S, Johnston J. A literature review on the representativeness of randomized controlled trial samples and implications for the external validity of trial results. Trials. 2015;16(1):495.
- Späth C, Schmidt B-M. An unclosed loop: Perspectives of community engagement in infectious disease clinical trials in sub-Saharan Africa. PLOS ONE. 2024;19(8):e0308128.
- World Health Organization. Guidance for best practices for clinical trials. 2024.