03 · What You Need to Know
Determining When Feasibility Evidence Justifies Stopping a Proposed Study
What Does a Decision Not to Proceed Actually Mean?
A decision not to proceed means that the available evidence does not support undertaking the proposed definitive study under its current conditions.
It does not necessarily mean that the research question is unimportant, the intervention is ineffective, or the broader research program should end.
For example, a university may be unable to conduct a cluster-randomized trial because too few independent classes are available. That finding concerns the feasibility of the proposed design, not the educational value of the intervention being investigated.
The research team may later develop a different design or undertake the investigation in another setting.
Researchers should distinguish stopping a particular study from abandoning the underlying scientific question.
Can a Feasibility Study Be Successful Even When It Recommends Stopping?
Yes. The success of feasibility research should be evaluated according to whether it answers its preliminary questions and supports an informed decision.
Eldridge and colleagues describe feasibility and pilot research as preparation for future definitive evaluation. The purpose is to establish whether and how that evaluation can be conducted.
If a feasibility study demonstrates that an essential procedure cannot be implemented, it may prevent an expensive or ethically problematic definitive investigation.
That contribution can be scientifically valuable even when the original project does not proceed.
However, a decision not to proceed should follow credible evidence rather than frustration with preliminary difficulties or an arbitrary threshold.
What Types of Problems May Justify Not Proceeding?
The most consequential problems are those that prevent the definitive study from answering its intended research question or satisfying essential ethical and operational requirements.
| Feasibility problem |
Potential consequence |
Possible decision |
| Insufficient eligible participants |
The required sample or number of clusters cannot be obtained. |
Expand access, redesign, or stop the proposed design. |
| Unacceptable participant burden |
Procedures may be impractical or ethically inappropriate. |
Reduce burden, revise procedures, or discontinue. |
| Unreliable intervention delivery |
Essential intervention components cannot be implemented. |
Improve delivery or reconsider the intervention. |
| Unusable primary outcome data |
The definitive research question may remain unanswered. |
Revise measurement or stop if no suitable alternative exists. |
| Compromised allocation procedures |
The intended randomized comparison may lack adequate safeguards. |
Correct allocation procedures or redesign the trial. |
| Unresolved privacy or ethical concerns |
Participant protections may be inadequate. |
Do not proceed until the concerns are resolved and required approvals obtained. |
| Unsustainable resource requirements |
The study cannot be completed within available capacity. |
Secure resources, reduce nonessential demands, or stop. |
These problems vary in severity. Some may be corrected through relatively modest changes, while others challenge the basic assumptions of the study.
What Is the Difference Between a Remediable Barrier and a Fundamental Feasibility Problem?
A remediable barrier is a problem that can plausibly be addressed without undermining the essential purpose or integrity of the proposed study.
A fundamental feasibility problem is one for which no realistic corrective action has been identified under the available conditions.
Remediable feasibility barrier
A difficulty that may be addressed through justified changes to procedures, staffing, scheduling, recruitment, or implementation arrangements.
Fundamental feasibility problem
A limitation that prevents essential requirements from being met and cannot reasonably be resolved within the proposed study's constraints.
For example, recruitment delays caused by an inconvenient invitation schedule may be remediable.
By contrast, a study requiring 30 independent institutions may be infeasible if only four eligible institutions exist and no defensible alternative sampling frame is available.
The distinction depends on evidence about the available options, not merely the severity of the observed shortfall.
Should Recruitment Failure Lead to a No-Go Decision?
Potentially, when recruitment cannot meet the definitive study's requirements despite realistic corrective options.
Researchers should first determine whether the problem concerns access, eligibility, willingness to participate, or recruitment speed.
For example, low enrollment may result from an ineffective communication strategy rather than an insufficient eligible population.
Adding suitable recruitment sites or extending the enrollment period may be defensible if the resources and design permit.
However, expanding eligibility merely to increase numbers may alter the target population and undermine the original research question.
A no-go decision becomes more defensible when the necessary recruitment capacity cannot be established without unacceptable methodological changes.
This decision depends on the recruitment rate and recruitment pathway, not simply the total number enrolled during the feasibility period.
Can Poor Retention Justify Stopping the Definitive Study?
Yes, when the proposed study depends on continued participation and the feasibility findings suggest that essential follow-up information cannot be obtained.
For example, a longitudinal investigation may recruit sufficient participants but lose many before its primary outcome assessment.
Researchers should examine whether attrition reflects scheduling, participant burden, accessibility, or circumstances that cannot readily be changed.
Some problems may be addressed through revised follow-up arrangements.
However, increasing initial enrollment does not necessarily correct bias associated with missing outcomes.
If the study cannot obtain interpretable follow-up evidence, proceeding may be difficult to justify.
When Does Data Completeness Become a Critical Barrier?
Data completeness becomes particularly consequential when essential observations are missing or unusable.
For example, a trial may retain most participants while failing to collect valid primary outcome measurements because of assessment or technical problems.
A feasibility study should identify whether the problem can be corrected through improved procedures, alternative measurement arrangements, or technical verification.
However, replacing the primary outcome may substantially change the definitive study.
If no appropriate measurement strategy is available, the research may not be ready to proceed.
The issue should be assessed through the availability of usable data rather than an overall percentage of nonblank records.
Can Intervention Delivery Problems Justify Not Proceeding?
Yes. A definitive effectiveness study may be difficult to interpret if the intervention cannot be delivered in a manner consistent with its essential components.
For example, an AI-supported formative feedback intervention may require teachers to review generated feedback before students receive it.
If the research team cannot provide the necessary review capacity, the intervention may be delivered differently from what the study intends to evaluate.
Researchers may consider additional training, staffing, or workflow modifications.
However, removing an essential component solely to make implementation easier may produce a substantially different intervention.
The decision should consider whether the revised intervention still represents the intended object of evaluation.
What If Participants Find the Intervention Unacceptable?
Participant acceptability can become a substantial feasibility concern when the intervention imposes burdens or raises objections that cannot be addressed appropriately.
For example, students may object to intrusive learning analytics or extensive monitoring of their activities.
Researchers should investigate the nature of these concerns rather than rely exclusively on average satisfaction scores.
Some problems may be resolved through clearer communication or less intrusive procedures.
Others may require substantial redesign or indicate that the proposed intervention is inappropriate for the intended population.
Ethical concerns should not be dismissed simply because most participants express favorable opinions.
The relevant evidence concerns acceptability across the affected participants and procedures.
Can Resource Constraints Justify a No-Go Decision?
Yes, particularly when essential staffing, funding, infrastructure, or technical support cannot be secured.
For example, a multisite study may require specialized assessors at every institution, but the available budget may support only a fraction of that requirement.
Researchers should distinguish avoidable inefficiencies from resources necessary to preserve methodological quality and participant safeguards.
Some costs may be reduced through better scheduling or shared infrastructure.
However, eliminating essential measurements or security procedures merely to meet the budget may make the study less defensible.
A decision not to proceed may be appropriate when the necessary resources cannot be provided sustainably.
Can Ethical Concerns Override Favorable Feasibility Results?
Yes. Favorable recruitment, retention, or intervention adherence does not justify proceeding when essential ethical requirements remain unresolved.
For example, a study may recruit participants efficiently while relying on a data collection arrangement that lacks necessary authorization or appropriate confidentiality safeguards.
Such a problem cannot be offset by favorable performance elsewhere.
Researchers must comply with applicable ethical review, consent, data protection, and institutional requirements.
A feasibility study does not replace formal approval or authorization.
Where a critical safeguard cannot be implemented, the affected procedures should not proceed until the problem has been appropriately resolved.
Should Progression Criteria Determine Whether the Study Stops?
Progression criteria can provide a structured basis for deciding whether to proceed, modify, investigate further, or stop.
However, they should be justified by the definitive study's requirements and interpreted in context.
For example, a recruitment threshold may reflect the minimum enrollment capacity needed to complete the definitive study within its available period.
Falling below the threshold may indicate a problem, but the implications depend on uncertainty and whether realistic modifications are available.
Some conditions, such as essential ethical authorization, may be non-negotiable.
The role of prespecified progression criteria is to support transparent decisions, not to eliminate methodological judgment.
Does a Red Progression Classification Mean the Study Must Stop?
Not automatically. A red classification commonly indicates that the current design faces substantial feasibility problems.
It may support stopping the proposed design, undertaking major modifications, or conducting additional preliminary research.
However, the meaning of red should be established within the study's decision framework.
Researchers should distinguish problems that can be corrected from conditions that make progression inappropriate.
The green-amber-red progression framework can clarify these possibilities, but the color itself is not evidence.
How Should Statistical Uncertainty Affect a No-Go Decision?
Feasibility estimates are often based on relatively small samples and may be imprecise.
For example, observing that 14 of 20 participants complete follow-up gives a completion proportion of 70%.
However, this estimate does not establish that the underlying completion probability is exactly 70%.
Researchers should consider the range of plausible values and the consequences of making an incorrect decision.
A point estimate slightly below a threshold may justify further investigation rather than immediate discontinuation.
Conversely, severe procedural failures may justify action even when a numerical estimate is uncertain.
Should a Feasibility Study Test Intervention Effectiveness Before Recommending Stopping?
Generally, a feasibility study should not base its progression decision on conventional significance testing of preliminary intervention effects.
Small feasibility studies are commonly designed to assess whether a definitive evaluation can be conducted rather than to provide conclusive evidence of effectiveness.
A statistically nonsignificant preliminary effect does not establish that the intervention is ineffective.
Similarly, a favorable preliminary effect cannot compensate for an infeasible or ethically problematic design.
Researchers should distinguish decisions about study feasibility from conclusions about intervention efficacy or effectiveness.
Can a Feasibility Study Demonstrate That the Research Question Is Unanswerable?
It may demonstrate that the question cannot be answered using the proposed design under the available conditions.
However, that conclusion should not automatically be generalized to every possible methodology.
For example, a randomized trial may be infeasible because the number of independent clusters is insufficient.
A different research design might still provide useful evidence, although it may support different forms of inference.
Researchers should explain which methodological requirements cannot be satisfied and what alternatives remain plausible.
Changing the design may also require revising the claims the study can support.
How Should Researchers Distinguish Redesign From Discontinuation?
Redesign involves changing the proposed study to address feasibility problems while retaining a defensible research purpose.
Discontinuation means deciding not to undertake the proposed definitive investigation under the current circumstances.
These options may overlap. A research team may discontinue one design and subsequently develop another.
For example, a proposed intervention trial may be replaced by an implementation study if the intervention is not sufficiently established for effectiveness evaluation.
However, the new study should be justified on its own methodological terms.
Researchers should not describe a substantially different investigation as the original study merely to preserve continuity.
When Is Additional Feasibility Testing Preferable to Stopping?
Additional testing may be appropriate when the preliminary findings identify a plausible corrective action whose effectiveness remains uncertain.
For example, a recruitment pilot may reveal that institutional invitations are poorly timed.
Revising the invitation schedule could be straightforward, but the researchers may still need evidence that the new procedure produces sufficient enrollment.
A focused follow-up test may answer that question without repeating the entire feasibility study.
However, repeated testing should not continue indefinitely without a credible prospect of resolving the central problem.
The question of repeating feasibility research after redesign depends on whether consequential uncertainty remains.
Can the Sunk-Cost Problem Influence Progression Decisions?
Yes. Researchers may feel pressure to continue because substantial time, funding, or professional effort has already been invested.
However, resources already spent cannot establish that additional investment is justified.
The relevant question is whether proceeding now is likely to produce meaningful evidence under acceptable methodological and ethical conditions.
A feasibility study may prevent further expenditure on a design that cannot meet those conditions.
Researchers should distinguish the scientific value of the preliminary work from the feasibility of the proposed next phase.
Who Should Participate in a Decision Not to Proceed?
The appropriate decision-making arrangement depends on the research design, governance requirements, and consequences of the decision.
Relevant contributors may include investigators, statisticians, implementation specialists, funders, institutional representatives, and participant advisers.
For a trial, an independent steering or oversight group may provide valuable scrutiny.
However, consultation should not obscure responsibility for the final decision.
Researchers should identify who has authority to approve progression, redesign, or discontinuation under the applicable arrangements.
How Should Researchers Communicate a No-Go Decision?
The explanation should identify the original feasibility questions, observed findings, progression criteria, and reasons the proposed definitive study is not supported.
It should distinguish problems that were investigated from uncertainties that remain unresolved.
For example, researchers may conclude that recruitment capacity is insufficient under the available institutional arrangements, rather than claim that the target population is unwilling to participate generally.
Where appropriate, the report may identify conditions under which a redesigned study could become feasible.
The conclusion should be proportionate to the evidence and should not characterize the preliminary investigation as a scientific failure merely because the definitive study will not proceed.
Should Negative Feasibility Findings Be Published?
They may warrant publication when the study was methodologically sound and its findings provide useful information for researchers or practitioners.
For example, evidence about recruitment barriers, intervention workload, or impractical outcome assessments may inform future studies in similar settings.
However, publication suitability depends on the study's contribution, reporting quality, and the requirements of the intended journal.
Researchers should report the feasibility objectives and findings transparently, including limitations.
The CONSORT extension for randomized pilot and feasibility trials provides relevant reporting guidance for studies within its scope.
How Should a Decision Not to Proceed Be Documented?
The research record should preserve the original objectives, prespecified progression criteria, findings, and reasoning behind the decision.
It should also describe alternative actions considered and explain why they were insufficient or inappropriate.
Where the study was registered or subject to institutional oversight, relevant reporting and notification obligations should be followed.
Researchers should avoid retrospectively changing the objectives to make unfavorable findings appear consistent with an originally preferred outcome.
Transparent documentation allows others to understand what was learned and how the decision was reached.
Watch Out
Do not conclude that an intervention is ineffective merely because the proposed definitive study is infeasible. Likewise, do not continue an impractical study solely because substantial resources have already been invested. Feasibility evidence concerns whether and how the research can be conducted, and its implications should remain within that scope.
