Pilot and feasibility studies
Why this level?
The design family and its defining features have not been specified for this hierarchy. NHMRC 2009 sourceA pilot answers whether the study can be run, not whether the intervention works. It is not underpowered evidence about an effect; it is evidence about feasibility. Reporting a pilot's effect estimate as a finding is the mistake this page exists to prevent.
What question does this help answer?
Not "does the treatment work". It answers the operational questions whose answers determine whether the real trial is viable, and it is judged against progression criteria rather than a p-value.
When to use it
Use it before a large trial in an untested setting, with an untested intervention, or where recruitment is genuinely uncertain. It is most valuable where the definitive trial is expensive enough that failing early is worth paying for.
When not to use it
Do not use it to estimate a treatment effect. It is underpowered for that by construction, and a promising effect from a pilot is a common reason a definitive trial later disappoints. Do not run one where feasibility is not actually in doubt.
How it works in plain language
The pilot mimics the definitive trial in miniature and measures process outcomes: recruitment rate, adherence, retention, completeness of outcome data. PROGRESSION CRITERIA are set in advance, stating what results would mean go, amend, or stop, so that the pilot is capable of failing.
Why Mallard may recommend it
Mallard may recommend a pilot when the investigator describes an untested setting or a recruitment target that looks optimistic against the described resources.
Key assumptions and design requirements
- The objectives are feasibility outcomes, stated as such, not efficacy.
- Progression criteria are prespecified, including what would stop the programme.
- The report does not present the treatment effect as a finding, and any estimate carries its imprecision plainly.
- The sample size is justified by what is needed to estimate a recruitment rate or a standard deviation, not by power.
- Whether the pilot is internal (its data roll into the main trial) or external is stated up front.
What to watch out for
A pilot is not a small trial of effectiveness. Estimating a treatment effect from it is underpowered by construction, and a promising effect estimate from a pilot is one of the commonest reasons a definitive trial is later disappointed. Progression criteria should be set in advance so the pilot can fail.
Clinical example
Question: Can we recruit 200 patients for a discharge-support trial across three hospitals in a year? Design: Run a six-month pilot in one hospital, measure recruitment rate, consent rate and outcome completeness against prespecified progression criteria, and report the effect estimate only as an imprecise by-product.
- CONSORT 2010 statement: extension to randomised pilot and feasibility trials (opens in a new tab) · BMJ · doi 10.1136/bmj.i5239
- Overview of clinical study designs (opens in a new tab) · Clin Exp Emerg Med · doi 10.15441/ceem.23.036