Pilot Study in MD/MS Thesis: Purpose, Size and Reporting

"Do a pilot study first" is advice almost every PG resident hears, usually when the sample size calculation is stuck because nobody knows the standard deviation of the outcome, or when a questionnaire has never been used in the local language. A well-planned pilot can save a thesis from months of trouble. A poorly planned one wastes time, needs its own ethics approval anyway, and sometimes produces numbers that mislead the main study.

This guide explains what a pilot study is for, how large it should be (and why there is no single correct number), how to use pilot estimates for sample size safely, whether pilot data can go into your main analysis, and how to report the pilot in your synopsis and thesis. Your guide and department may have their own expectations; discuss the plan with them before you start.

1What a Pilot Study Is (and Is Not)

A pilot study is a small-scale version of the planned study, or of part of it, done to find out whether the main study can be carried out as designed. Methodologists distinguish a few related terms:

• Feasibility study: asks whether the study can be done at all. Are there enough eligible patients? Will they agree to take part? Can the procedures be done in your setting?
• Pilot study: a miniature run of the main study, testing the processes together: recruitment, consent, randomisation, data collection and follow-up.
• Pretesting a tool: trying a questionnaire or proforma on a small group to check that questions are understood, the order works and the time taken is acceptable.

What a pilot is not: a small study to test the hypothesis. A pilot is not designed or powered to show whether a treatment works, so its p-values for effectiveness mean very little. If a pilot "shows" a significant difference, that is not a reason to skip or shrink the main study.

2Why Do a Pilot in a Thesis?

Feasibility of recruitment and procedures

Thesis timelines are fixed by your university and NMC regulations. A short pilot tells you how many eligible patients actually present per month, how many consent, and whether your inclusion criteria are realistic. If you need 120 patients and the pilot suggests only a few eligible patients a month, you know early that the criteria, setting or duration must change.

Testing tools and the CRF

A pilot shows whether your case record form captures every variable, whether questions are understood in the local language, how long the interview takes, and whether measurement procedures (for example, timing of a blood sample or a scoring method) work in your OPD or ward. Our guide on CRF, proforma and master chart explains what to look for.

Estimates for the sample size

Sample size formulas need assumptions: an expected prevalence for a descriptive study, or a standard deviation (SD) and a clinically important difference for comparing means. When no published study from a similar population gives these values, a pilot can provide an estimate. See our sample size guide for the formulas, and try the sample size calculator with different assumptions.

3How Big Should a Pilot Be?

There is no universal rule. Different methodologists have proposed different rules of thumb, and published suggestions range from about a dozen participants per group to several dozen, depending on what the pilot is meant to estimate. Any single number quoted as "the rule" in your department is a convention, not a requirement.

A more defensible approach is to size the pilot according to its purpose:

Write down your reasoning in the protocol: "A pilot of __ participants will be conducted to estimate the standard deviation of ___ and to assess recruitment rate, because no published data from a comparable population were found." Reviewers accept a clear justification far more readily than an unexplained number.

4Using Pilot Estimates for Sample Size Safely

Small samples give unstable estimates. An SD from a small pilot can easily be smaller than the true value, which leads to an underpowered main study. To reduce this risk:

• Be conservative with the SD. Some statisticians suggest using a value larger than the pilot SD, such as its upper confidence limit, rather than the point estimate.
• Do not use the pilot effect size as the expected difference. The difference you power the main study for should be the smallest clinically important difference, based on clinical judgement and literature, not the difference seen in a small pilot.
• Compare with published values where any exist, and justify the value you choose.
• Allow for dropouts using the loss-to-follow-up rate the pilot showed, if it gave one.

5Can Pilot Data Go into the Main Analysis?

It depends on the type of pilot and on what you changed afterwards.

External pilot

An external pilot is a separate, small study done before the main study. Its data are generally not included in the main analysis, especially if you changed the protocol, tools, criteria or procedures as a result. Including them would mix participants studied under different conditions.

Internal pilot

An internal pilot is the first phase of the main study, planned in advance as part of it, often to check recruitment or re-estimate sample size. Its participants are usually included in the final analysis, provided the protocol did not change in a way that affects the outcome, and provided this was specified before the study began.

Whatever you decide, state it in the protocol before you begin, so the decision cannot be influenced by the results.

6Ethics Approval and Timing

A pilot that involves participants, their samples or their records is research and needs IEC approval and informed consent, just like the main study. In a thesis, the usual approach is to describe the pilot within the protocol you submit to the IEC, so that one approval covers both the pilot and the main study. If the pilot leads to changes in the protocol, tools or consent documents, submit them as an amendment before continuing. Our guide on IEC approval for an MD/MS thesis covers amendments.

Plan the time carefully. A pilot adds weeks to the schedule: recruitment, analysis, revisions, and possibly an IEC amendment. Start early enough that the main study still fits within your thesis period.

7Reporting the Pilot in Your Synopsis and Thesis

In the synopsis or protocol, describe:

• The purpose of the pilot (feasibility, tool testing, estimate of SD or prevalence).
• Its size and how it was decided.
• Whether it is external or internal, and whether its data will be included in the main analysis.
• What criteria will lead to changes (for example, recruitment below a planned rate).

In the thesis Methods chapter, report:

• When and where the pilot was done, and how many participants took part.
• What was learnt: recruitment and consent rates, problems with tools, the SD or prevalence obtained.
• What was changed as a result, and how the final sample size was calculated.

Report feasibility findings descriptively (numbers, proportions, confidence intervals where useful), not as tests of effectiveness. If you publish the pilot separately, the CONSORT extension for randomised pilot and feasibility trials gives a useful checklist.

8Getting Help with Your Pilot and Sample Size

If you are unsure whether your thesis needs a pilot, how to justify its size, or how to turn pilot results into a defensible sample size, the PubMedico team works with MD, MS and DNB residents on these protocol decisions. Message us on WhatsApp at +91 78787 33004.