Research Methodology
How to write a research protocol
A protocol is the written plan for a study. This guide explains each section, what it must say, and how to avoid the mistakes reviewers see most often.
- Author
- By ToProt editorial team
- Published
- Updated
- Reading time
- 7 min read
Reviewed by ToProt editorial team on
What a research protocol is for
A research protocol is the written plan for a study, prepared before any data are collected. Supervisors, ethics committees and funders read it to judge two things: whether the question is worth answering, and whether this design can answer it with the people, time and money available. The research team uses it afterwards as the single reference for how the study is carried out, so that every site and every team member works the same way.
A protocol also protects the research. When the question, the primary outcome and the analysis are written down in advance, it is much harder to change them after seeing the results, which is one of the commonest sources of misleading findings.
The standard checklist for trial protocols is SPIRIT [1, 2]. Observational studies have no protocol checklist of their own, but the STROBE reporting guideline [3, 4] lists what a finished observational report must contain, so a protocol that plans for those items is easier to write up later.
Start with a question you can answer
Before writing sections, write one sentence that states the question. Check it against five criteria that are often remembered as FINER:
- Feasible: enough participants, time, money and skills.
- Interesting: to you and to the people who will read the results.
- Novel: it confirms, refutes or extends what is known.
- Ethical: the risks to participants are justified and consent is possible.
- Relevant: it can change knowledge, practice or policy.
For health questions it helps to structure the sentence by population, exposure or intervention, comparison and outcome (PICO for interventions, PECO for exposures). For example: "Among pregnant women attending antenatal clinics in a district (population), how common is anaemia (outcome), and is it associated with iron supplement use (exposure)?"
The sections of a protocol
Institutions use different templates, but nearly all contain the sections below.
| Section | What it must say |
|---|---|
| Title and summary | A specific title, and a short summary of the question, design, population and plan. |
| Background and rationale | What is known, what is missing, and why this study is needed. Cite the evidence. |
| Research question and objectives | One question, one primary objective, a few secondary objectives. |
| Study design | The design (for example cross-sectional, case-control, cohort, randomised trial) and why it fits the question. |
| Population and setting | Who is eligible, where, and how participants are selected. |
| Variables and measurement | Each variable, its definition, how it is measured and its unit or categories. |
| Sample size | The calculation, the inputs and where each input came from. |
| Data collection and management | Instruments, who collects data, storage, protection and quality checks. |
| Analysis plan | The method for each objective, written before the data are seen. |
| Ethics | Approval route, consent, risks and benefits, confidentiality. |
| Timeline and responsibilities | Milestones, who does what, and what could delay the work. |
Write objectives that can be tested
An objective is a specific statement of what the study will estimate or compare. Start each with a verb such as estimate, compare or describe, and name the population, the measure and the setting. Keep one primary objective: the sample size and the main analysis should be built around it.
Choose the design and define the population
Match the design to the question. Estimating how common something is calls for a cross-sectional design; comparing risks between exposed and unexposed groups calls for a cohort or a trial; studying a rare outcome often calls for a case-control design. The companion guide on choosing a study design sets out the options and their limits.
Define the population so that someone else could apply the same rules. List inclusion criteria (who can take part) and exclusion criteria (who cannot, and why), and state how participants will be selected, for example by consecutive attendance or by random sampling from a register. The sampling method decides whether the results can be generalised, so say what it is and what its limits are.
Define variables and how you will measure them
For every variable give its name, role, definition, measurement method and scale. Roles matter: the outcome is what you measure as the result, the exposure (or intervention) is what you compare, and confounders are other factors linked to both that must be measured or controlled.
| Variable | Role | Measurement | Scale |
|---|---|---|---|
| Anaemia | Outcome | Haemoglobin below 11.0 g/dL | Binary |
| Iron supplement use | Exposure | Self-report at interview, with the clinic card checked | Binary |
| Age | Confounder | Years at last birthday | Continuous |
| Parity | Confounder | Number of previous births | Integer |
Use validated instruments and standard definitions where they exist, and say which. A variable with no stated definition cannot be analysed consistently.
Plan the sample size
State how many participants you need and show the calculation: the formula, every input, and where each input came from, such as a pilot study or published estimates. Add an allowance for people who drop out or give incomplete data. For the example above, a single-proportion calculation with an expected prevalence of 40%, a precision of plus or minus 5 percentage points and 95% confidence gives 369 women, and allowing for 10% non-response raises this to 410. The sample size guide shows the working. Do not choose a number because it is convenient: a study that is too small cannot answer its question, and one that is too large exposes more people to the study than necessary.
Pre-specify the analysis
For each objective write the analysis in advance: the descriptive measures, the test or model, the measure of effect and its confidence interval, and how you will handle missing data. State the primary outcome and the primary analysis explicitly. If you later add analyses that were not planned, label them as exploratory when you report them. If you will adjust for confounders, name them and say how.
Plan data management and ethics
Describe how data are recorded, entered, checked and stored; who can see identifiable data; and how long they are kept. Use codes instead of names in analysis files and keep the key separately.
For ethics, state the committee that will review the protocol, how consent will be obtained and documented, the risks and benefits to participants, and how confidentiality will be protected. The Declaration of Helsinki sets out the ethical principles for research involving people [5], and your institution and country add their own requirements. Do not start recruiting before approval.
Plan the timeline and name the risks
List the milestones with dates: ethics approval, start of recruitment, end of data collection, analysis and report. Name who is responsible for each. Then list the main risks, such as slow recruitment or missing equipment, and what you will do if they occur. Reviewers trust a protocol that shows its authors have thought about what could go wrong.
Common mistakes
- A question too broad to answer with the planned design.
- Objectives, variables and analysis that do not match one another.
- A sample size taken from a convenient number rather than a calculation.
- Changing the primary outcome after looking at the data without saying so.
- No plan for missing data or for participants lost to follow-up.
- Vague inclusion criteria that two people would apply differently.
- Ethics treated as a final paragraph rather than a plan.
Check it before you submit
Read the protocol as a reviewer would. For every objective, find the variable, the measurement and the analysis; if any is missing, the protocol has a gap. Check the sample size against the primary objective. Ask your supervisor to review the whole protocol, and ask a statistician to review the analysis plan and sample size. Keep a dated copy of the approved version, and record any later change as a documented amendment.
Build it in ToProt
The protocol builder in ToProt follows these sections, saves as you type and shows which sections still need attention. You can work out the sample size with the sample size calculator, frame the question with the research question builder, and plan the analysis with the statistical analysis plan generator.
Limitations of this guide
This guide is a general introduction. Trial protocols have extra requirements covered by SPIRIT, and regulated studies must follow the rules of the regulator that applies. Always follow the template and rules of your own institution and ethics committee.
References
- 1.Chan A-W, Tetzlaff JM, Altman DG, et al. SPIRIT 2013 statement: defining standard protocol items for clinical trials. Ann Intern Med. 2013;158(3):200-207. doi:10.7326/0003-4819-158-3-201302050-00583
- 2.Chan A-W, Tetzlaff JM, Gotzsche PC, et al. SPIRIT 2013 explanation and elaboration: guidance for protocols of clinical trials. BMJ. 2013;346:e7586. doi:10.1136/bmj.e7586
- 3.von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-1457. doi:10.1016/S0140-6736(07)61602-X
- 4.Vandenbroucke JP, von Elm E, Altman DG, et al. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE): explanation and elaboration. PLoS Med. 2007;4(10):e297. doi:10.1371/journal.pmed.0040297
- 5.World Medical Association. World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects. JAMA. 2013;310(20):2191-2194. doi:10.1001/jama.2013.281053
This guide is general educational information. It is not a substitute for advice from your supervisor, a statistician or your ethics committee.