Most stalled research projects don't fail because of statistics or writing. They fail because the question was never tight enough to design a study around. Here's how to turn a clinical observation into a researchable question before you write a single page of protocol.
Start with the problem, not the design
A good research question begins with something you have seen: a gap in local practice, an inconsistency between guidelines, an outcome nobody seems to measure. Write that problem in plain words first. Only then ask what evidence would change what clinicians do.
At this stage, resist naming a study design. Design follows from the question, not the other way round, and picking a design early tends to bend the question to fit it.
Turn it into PICO
Structure the question into Population, Intervention, Comparison, and Outcome. For example, in a project on chronic low back pain:
| Element | Example |
|---|---|
| Population | Adults with chronic non-specific low back pain attending outpatient physiotherapy |
| Intervention | A supervised eight-week core-stabilisation programme |
| Comparison | Usual physiotherapy care |
| Outcome | Pain intensity on a numeric rating scale at eight weeks |
If you cannot fill in every part of PICO, the question is not ready for a protocol.
Stress-test with FINER
Once the question is structured, test it against the FINER criteria used in clinical research design. Ask whether it is:
- Feasible: enough eligible participants, time, skills and budget.
- Interesting: to you, and to the clinicians who will read the result.
- Novel: it confirms, refutes or extends what is already known.
- Ethical: it can pass review and protects participants.
- Relevant: it could change practice, policy or future research.
Feasibility is where most student and trainee projects fail. Estimate how many eligible patients you will realistically see in your data-collection window before you commit.
Write it as one sentence
Combine the four PICO parts into a single sentence. That sentence becomes the backbone of your protocol's introduction and objectives, and the fastest way to spot a question that isn't actually answerable yet. A short consultation at this stage costs far less than a redesign after data collection.
Common traps
Questions that are too broad to design a single study around. Comparisons that don't exist in your setting. Outcomes that can't actually be measured with the data you'll have access to. Catching these before you write the protocol saves months.