How to Turn a Clinical Observation Into a Research Question

A clinical observation becomes a research question when you convert "I noticed this" into a specific, testable question with a defined population, exposure or intervention, comparison and outcome. The observation alone is a starting point, not evidence.

Step by step

1. Describe the observation neutrally. What exactly did you see, how often, in whom?
2. Ask why it matters. Is it harmful, costly, or unexplained?
3. Check the literature to see what is known.
4. Name the elements: population, exposure/intervention, comparator, outcome (PICO), and timeframe.
5. Choose the verb carefully. "Is associated with" suits observational designs; "reduces" implies an intervention and comparison.
6. Check feasibility and ethics. Can the question be answered with data you can access?

Illustrative example

Invented for teaching. Observation: "Several patients on proton pump inhibitors seemed to develop low magnesium." Question: "Among adults on long-term proton pump inhibitors (P), is duration of use (E) compared with non-use (C) associated with hypomagnesaemia (O) on routine blood tests?" Design: retrospective cohort or cross-sectional analysis of laboratory records. Note that the original impression may be biased by what you happened to notice, which is why you test it.

Testing the idea with a quick look at the data

Before writing a protocol, count how many patients have the exposure and the outcome in the records you could use. A rough two-by-two count (exposed or not, outcome or not) shows whether the pattern you noticed is visible at all, and whether the numbers are large enough to study. Treat this look as a feasibility check only; the formal analysis should be planned in advance to avoid fitting a hypothesis to the data you have already seen.

Common pitfalls

• Treating a memorable case as a trend (availability bias).
• Writing a question that implies a cause without a design that can test it.
• Not defining the outcome measurement.

Key takeaways

• Observation generates a hypothesis; study design tests it.
• Be specific about who, what, compared with what, and measured how.
• Validate the idea against the literature and available data.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.