How to Write the Results Section

The Results section reports what you found, in the order of your objectives, without interpretation. It should tell the story of the data in text that points to tables and figures, and it should present effect sizes with measures of uncertainty.

Recommended order

1. Participants: numbers screened, included and analysed, reasons for exclusion; a flow diagram when relevant; baseline characteristics (Table 1).
2. Primary outcome: the effect estimate and its confidence interval, with the p-value if used.
3. Secondary outcomes.
4. Subgroup, sensitivity and exploratory analyses, labelled as such.
5. Harms or adverse events, where relevant.

Writing tips

• Match each result to the objective or outcome described in Methods.
• Do not repeat every number in tables; highlight the key ones.
• Report absolute numbers and percentages as well as relative measures.
• Report exact p-values and confidence intervals, and the test used.
• Keep interpretation (why, so what) for the Discussion.
• Use consistent units and decimal places.

Illustrative example

Invented for teaching. "Of 1,200 patients screened, 1,000 were included (Figure 1). Delirium occurred in 90 of 400 patients operated within 48 hours (22.5%) and 210 of 600 operated later (35.0%); the adjusted odds ratio was 0.65 (95% CI 0.49 to 0.86). Results were similar in sensitivity analyses (Table 3)." (Invented numbers.)

Reporting adverse events and harms

Harms deserve the same care as benefits. Report the number of participants with each important harm in each group, define how harms were identified, and avoid combining everything into a single "safety" statement. If no harm was observed, say how many people were exposed, because "no events" in a small group is compatible with a real risk. Present harms in a table if there are many categories.

Common pitfalls

• Listing only p-values.
• Reporting a result for an analysis that was not described in Methods.
• Discussing why in the Results.

Key takeaways

• Present, don't interpret.
• Effect size plus uncertainty beats a bare p-value.

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.