.jpg){kind=link}
- Published
- Updated
A confidence interval is more than a line indicating statistical significance. Read it alongside clinically relevant boundaries when assessing estimation precision and the range of effect sizes compatible with the data under the stated analytical assumptions.
Define the question and assumptions
First identify the effect measure and its direction. A ratio and a difference have different null values. Explain how the interval was calculated and which assumptions it uses. If a clinical importance threshold is applied, justify it independently of results and distinguish it from the statistical null boundary.
Hypothetical study example
In a hypothetical trial, the mean difference favours one arm but the interval includes both small and important effects. The team avoids labelling the finding as certain benefit or certain absence of effect. It presents the estimate, uncertainty and additional information needed for a decision together.
Design review checks
- Confirm the effect measure, units, direction and relevant null value.
- Examine whether the interval allows materially different clinical interpretations.
- Report the calculation method and population included in the corresponding analysis.
Record the decision and its limits
Accepted interpretation is not reduced to one p-value and does not describe the interval as the range of outcomes for every future participant. Wide intervals may signal limited information; narrow intervals do not remove systematic bias. Discuss measurement error, missing follow-up and design limitations alongside the numerical estimate.
Frequently asked questions
- Does including the null mean there is no effect?
- No. Several effect sizes may remain compatible with the data.
- Does a narrow interval guarantee accuracy?
- No. Precision and freedom from bias are different properties.