Research methods guide
How to report confidence intervals and P values
Report the effect, comparison, units, and confidence interval for the main result. When a journal requests a P value, identify the test or model and report the value or threshold.
These are synthetic contrasts defined as intervention minus control. Identify scale, unit, null and any clinically relevant threshold before reporting.
Interpreting intervals and P values together
Verify model, estimate and test before comparing the interval and P value.
| Item | Worked example | Verification |
|---|---|---|
| Continuous outcome | Adjusted six-month difference −3.4 mmHg (95% CI −5.9 to −0.9). | State direction. With a prespecified 4 mmHg threshold, the interval includes smaller and larger benefits. |
| Ratio outcome | 30-day readmission RR 0.82 (95% CI 0.58–1.17). | Null is 1. The interval allows benefit and harm; report modeled absolute effects where available. |
| Precision and presentation | Give effect, bounds, unit, contrast and model. Use journal precision for very small P values, for example P<0.001. | Identify adjusted versus crude estimates and one- versus two-sided tests. |
Apply the method to your materials
Verify calculation and rounding
Check bounds, direction, confidence level and interval method. Investigate apparent discrepancies using unrounded values and multiplicity rules.
Reuse the same output across the manuscript
Carry units and references to the abstract and graphics and mark the null clearly on forest plots.
Questions that arise in practice
Is there a 95% probability the true value is in this interval?
For a conventional frequentist interval, confidence refers to long-run coverage of the procedure. Report the estimate, interval and justified interpretation.
Verified sources
The official sources below were checked on 2026-11-12. Access conditions and update dates are rechecked before publication.
RigorMed workspace
Start with your research materials
Bring your research question, available materials, and the steps you want to check.
