Research methods guide
How to extract an odds ratio and confidence interval
When extracting an odds ratio and 95% confidence interval, also record the comparator, reference category, adjustment model, unit, analysis set, and source table. One outcome can have several estimates.
Suppose a results table contains a crude OR and two adjusted models. Select the model specified in the review protocol and extract its unit, direction, and interval together. Values are synthetic.
A row another extractor can reproduce
The worked example uses Table 2, model 3, with exposure measured per 10-unit increase.
| Item | Worked example | Verification |
|---|---|---|
| Location and model | Table 2, model 3, adjusted for age, sex, and site. | Verify footnotes and Methods. Record why this model was selected over model 2. |
| Estimate and direction | OR 1.20 per 10 units, 95% CI 0.98–1.47. | Preserve unit and event. For the reverse contrast, invert the estimate and swap interval endpoints, about 0.83 (0.68–1.02). |
| Analysis set | 428 complete cases, including 36 events. | Check the sample-size column or appendix and distinguish it from initial enrollment. |
Apply the method to your materials
Apply the extraction rule
Record study ID, outcome time point, exposure, and comparator before copying estimates. If several models qualify, use the prespecified hierarchy and retain alternatives separately.
Document transformations separately
If the table reports log(OR), preserve the original coefficient and exponentiated value. Record reported precision, interval, and any conversion for a second extractor to reproduce.
Questions that arise in practice
Can I reconstruct the CI from a P value alone?
Look for the original interval in tables, supplements, or author-provided results. Any reconstruction depends on the test, scale, inputs, and assumptions.
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.
