The E-value calculator gives the minimum strength of unmeasured confounding that could explain away an observed risk ratio, odds ratio or hazard ratio, for the estimate and for its confidence limit.
How the calculation works
- Choose the type of estimate and enter it with its confidence limit closest to 1.
- Optionally enter the strengths of a hypothetical confounder.
- The calculator gives the E-value, the E-value of the limit, the bias factor and the adjusted estimate.
Example
A study finds a risk ratio of 1.8, with a lower confidence limit of 1.2. The E-value is 1.8 + √(1.8 × 0.8) = 3.000, so a confounder associated with both exposure and outcome by a risk ratio of 3 each could fully explain the result, but a weaker one could not. For the limit of 1.2, the E-value is 1.690. A hypothetical confounder with risk ratios of 2 and 2 gives a bias factor of 1.333 and would reduce the estimate to 1.350.
Why it matters
The calculator helps epidemiology students and researchers who interpret observational studies, reviewers who judge claims of causality, and readers of health news.
Good to know
The E-value does not say that confounding exists, only how strong it would have to be. Compare it with the strengths of known confounders in the field: if the E-value is larger than any measured confounder’s association, the result is harder to explain away. It is a sensitivity analysis, not a correction, and it does not cover selection bias or measurement error. The bias factor formula assumes a single binary confounder. Always report the E-value for the confidence limit as well as for the estimate, since the limit is what often matters for the claim.
Related tools
The relative risk calculator compares two risks, the attributable risk calculator measures the share due to exposure, the odds calculator converts between odds and probabilities and the Bayes theorem calculator updates beliefs.