Risk Calculator: Relative Risk, ARR, NNT and Odds Ratio
Enter event counts for an exposed group and a control group (or enter event rates directly) to calculate relative risk, absolute risk reduction, relative risk reduction, number needed to treat, and odds ratio, all with 95% confidence intervals. The results update as you type and a worked-steps panel shows every formula with your actual numbers substituted in.
What this calculator computes
This tool computes five core risk statistics from a 2x2 contingency table or from raw event rates: Relative Risk (RR), Absolute Risk Reduction (ARR), Relative Risk Reduction (RRR), Number Needed to Treat (NNT), and Odds Ratio (OR). Each metric answers a different question. RR tells you how many times more likely an event is in the exposed group compared to the control group. ARR captures the real-world magnitude of that difference in percentage points. RRR expresses ARR as a fraction of the baseline risk, making it easy to compare across conditions. NNT inverts the ARR to show how many people you must treat (or expose) to see one additional event prevented. OR is used in case-control studies and is always further from 1 than RR, so mixing the two overstates effect size. All estimates come with a confidence interval and a two-tailed p-value based on the normal approximation for the log-transformed statistic.
How to fill in the 2x2 table
A standard 2x2 table has four cells. Cell a is the number of people in the exposed (treatment) group who experienced the outcome. Cell b is the number in the exposed group who did NOT experience it. Cell c is the number in the control group who experienced the outcome. Cell d is the number in the control group who did NOT experience it. If any cell is zero, the calculator adds 0.5 to all cells (Haldane-Anscombe continuity correction) to avoid division by zero. Alternatively, switch to "rates" mode and enter the percentage of each group that experienced the outcome directly - useful when you already have summary statistics but not the raw counts.
Relative Risk vs Odds Ratio: which to use
Relative Risk is the natural measure for cohort studies and randomised controlled trials, where you follow people forward in time and observe who develops the outcome. The OR is preferred in case-control studies, where you start with outcomes and look backwards at exposures, because the design does not allow you to estimate absolute rates. When the event is rare (under about 10%), OR closely approximates RR, so the distinction matters less. When the event is common, OR diverges from RR and always looks more extreme, so confusing the two inflates the apparent effect. This calculator reports both so you can choose the appropriate one for your study design.
Confidence intervals and statistical significance
A 95% confidence interval (CI) means that if the study were repeated many times under the same conditions, 95% of the calculated intervals would contain the true value. A CI for RR that straddles 1.0 (or for OR that straddles 1.0) indicates the result is consistent with no effect at that confidence level. The p-value is calculated from the z-statistic z = ln(RR) / SE(ln RR) and converted to a two-tailed probability using the standard normal distribution. Statistical significance (p < 0.05) does not imply clinical or practical importance: a tiny, irrelevant difference can be highly significant in a large study. Always pair the p-value with the absolute risk difference and NNT to judge practical importance.
Interpreting Relative Risk and Odds Ratio
| RR or OR value | Interpretation | Direction |
|---|---|---|
| < 0.5 | Strong protective effect | Benefit |
| 0.5 - 0.79 | Moderate protective effect | Benefit |
| 0.8 - 0.99 | Weak protective effect | Benefit |
| 1.00 | No difference | Neutral |
| 1.01 - 1.24 | Weak increased risk | Harm |
| 1.25 - 1.99 | Moderate increased risk | Harm |
| >= 2.0 | Strong increased risk | Harm |
Common thresholds used in epidemiology to characterise the strength and direction of an association.
Frequently asked questions
What is a Relative Risk of 2?
An RR of 2 means the exposed group had twice the risk of experiencing the outcome compared to the control group. If the control event rate was 10%, the exposed rate was approximately 20%. Whether that doubling is clinically important depends on the absolute numbers: going from 0.1% to 0.2% (ARR = 0.1%) has very different practical consequences than going from 10% to 20% (ARR = 10%).
What does a Relative Risk below 1 mean?
An RR below 1 indicates a protective association: the exposed group experienced fewer events than the control group. An RR of 0.6, for example, means the exposed group had 40% lower risk than the control group. In clinical trials this often reflects a treatment benefit, with the treatment group having fewer adverse events than the placebo group.
Why is the Number Needed to Treat important?
NNT translates the abstract risk difference into a concrete number of patients. A treatment that cuts event rate from 20% to 15% has an ARR of 5% and an NNT of 20: you need to treat 20 patients to prevent one event. NNT makes it easier to weigh benefits against costs and side effects. NNT below 10 is generally considered highly effective; above 100 suggests a modest benefit that may not justify widespread use depending on the condition.
When should I use Odds Ratio instead of Relative Risk?
Use OR when your study design is a case-control study, where you recruited people based on whether they already had the outcome and then looked back at exposure. In a case-control design you cannot calculate RR directly because the proportion of cases to controls is set by the researcher, not by the actual disease frequency in the population. OR is valid for cohort studies too, but RR is more intuitive there. When the outcome is rare (below about 10%), OR approximates RR closely.
What does a confidence interval crossing 1 mean for RR?
If the 95% CI for RR includes 1 (for example, 0.85 to 1.20), the result is not statistically significant at the 5% level: the data are consistent with no effect. This typically corresponds to a p-value above 0.05. It does not prove there is no effect; it means the sample size was insufficient to rule one out, or the true effect is very small. Always report CIs alongside p-values so readers can see the range of plausible effect sizes.
What is the continuity correction and when is it applied?
When any of the four cells in the 2x2 table is zero, the formulas for RR, OR, and their standard errors are undefined because of division by zero. The Haldane-Anscombe correction adds 0.5 to all four cells before computing. This is a pragmatic fix that slightly biases estimates toward the null (RR toward 1) but is far better than producing an error. The calculator applies it automatically and you can see which cells were adjusted in the steps panel.