ToolsConductScience tool
Cohort & Case-ControlFree in-browser calculator

OR / RR / NNT Calculator.

Compute odds ratio, relative risk, absolute risk reduction, and NNT from a 2×2 table. Study-design-aware with OR≠RR warnings. Data never leaves your browser.

PrivateData stays in your browser
LiveNo sign-up required
Open methodFormula shown on this page
CitableMethods and citation included

Calculator

Results update in place

Try it out

Load example OR/RR/NNT data to see the full workflow

Enter events and non-events for exposed (treatment) and unexposed (control) groups.

Events
Non-events
Exposed
Unexposed

When to use

  • Compute odds ratio, relative risk, and NNT from a 2x2 contingency table
  • Evaluate treatment effects in cohort studies or randomized controlled trials
  • Calculate odds ratios for case-control studies where RR is not estimable
  • Determine the number needed to treat (NNT) or harm (NNH) for clinical decision-making
  • Generate a manuscript-ready methods paragraph with effect measures and CIs

Do not use for

  • Diagnostic accuracy analysis (sensitivity, specificity) — use the Diagnostic Test Calculator
  • Continuous outcome data (means, not counts) — use the General Stats Calculator for t-tests
  • Survival / time-to-event data — use a Kaplan-Meier or Cox regression tool

OR overestimates RR when events are common

When the event rate exceeds 10%, the odds ratio substantially overestimates the relative risk. A common mistake is interpreting OR as if it were RR. Always check the event rate and report the appropriate measure for your study design.

RR is invalid in case-control studies

Case-control designs sample on outcome, not exposure, so you cannot estimate true incidence rates. Only the odds ratio is valid. Reporting RR from a case-control study is a fundamental methodological error.

NNT confidence intervals can include infinity

When the absolute risk reduction CI crosses zero, the NNT CI passes through infinity — meaning the treatment might have no effect or even cause harm. Report the full CI including the direction change (NNT to NNH).

Zero cells require the Haldane-Anscombe correction

A zero cell makes OR or RR undefined. Adding 0.5 to all four cells (Haldane-Anscombe correction) produces finite estimates with valid confidence intervals. This is standard practice in meta-analysis.

1

Method

OR is computed as (a×d)/(b×c) with log-scale normal approximation CIs. RR uses risk ratios with log-scale CIs. ARR is the difference of proportions with normal approximation CIs. NNT = 1/|ARR| with CI bounds derived from ARR CI inversion. The Haldane-Anscombe correction (0.5 added to all cells) is applied when any cell is zero. Study design selection gates which measures are displayed. All calculations run client-side in the browser.

2

Published

Published 2026-04-05. Calculations support planning and documentation; verify procurement decisions against manufacturer specifications or institutional SOPs.

3

How to cite

How to Cite

ConductScience OR/RR/NNT Calculator. ConductScience, Inc. 2026. Available at: https://conductscience.com/tools/or-rr-nnt-calculator

Altman DG. Confidence intervals for the number needed to treat. BMJ. 1998;317(7168):1309-1312. doi:10.1136/bmj.317.7168.1309

Effect Measure Fundamentals

Effect measures quantify the strength and direction of an association between exposure and outcome:

• OR = (a×d)/(b×c) — odds of event in exposed vs unexposed • RR = risk_exposed / risk_unexposed — probability ratio • ARR = risk_exposed − risk_unexposed — absolute difference • RRR = ARR / risk_unexposed — proportional reduction • NNT = 1 / |ARR| — patients to treat for one benefit/harm

OR and RR measure relative effects (how many times more/less likely). ARR and NNT measure absolute effects (how much difference in practice). Both perspectives are needed for clinical decision-making.

Frequently asked

What is the difference between odds ratio and relative risk?
Odds ratio (OR) compares the odds of an event between groups: OR = (a×d)/(b×c). Relative risk (RR) compares event probabilities: RR = (a/(a+b))/(c/(c+d)). OR is always valid for both cohort and case-control designs. RR is only valid for cohort/RCT designs where you can estimate true incidence. When the event rate is low (<10%), OR ≈\approx RR. At higher event rates, OR overestimates RR.
What is NNT?
Number Needed to Treat (NNT) is the reciprocal of the absolute risk reduction: NNT = 1/ARR. It answers: "How many patients must be treated with the experimental intervention for one additional patient to benefit?" NNT = 10 means you treat 10 patients to prevent one event. When the exposure increases risk, the reciprocal is called NNH (Number Needed to Harm).
When should I use OR vs RR?
Use RR for cohort studies and RCTs — it is directly interpretable as a ratio of risks. Use OR for case-control studies — it is the only valid measure because you cannot estimate true incidence from a case-control design. In meta-analyses, OR is often preferred because it has better mathematical properties (symmetry, no boundaries issues). Always report the measure appropriate to your study design.
What happens when a cell is zero?
A zero cell makes the OR or RR undefined (0 or ∞). The Haldane-Anscombe correction adds 0.5 to all four cells, producing finite estimates with valid confidence intervals. This correction is standard in meta-analysis. The tool applies it automatically when enabled.
How are confidence intervals computed?
OR and RR confidence intervals use the log-scale normal approximation: ln(OR) ±\pm z ×\times SE(ln(OR)), then exponentiated. ARR uses a normal approximation to the difference of proportions. NNT CIs are the reciprocals of the ARR CI bounds (which may include infinity if the ARR CI crosses zero).
325
Free tools
1,200+
Institutions
100%
Client-side
0
Uploads required