How to use the number needed to treat calculator
Find how many people must be treated to prevent one extra bad outcome, from the event rates in two groups. Enter event rate in control group, event rate in treatment group in the fields above; the absolute risk reduction, number needed to treat are computed instantly and plotted on the chart so you can see how the answer changes.
Common searches this calculator answers include “number needed to treat”, “NNT calculator”, “absolute risk reduction”, “calculate number needed to treat”, “find absolute risk reduction”.
Formula
Variables
- pc — Event rate in control group
- pt — Event rate in treatment group
Worked example
With event rate in control group = 0.2, event rate in treatment group = 0.15, the calculator gives absolute risk reduction ≈ 5 %; number needed to treat ≈ 20 people.
Frequently asked questions
What is the number needed to treat calculator used for?
Find how many people must be treated to prevent one extra bad outcome, from the event rates in two groups. It helps with questions such as “number needed to treat”, “NNT calculator”, “absolute risk reduction”.
What is the formula for the number needed to treat calculation?
Absolute risk reduction (%) = (pc-pt) × 100. Number needed to treat (people) = 1 / (pc-pt). Here pc is event rate in control group, pt is event rate in treatment group.
What inputs do I need for the number needed to treat calculator?
You need event rate in control group, event rate in treatment group. Each field is pre-filled with a typical example so you can see how the result responds as you edit the numbers.
How do I use the number needed to treat calculator?
Type values into the input fields. Results and the chart update instantly. Use Reset to restore the example values, Copy results to paste them elsewhere, or Share to copy a link that preserves your inputs.
Is the number needed to treat result exact?
Calculations use standard formulas and double-precision arithmetic, then display about seven significant figures. Real-world measurements carry uncertainty, so treat results as estimates unless your inputs are exact.