01
What the evidence says
If risk falls from 2 in 100 to 1 in 100, the relative reduction is 50% and the absolute reduction is 1 percentage point. Over that period, about 100 similar people would need treatment to prevent one event if the estimate is causal and stable. If baseline risk were 20 in 100 and the same relative effect held, the absolute difference would be much larger.
Sources for this section: [1]
02
How to use the finding
Always record the numerator, denominator, outcome definition, comparison, and time. Hazard ratios and odds ratios are not automatically the same as risk ratios, especially when events are common or timing matters. Number needed to treat is derived from an absolute difference and inherits its uncertainty; it should never be reported without the duration and population.
Sources for this section: [1]
03
Limits and open questions
Benefits and harms may use different outcomes and time horizons. A composite can hide components moving in different directions, and competing risks can change interpretation. Apply a trial estimate only after asking whether the person’s baseline risk resembles the study population and whether implementation changes adherence, co-treatment, surveillance, cost, or exposure to adverse effects.
Sources for this section: [1]
Sources
- Understanding statistics: riskCochrane · 2022