Number needed to harm

Number needed to harm is the mirror image of number needed to treat, applied to unwanted effects. You calculate it as 1 divided by the absolute risk increase, the extra rate of harm the treatment causes. A large NNH is reassuring because harm is rare; a small NNH signals frequent harm. For example, if a drug raises the rate of a serious bleed from 1 percent to 3 percent, the absolute risk increase is 0.02 and the NNH is 50, meaning one extra bleed for every 50 people treated. When reading a study, compare NNH directly against NNT over the same time period and for outcomes of similar seriousness, since preventing a mild event is not worth causing a severe one at the same frequency. Watch for trials that are too short or too small to detect harms reliably, which can make NNH look larger, and therefore safer, than longer or bigger studies would show.