The reciprocal of the reduction in absolute risk between treated and control groups in a clinical trial. It is interpreted as the number of patients who need to be treated to prevent one adverse event.