[Number needed to treat: a useful toolto evaluate the effectiveness of a treatment?]

This article reviews the clinical implications and limitations of the number needed to treat (NNT). Clinicians can quickly use this rather intuitive statistical value to assess the expected effectiveness of a treatment and explain it to patients. However, careful attention must be paid to the outcomes used in defining an NNT, as well as, the rate of the specific event in the population and the duration of observation. Conflicts of interest may also affect this easily manipulated statistical tool. Some often prescribed treatments have an NNT well above 20, implying an uncertain benefit for the patient, which emphasizes the need to carefully weigh the risk-benefit balance (NNT vs. NNH: number needed to harm) when prescribing. This review shows particularly low NNTs for anti-infectious agents compared to other drugs frequently used in medical practice.