TY - JOUR T1 - NNTs and NNHs: handle with care JF - British Journal of General Practice JO - Br J Gen Pract SP - 133 LP - 133 DO - 10.3399/bjgp17X689797 VL - 67 IS - 656 AU - Adrian A Root AU - Liam Smeeth Y1 - 2017/03/01 UR - http://bjgp.org/content/67/656/133.abstract N2 - Best medical practice should be informed by research evidence. However, translating study findings into clinical decisions is not straightforward. Ratio measures of effect such as the risk ratio are most commonly reported, but tell us little about absolute effects. For example, taking a statin reduces your risk of vascular events by about 20% a year across a wide range of baseline vascular risk.1 If your initial 10-year risk of such an event was 1% then taking a statin would reduce this risk by 0.2%, whereas if your initial risk was 20%, the reduction would be 4%. This shows how the absolute risk difference can be more clinically informative.The number needed to treat (NNT) statistic was introduced over 30 years ago and has gained popularity as a useful measure to help clinicians understand research findings.2 Abstract probabilities are converted into a more tangible quantity: number of patients treated. In the above statins example, the NNT for 10 years to avert one vascular event would be … ER -