Physiotherapy 92 (2006) 240–246 An under-used yet easily understood statistic: the number needed to treat (NNT) Deborah J. Hilton a,∗ , Christopher M. Reid b , Jennifer Paratz c a Baker Heart Research Institute, P.O. Box 6492, St. Kilda Road Central, Melbourne, Vic. 3004, Australia b NHMRC Centre of Clinical Research Excellence, Department of Epidemiology & Preventive Medicine, Monash University, Alfred Hospital, Commercial Rd, Melbourne Vic. 3004, Australia c Department of Anaesthesiology & Intensive Care, University of Queensland, Herston, Qld, 4029, Australia Abstract Objectives Interpretation of the research literature can be difficult, yet it is essential in order for physiotherapists to convey the details of risks and benefits associated with therapeutic interventions to patients. The number needed to treat (NNT) statistic is one such statistic that is easily interpretable. Using several examples from the physiotherapy literature representing a range of conditions, this article demonstrates how to convert more commonly used statistics into the NNT statistic. The purpose of this is to show how the NNT can help clinicians to converse with patients to convey details about the likelihood of benefit with treatment and/or the likelihood of risk, in order that a decision may be made with respect to therapy. Design The Australian Physiotherapy Evidence Database (PEDro) was searched in order to locate a selection of physiotherapy research articles that reported various dichotomous outcomes that could be converted to the NNT statistic for the purpose of this analysis. Results The NNT statistic for nine studies with a PEDro score ≥6 was calculated using the Internet-based downloadable spreadsheet on the PEDro website. For six studies, the NNT point estimates ranged from 2 to 4 (95% confidence interval 1–10). One study had a NNT of 8, while two other studies produced number needed to harm values. Conclusion The NNT can be calculated quickly and efficiently using Internet-based calculators and/or other decision-making tools, and may be an alternative that provides readily interpretable information to assist in conveying the likely benefits (and/or risks) of treatment to patients. © 2006 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved. Keywords: Evidence-based medicine; Physical therapy; Number needed to treat (NNT) Introduction Evidence-based medicine (EBM) with philosophical ori- gins in the mid-19th Century and earlier [1] is now an essential element of healthcare decision-making. EBM is useful for researchers, clinicians, public health practition- ers, managers, planners and the general public. The Euro- pean Region of the World Confederation for Physical Ther- apy has defined evidence-based practice as a commitment to using the best-available evidence to inform decision- making about the care of individuals that involves integrat- ing practitioners’ individual professional judgement with ∗ Correspondence author. Address: 1 Aweta Street, Ashwood, Melbourne, Vic. 3147, Australia. E-mail address: dj hilton@telstra.com (D.J. Hilton). evidence gained through systematic research [2]. Phys- iotherapists are increasingly accepting of this approach that has been termed ‘evidence-based physical therapy’ [3]. Evidence can be graded using hierarchical guidelines such as the Oxford Centre for Evidence-based Medicine levels of evidence [4] or the National Health and Medical Research Council designation of levels of evidence [5]. Both of these guidelines stipulate the suitability of research design consid- erations in relation to the clinical question being focused on aetiology, diagnosis, prognosis or treatment or another clini- cal question such as economic analyses. The Physiotherapy Evidence Database (PEDro) [6], an ini- tiative of the Centre of Evidence-Based Physiotherapy, at the University of Sydney (Australia), collates research literature relevant to physiotherapy practice. 0031-9406/$ – see front matter © 2006 Chartered Society of Physiotherapy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.physio.2006.06.003