NUMBER NEEDED TO TREAT STATISTIC REPORTING FREQUENCY IN SYSTEMATIC REVIEW AND CLINICAL TRIAL ABSTRACTS ON THE PHYSIOTHERAPY EVIDENCE DATABASE [PEDRO]

D. Hilton11Deborah Hilton Statistics Online, Ashwood, AustraliaBackground: Hilton and colleagues published a manuscript titled; An under used yet easily understood statistic: the number needed to treat [NNT] [2006]. The author describes how to convert more commonly reported statistics [dichotomous outcomes] into the NNT [the mathematical inverse of the absolute risk reduction], using nine manuscripts from the physiotherapy evidence database [PEDro] [varied clinical disciplines] converting reported statistics to the NNT utilising the PEDro downloadable spreadsheet. This was presented at the Australian Physiotherapy Association Conference [Cairns 2007] and delegate comments predominantly were while physiotherapists knew of NNT, few utilised it in everyday clinical work when discussing with patient’s their likelihood of benefit. The NNT is easily interpretable, explicit and describes how many patients need to be treated with the new intervention in order for one additional patient to have the desired outcome compared with the number expected to have that outcome under the standard treatment or placebo.Purpose: To comprehend and understand how often there is reporting of NNT in systematic reviews and clinical trials listed on the PEDro database.Methods: A PEDro search was conducted in order to ascertain how many systematic reviews and clinical trials report outcomes with NNT. A simple search using the search words ‘number needed to treat’ identified 236 records [1 practice guideline, 98 systematic reviews and 137 clinical trials]. Results: When the abstracts were scoured to identify which records reported NNT the following was obtained. 142 records [60.1%] obtained reported NNT, 23 records did not report NNT [9.7%], while 70 of the records obtained stated that copyright release for this abstract has not been granted hence no details were available [29.6%]. 2 records reported number needed to prevent. In terms of the categorisation of reports whereby NNT was recorded in terms of clinical speciality, the breakdown was; musculoskeletal [eg spinal injury, after anterior cruciate repair, use of orthoses and splints] (36.1%); cardiothoracic / respiratory [eg neonatal care, adult care, ICU, asthma] (24.3%); rheumatic/ inflammatory diseases (3.5%); general pain syndromes (6.3%); women’s health [eg gynaecological, osteoporosis after menopause] (6.9%); men’s health (0.7%); neurological (4.9%); mental health [eg depression, quality of life, generalised anxiety disorder] (3.5%) and general exercise therapy [eg yoga, exercise for overweight or obesity, strength training, specific exercises to prevent falls in older adults, aquatic therapy] (8.3%). There were also several adhoc titles whereby NNT was reported (5.5%) including one clinical trial titled; Mobile telephone short message service reminders can reduce nonattendance in physical therapy outpatient clinics and another titled; Individually fitted sports shoes for overuse injuries among newspaper carriers while several were on injury prevention exercise programmes or physical activity promotion in primary care and/or diabetes prevention.Conclusion(s): NNT is often reported and utilised as a statistical means of indicating how likely the patient is to benefit from treatment in records listed on the PEDro website.Implications: Considering that NNT is often reported, physiotherapists should utilise this statistic, not just to understand themselves how likely a patient will benefit from treatment, but to use NNT when conveying likelihood information to patients.Funding, acknowledgements: No funding source. Keywords: probability, statistics, likelihoodTopic: Research methodology, knowledge translation & implementation scienceDid this work require ethics approval? NoInstitution: N/ACommittee: N/AReason: This is an analysis of a statistic reported in published work, there is no direct clinical contact with patients.All authors, affiliations and abstracts have been published as submitted.