Relative efficacy of different types of exercise for treatment of knee and hip osteoarthritis: protocol for network meta-analysis of randomised controlled trials

Siew-Li Goh, Monica S. M. Persson, Archan Bhattacharya, Michelle Hall, Michael Doherty, Weiya Zhang

'Exercise' is universally recommended as a core treatment for knee and hip osteoarthritis (OA). However, there are very few head-to-head comparative trials to determine the relative efficacy between different types of exercise. The aim of this study is to benchmark different types of exercises against each other through the use of a common comparator in a network meta-analysis of randomised controlled trials (RCTs). This study will include only RCTs published in peer-reviewed journals. A systematic search will be conducted in several electronic databases and other relevant online resources. No limitations are imposed on language or publication date. Participants must be explicitly identified by authors as having OA. Interventions that involved exercise or comparators in any form will be included. Pain is the primary outcome of interest; secondary outcomes will include function and quality of life measures. Quality assessment of studies will be based on the modified Cochrane's risk of bias assessment tool. At least two investigators will be involved throughout all stages of screening and data acquisition. Conflicts will be resolved through discussion. Conventional meta-analysis will be performed based on random effects model and network meta-analysis on a Bayesian model. Subgroup analysis will also be conducted based on study, patient and disease characteristics. This study will provide for the first time comprehensive research evidence for the relative efficacy of different exercise regimens for treatment of OA. We will use network meta-analysis of existing RCT data to answer this question. PROSPERO CRD42016033865.