Background All osteoarthritis (OA) guidelines recommend exercise as one of the core treatments for OA.1 However, it is unclear whether one exercise is better than another and for which outcome. Due to the limited evidence that compare different types of exercise, we undertook this network meta-analysis (NMA).
Objectives To determine the relative efficacy of different exercises for pain and self-reported function at (or nearest to) eight weeks.
Methods Nine electronic databases were searched for eligible randomised controlled trials (RCTs) that compared any types of exercise. The search was first performed in December 2015 and was updated in December 2017. Studies comparing exercise with usual care or with another exercise were included for this NMA. Common comparators such as usual care were used to network different types of exercise. Frequentist NMA was used to estimate the relative effect size (ES), i.e. standard mean difference and its 95% confidence interval (CI).2
Results 217 RCTs (n=20419) met the inclusion criteria. Of these, 89 trials (n=7070; 97 comparisons) were analysed for pain outcome (figure 1), whilst 87 trials (n=7039; 97 comparisons) were analysed for function. Mind-body exercise was the most effective for pain relief, closely followed by aerobic exercise (See the last column, table 1). While mind-body remained the best for improving function, strength and flexibility/skills exercise were better than aerobic exercise (See the last row, table 1). Single exercises were consistently better than mixed exercise.
Table 1 Effect size (95% confidence interval) between different exercises
Note Exercises are in order from the top left to the bottom right. Pain outcomes are at the top and functional outcomes are at the bottom. Numbers in each cell represent effect size (95% confidence interval) between higher comparator versus lower comparator.
Aerobic: walking, cycling; FlexSkills: flexibility exercise, neuromotor training, proprioceptive training; Strength: resistance training; Mind-body: Tai-chi, Yoga; Mixed Ex: multi-component exercises
Conclusions This exercise hierarchy and their relative efficacy provide a useful basis for patients and clinicians to select the most appropriate exercise whilst taking treatment goals (pain relief or functional improvement) and patient preference into consideration. The reason for the relative poor efficacy of mixed exercise warrants investigation as it contradicts current guidelines.
References  Fernandes L, et al. EULAR recommendations for the non-pharmacological core management of hip and knee osteoarthritis. Annals of the Rheumatic Diseases2013;72(7):1125–1135.
 Caldwell DM. An overview of conducting systematic reviews with network meta-analysis. Systematic Reviews2014;3(109):1–4.
Disclosure of Interest None declared
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.