MODERATORS OF THE EFFECT OF THERAPEUTIC EXERCISE FOR KNEE AND/OR HIP OSTEOARTHRITIS: A SYSTEMATIC REVIEW AND INDIVIDUAL PARTICIPANT DATA META-ANALYSIS

M. Holden¹, M. Hattle¹, J. Runhaar², R. Riley¹, E. Healey¹, J. Quicke^3,1, D. van der Windt⁴, K. Dziedzic⁴, M. van Middelkoop², D. Burke¹, N. Corp¹, A. Legha¹, S. Bierma-Zeinstra², N.E. Foster^5,6,1, O.A. Trial Bank Exercise Collaborative⁷

¹Keele University, School of Medicine, Primary Care Centre Versus Arthritis, Keele, United Kingdom, ²Erasmus MC University, Medical Center, Rotterdam, Netherlands, ³Chartered Society of Physiotherapy, Research, London, United Kingdom, ⁴Keele University, Primary Care Centre Versus Arthritis, Keele, United Kingdom, ⁵The University of Queensland, Faculty of Health and Behavioural Sciences, Brisbane, Australia, ⁶Metro North Hospital and Health Service, STARS (Surgical Treatment and Rehabilitation Service), Brisbane, Australia, ⁷The OA Trial Bank Exercise Collaborative, School of Medicine, Primary Care Centre Versus Arthritis, Keele, United Kingdom

Background: Although therapeutic exercise is a recommended core treatment for osteoarthritis (OA), the average effects for pain and physical function seen in randomised controlled trials (RCTs) are small to moderate compared to non-exercise controls. This may be due to individual variability in response to exercise.

Purpose: We aimed to identify individual-level moderators of the effect of exercise on pain and function in people with knee and/or hip OA.

Methods: Systematic review and individual participant data (IPD) meta-analysis. To identify RCTs that compared exercise to non-exercise controls, a previous systematic search was updated to February 2019 in multiple electronic databases. Two independent reviewers screened titles, abstracts and full texts. Data were extracted into tables and risk of bias was assessed using the Cochrane Collaboration’s tool (version 1.0). In collaboration with the international OA Trial Bank, identified RCT leads were contacted and IPD requested. Potential moderators (pain severity, physical function, age, body mass index, physical activity, arthritis self-efficacy, mental wellbeing, co-morbidity, muscle strength (quadriceps), educational attainment, pain duration, radiographic joint structure) were explored within IPD meta-analyses to determine whether they were associated with short- (12 weeks), medium- (6 months) or long-term (12 months) effects of exercise on pain and function, compared to non-exercise controls. Overall intervention effects were also summarised. All IPD meta-analyses used a two-stage approach, where estimates were obtained from a longitudinal model for each RCT separately and then synthesised using a random-effects multivariate meta-analysis accounting for correlation across time-points. All analyses were on intention-to-treat principle, with summary meta-analyses estimates reported as mean differences for a standardised 0 to 100 scale with 95% confidence intervals.

Results: IPD were obtained from 37 RCTs and following data checking, IPD from 31 RCTs (n=4241 participants) were included in analyses. Most RCTs included knee OA (n=18) and tested varied exercise interventions versus varied non-exercise controls. Summary meta-analysis results showed that on average, compared to non-exercise controls, exercise reduced pain and improved physical function in the short-, medium-, and long-term, although the magnitude of effect was small and of questionable clinical importance in the medium- and long-term. There was evidence that baseline pain and physical function moderated the effect of exercise for pain and physical function outcomes. Those with a higher baseline pain score and physical function score (poorer physical function) benefited most, with evidence most certain in the short-term (12-weeks). None of the other factors tested moderated the effect of exercise on pain or function.

Conclusions: Of 12 potential moderators of exercise for knee and/or hip OA, two showed a differential response. Targeting exercise to those with higher levels of OA pain and disability may therefore be of merit. There was strong evidence of a small, positive overall effect of exercise on pain and function compared to non-exercise controls. Whilst this was a large international IPD meta-analysis, limitations include potential differences between RCTs that did/did not share IPD, possible lack of power when exploring some moderators, and large heterogeneity in included RCTs.

Implications: Targeting exercise to those with higher levels of OA pain and disability may be of merit.

Funding acknowledgements: Chartered Society of Physiotherapy Charitable Trust; National Institute for Health Research School of Primary Care Research; NIHR Senior Investigator Award

Keywords:
osteoarthritis
exercise
individual patient data meta-analysis

Topics:
Musculoskeletal: lower limb
Orthopaedics
Pain & pain management

Did this work require ethics approval? No

Reason: Ethical approval was not required as no new data were collected