| dc.description.abstract | Purpose: Therapeutic exercise is recommended as a core treatment for
adults with knee and hip osteoarthritis (OA). However, its average effect
sizes for pain and physical function within randomised controlled trials
(RCTs) tend to be small to moderate compared to no exercise controls.
This may be due to insufficient targeting of exercise to specific subgroups (moderators), and/or that exercise programmes may not sufficiently address factors leading to improvement (mediators). This study
aims to identify moderators and mediators of the effect of therapeutic
exercise for reducing pain and improving physical function in people
with knee or hip OA. This could lead to better targeting and refinement
of future exercise interventions.
Methods: Systematic review update and individual participant data
(IPD) meta-analyses, with ongoing patient and public involvement
and engagement (PPIE) (PROSPERO: CRD42017054049). The search
strategy from a previous systematic review (Uthman et al 2013) that
identified 60 RCTs of exercise for people with knee or hip OA was rerun from March 2012 (previous search date) in 10 electronic databases. Identified titles, abstracts and subsequent full texts were
reviewed against inclusion/exclusion criteria by two independent
reviewers. In collaboration with the OA Trial Bank, leads of included
RCTs are being invited to share their IPD. All meta-analyses, apart
from mediation analyses, will use a two-stage approach, where estimates are obtained for each trial and then synthesised using a random effects model. Mediation analyses will use Causal Mediation
Analyses. All meta-analyses will be on an intention-to-treat principle
with summary estimates reported as standardised mean differences
with 95% confidence intervals. To identify potential moderators for
analyses, RCT leads (n¼24) and PPIE representatives (n¼4) who were
named co-authors on the STEER OA protocol manuscript (collaborators) were invited to participate in a “moderator ranking exercise”.
Initially, collaborators were invited to a face-to-face meeting at Keele
University where, through discussion, a list of 43 potential moderators of the effect of exercise was developed. Collaborators were then
contacted via email or post, and asked to rank which moderators they
believed were the 10 “most important”. RCT leads were also asked to
state their hypothesised direction of effect on pain and function
outcomes for each selected moderator. Frequency counts were completed to identify the 10 moderators most commonly rated as “most
important”. Based on previous systematic review findings (Runhaar et
al 2015), meta-analyses will explore the potential mediating effects of
muscle strength (for knee and hip OA), proprioception (for knee OA)
and extension deficits (for knee OA) in the association between
therapeutic exercise and pain and physical function.
Results: The systematic review update identified 3943 unique references, which reduced to 272 following title and abstract screening.
Remaining full texts were combined with those included in the
original review (n¼60) and nine identified from other sources, and
were screened against the inclusion/exclusion criteria. In total, 114
RCTs met our criteria and are included in the review. In total, 60 RCT
leads have agreed in principle to share IPD (approximately 8500
participants in total); 38 of these have signed data sharing agreements, and 28 datasets have been obtained to date. In total, 17 collaborators (13 RCT leads; 4 PPIE representatives) completed the
“moderator ranking exercise”. Overall, 11 subgroups were most frequently rated as being “most important” for moderating the effect of
either pain or function following therapeutic exercise. These were:
motivation to exercise; outcome expectations; pain severity; body
mass index (BMI); anxiety/depression; self-efficacy; lower limb
muscle strength; co-morbidity; age; instability (buckling); and baseline physical activity. These subgroups will therefore all be included
in the analyses for both pain and function outcomes. The hypothesised direction of effect for some potential moderators was variable
(see Table 1).
Conclusions: This is the first study, of this scale, to combine IPD from
existing RCTs of therapeutic exercise for hip and knee OA. This will
increase the statistical power to identify moderators and mediators of
the effect of exercise. The moderators rated as “most important” will
be explored in the IPD meta-analyses for both pain and function
outcomes. The variation in hypothesised direction of effect of some
potential moderators may reflect the lack of previous robust evidence
in this area. | en_US |
