Mechanisms of action of therapeutic exercise for knee and hip OA remain a black box phenomenon

Key points

1. 01Knee extension strength gain mediated only 2.3% of the pain benefit and 2.0% of the function benefit in knee OA
2. 02Proprioception and range of motion changes showed no statistically significant mediating effect in knee OA
3. 03Knee extension strength did not significantly mediate the exercise benefit in hip OA
4. 04Exercise still produced statistically significant improvements in pain and function across the included trials
5. 05The underlying mechanisms of therapeutic exercise for OA remain largely unknown, sometimes called a 'black box'

How it was conducted

Design

Individual participant data (IPD) mediation analysis using the counterfactual framework, drawing on RCTs from the STEER OA project and OA Trial Bank

Participants

1,407 participants from 12 RCTs (1,113 with knee OA from 11 RCTs; 294 with hip OA from 4 RCTs)

Eligible RCT pool

12 of 31 RCTs that shared IPD with STEER OA and measured at least one potential mediator

Mediators tested

Knee extension strength (all knee and hip OA RCTs), knee joint proprioception (3 knee OA RCTs, N=163), and knee joint range of motion (1 knee OA RCT, N=70)

Primary outcomes

Self-reported pain and physical function (rescaled 0-100), change from pre- to immediately post-intervention

Statistical approach

Counterfactual framework linear regression adjusted for age, BMI, sex, radiographic OA, baseline mediator and outcome; percentage mediated calculated via R mediation software

What they found

• Therapeutic exercise significantly reduced pain in knee OA (beta -9.0, 95% CI -10.8 to -7.1) and physical function (beta -6.7, 95% CI -8.2 to -5.1) versus non-exercise controls
• Knee extension strength mediated 2.3% (95% CI 0.4% to 6.0%) of the pain effect and 2.0% (95% CI 0.0% to 5.0%) of the function effect in knee OA
• Change in knee extension strength was significantly associated with pain change (beta -0.03, 95% CI -0.05 to -0.01) and function change (beta -0.02, 95% CI -0.04 to -0.00) in knee OA
• Exercise significantly improved knee extension strength (beta 8.0, 95% CI 2.8 to 13.1 for pain model; beta 8.0, 95% CI 2.7 to 13.3 for function model) in knee OA
• Proprioception showed no significant association with pain (beta 0.00, 95% CI -0.06 to 0.06) or function (beta -0.01, 95% CI -0.06 to 0.04) change in knee OA
• Range of motion showed no significant association with pain (beta -0.2, 95% CI -0.9 to 0.6) or function (beta -0.2, 95% CI -0.9 to 0.5) change; exercise did not significantly change ROM (beta 0.7, 95% CI -2.1 to 3.5)
• In hip OA, exercise significantly improved pain (beta -4.8, 95% CI -8.0 to -1.6) and function (beta -3.1, 95% CI -6.1 to -0.2), but knee extension strength change was not a significant mediator for pain (beta -0.01, 95% CI -0.04 to 0.03) and showed only a borderline association with function (beta -0.03, 95% CI -0.07 to -0.00) with no significant exercise effect on strength change

Limitations

• Only 12 of 31 eligible RCTs shared IPD, and most did not measure any potential mediator, limiting statistical power and representativeness
• Hip OA was substantially underrepresented, with only 4 RCTs and 294 participants
• Most RCTs assessed mediator changes at the same time as outcome changes rather than before, violating strict causality principles
• Heterogeneous measurement protocols for muscle strength across RCTs may have introduced noise and reduced the detectable mediating effect

Why it matters

For patients

Exercise reliably reduces pain and improves function in knee and hip OA, but the physical changes traditionally assumed to drive that improvement, such as getting stronger or more flexible, explain almost none of it, so you should not be discouraged if your strength gains seem modest.

For clinicians

Clinicians should not assume that optimising muscle strength, proprioception, or ROM gains is the key to better OA exercise outcomes; contextual, psychological, and possibly anti-inflammatory factors warrant investigation as potential targets.

For readers

This is the largest IPD mediation study of its kind and demonstrates a major gap in mechanistic understanding of exercise therapy for OA, highlighting the need for future trials to measure psychological, social, and inflammatory mediators alongside biomechanical ones.