Patient-friendly summary
If you read nothing else
Bottom line
Exercise produces moderate short-term improvements in pain, function, and performance and a smaller benefit for quality of life in knee and hip osteoarthritis versus usual care, peaking around 2 months and diminishing to no better than usual care by 9 to 18 months.
Moderate evidencePublished
Evidence hierarchy
Study participants
Adults with knee or hip osteoarthritis who had not undergone joint replacement surgery
Study Summary
This systematic review and meta-analysis of 77 randomised controlled trials (6472 participants) compared exercise-only interventions with usual care in people with knee and hip osteoarthritis. At or near 8 weeks, exercise produced statistically significant moderate benefits for pain (effect size 0.56), function (0.50), and objective performance (0.46), and a smaller but significant benefit for quality of life (0.21). The effects appeared to peak around 2 months and then gradually declined, becoming no better than usual care by 9 to 18 months depending on the outcome. Study-level analyses suggested that younger participants, those with knee (rather than hip) OA, and those not awaiting joint replacement surgery tended to report greater pain relief, though the authors caution these determinants may reflect ecological bias and need confirmation with individual patient data.
Key Findings
| Finding | Detail | Impact |
|---|---|---|
| Exercise gave a moderate, statistically significant benefit for pain versus usual care at ~8 weeks | Effect size 0.56 (95% CI 0.44-0.68), which the authors classify as a moderate effect using Cohen's thresholds (0.2 small, 0.5 moderate, 0.8 large). | High |
| Function and objective performance also improved moderately | Function effect size 0.50 (95% CI 0.38-0.63) and performance 0.46 (95% CI 0.35-0.57) at or nearest to 8 weeks. | High |
| Quality of life benefit was smaller but still significant | Quality of life effect size 0.21 (95% CI 0.11-0.31); the benefit was not influenced by whether a generic or disease-specific instrument was used. | Medium |
| Benefits peaked around 2 months and then diminished over time | Across all outcomes effects appeared to peak around 2 months, then gradually decreased and became no better than usual care by 9 to 18 months depending on the outcome. | High |
| Greater pain relief was associated with younger age, knee OA, and not awaiting joint replacement | On multivariate meta-regression (significance threshold P <= 0.10), younger age (mean <60 years), knee OA, and participants not on a waiting list for joint replacement remained significant determinants of better pain relief. These are study-level findings that may carry ecological bias. | Medium |
| Results were relatively robust in sensitivity analyses | Restricting to more homogeneous RCTs (I-squared <30%) still showed significant benefits: pain 0.50 (0.43-0.58), function 0.43 (0.35-0.51), performance 0.32 (0.25-0.39), and QoL 0.18 (0.09-0.27). | Medium |
Effect size 0.56 (95% CI 0.44-0.68), which the authors classify as a moderate effect using Cohen's thresholds (0.2 small, 0.5 moderate, 0.8 large).
Function effect size 0.50 (95% CI 0.38-0.63) and performance 0.46 (95% CI 0.35-0.57) at or nearest to 8 weeks.
Quality of life effect size 0.21 (95% CI 0.11-0.31); the benefit was not influenced by whether a generic or disease-specific instrument was used.
Across all outcomes effects appeared to peak around 2 months, then gradually decreased and became no better than usual care by 9 to 18 months depending on the outcome.
On multivariate meta-regression (significance threshold P <= 0.10), younger age (mean <60 years), knee OA, and participants not on a waiting list for joint replacement remained significant determinants of better pain relief. These are study-level findings that may carry ecological bias.
Restricting to more homogeneous RCTs (I-squared <30%) still showed significant benefits: pain 0.50 (0.43-0.58), function 0.43 (0.35-0.51), performance 0.32 (0.25-0.39), and QoL 0.18 (0.09-0.27).
Strengths
- Large evidence base: 77 RCTs with 6472 participants, more studies and more outcomes than many previous meta-analyses
- Searched 9 databases from inception to December 2017 with no language restriction
- Included all types of exercise and standardised the comparator to usual care, isolating the exercise effect
- Examined time-dependent effects across multiple follow-up intervals
- Robustness checked with multiple sensitivity analyses (end-point vs change scores, imputed SDs, single index knee, translated publications, and removal of heterogeneous studies)
Limitations
- Observational nature of meta-analysis; results depend on accuracy and completeness of the primary reports, with missing covariate data (e.g., mean age, sex distribution) limiting adjustment for multiple comparisons
- Study-level (ecological) analyses mean the identified determinants may not hold at the individual-patient level and need confirmation with individual patient data
- Generally poor quality of exercise trials: exercise cannot be blinded, and many trials had high risk of bias in randomisation, allocation concealment, reporting, and small sample size
- Egger's test indicated a small study effect / publication bias for all outcomes except QoL
- No unified definition of 'usual care' across trials and substantial heterogeneity between studies
- A higher significance threshold (P <= 0.10) was used to avoid missing potential determinants, so the determinant findings are hypothesis-generating
Key Takeaways for Patients
What This Means for You
- 01For knee and hip osteoarthritis, exercise can meaningfully reduce pain and improve function, movement performance, and quality of life compared with no new treatment, with benefits typically seen within about 8 weeks.
- 02The gains tend to be strongest around 2 months and then slowly fade, so keeping up with exercise over the long term appears important to maintain benefit.
- 03People who are younger, have knee (rather than hip) osteoarthritis, and are not waiting for joint replacement surgery may get more pain relief from exercise, though this comes from study-level patterns and is not yet proven for individuals.
- 04Exercise is generally considered safe compared with medications and is recommended internationally as a core treatment for osteoarthritis.
Read the Full Paper
Access the complete peer-reviewed study from Annals of Physical and Rehabilitation Medicine
View Full Study