Summary by John Baur, PT, DPT, OCS, FAAOMPT
This Cochrane review evaluates the effectiveness of exercise for adults with knee osteoarthritis (OA), a common condition associated with chronic pain, reduced physical function, and poorer quality of life. Exercise is recommended in clinical guidelines as a nonpharmacologic treatment, yet earlier evidence often grouped all control conditions together. The authors argue this can be misleading because exercise may seem more effective against usual care or no treatment than against attention controls or placebo interventions. Accordingly, this review compared exercise with three control approaches: attention control or placebo, no treatment/usual care/limited education, and exercise added to another co-intervention versus that co-intervention alone.
The review updated a 2015 Cochrane analysis by searching CENTRAL, MEDLINE, Embase, ClinicalTrials.gov, and the World Health Organization trials registry from May 2013 through January 4, 2024, without language restrictions. Only randomized controlled trials were included. Eligible studies involved adults with knee OA participating in exercise programs that varied in type, supervision, delivery, and duration. Primary outcomes were pain severity, self-reported physical function, and quality of life; secondary outcomes were treatment success, withdrawals, and adverse events. Outcomes were assessed immediately after treatment and, when available, up to or beyond 12 months after the intervention ended.
The review included 139 trials involving 12,468 participants. Thirty trials compared exercise with attention control or placebo, 60 compared exercise with no treatment, usual care, or limited education, and 49 examined exercise added to another conservative intervention such as weight loss, physical therapy, or education. Programs ranged from 2 to 104 weeks, and participants were 41 to 81 years old. Exercise interventions were heterogeneous, including strengthening, aerobic, mind-body, stretching, balance, and mixed programs delivered individually, in groups, in person, or remotely.
A major theme of the review is methodological weakness. Most trials were judged to be at unclear or high risk of bias. The biggest problems were performance and detection bias because participants knew whether they were exercising and outcomes were self-reported. More than half of studies also had concerns about selection bias or selective reporting. These weaknesses lowered certainty in the findings and may have exaggerated the benefits of exercise.
Compared with attention control or placebo, exercise produced modest short-term effects. Pain improved by 8.7 points on a 0 to 100 scale, physical function by 11.27 points, and quality of life showed little to no improvement. Exercise likely increased treatment success and did not clearly increase withdrawals or adverse events. However, this evidence was low to moderate certainty, and longer-term data were sparse.
Compared with no treatment, usual care, or limited education, exercise showed somewhat larger short-term benefits. Pain improved by 13.14 points, physical function by 12.53 points, and quality of life by 5.37 points. These gains were larger than those seen against attention control, reinforcing the importance of comparator type. Still, the review found little evidence that exercise improved treatment success, and adverse events may have increased. Benefits up to 12 months were present but smaller than immediate post-treatment effects.
When exercise was added to another co-intervention, it again improved outcomes. Pain improved by 10.43 points, physical function by 9.66 points, and quality of life by 4.22 points. Exercise also increased treatment success and slightly reduced withdrawals, although adverse events increased slightly. These effects were supported mostly by moderate-certainty evidence and were sustained up to 12 months, though the magnitude remained small.
An important interpretive feature of the review is its emphasis on clinical significance rather than statistical significance alone. The authors compared pooled effects with minimal important difference thresholds: 12 points for pain, 13 for function, and 15 for quality of life on 0 to 100 scales. In most analyses, confidence intervals either failed to reach these thresholds or crossed both clinically important and unimportant ranges. Sensitivity analyses that removed low-quality or underpowered trials generally reduced effect sizes further. The authors therefore conclude that exercise probably improves pain, function, and quality of life in the short term, but the real-world size of these benefits is uncertain.
Subgroup analyses and meta-regression found no meaningful differences among exercise types and no relationship between outcomes and either the total number of prescribed sessions or the ratio of live clinician contact between groups. Overall, the review supports exercise as a first-line treatment for knee OA, but it also cautions clinicians against overstating its effects. Exercise appears to help, especially when compared with doing little or nothing, yet its benefits may be modest and influenced by contextual factors, trial bias, and patient expectations. Future research should emphasize stronger trial design, clearer adverse-event reporting, and long-term follow-up.
Reference
Lawford BJ, Hall M, Hinman RS, et al. Exercise for osteoarthritis of the knee. Cochrane Database Syst Rev. 2024;(12):CD004376. doi:10.1002/14651858.CD004376.pub4