Patient-friendly summary
If you read nothing else
Bottom line
Exercising into pain offers a small, short-term pain advantage over pain-free exercise and appears not to be harmful, but the evidence base is small and the two approaches perform similarly over the medium and long term.
Moderate evidencePublished
Evidence hierarchy
Study participants
Adults from the general population with chronic musculoskeletal pain lasting more than 3 months (low back, shoulder, Achilles and plantar heel pain); fibromyalgia and other widespread pain disorders excluded
Full research — for clinicians and curious readers
Study Summary
This systematic review and meta-analysis pooled nine papers (from seven randomised controlled trials, 385 participants) to compare therapeutic exercises where pain is allowed or encouraged ("exercise into pain") against pain-free exercises for adults with chronic musculoskeletal pain lasting more than three months. In the short term (≤3 months), painful exercises showed a small but statistically significant benefit for pain (standardised mean difference −0.27, 95% CI −0.54 to −0.05 in the sensitivity analysis), with moderate-quality evidence. No significant difference between approaches was found for pain in the medium or long term, or for function and disability at any time point. The authors conclude that pain during therapeutic exercise need not be a barrier to successful outcomes, while cautioning that the evidence base is small and that further research is warranted.
Key Findings
| Finding | Detail | Impact |
|---|---|---|
| Small short-term pain benefit favouring painful exercises | Six trials with 385 participants showed a significant short-term benefit for exercises into pain over pain-free exercises, with a small effect size (SMD −0.28, 95% CI −0.49 to −0.08; I²=0%). The sensitivity analysis gave SMD −0.27 (95% CI −0.54 to −0.05) and was rated moderate-quality evidence. | High |
| No clear difference in the medium or long term | Long-term follow-up showed no statistical difference between approaches (effect size 0.01, 95% CI −0.39 to 0.41; I²=70%, very low quality). The medium-term pooled estimate favoured painful exercise (SMD −0.59, 95% CI −1.03 to −0.15) but was rated low quality and was not robust to sensitivity analysis. | High |
| No difference in function or disability | For function and disability outcomes in the short, medium and long term there was no significant difference between painful and pain-free exercise; clinically significant improvements were reported across nearly all interventions and control exercises with no clear superiority. | Medium |
| Pain need not be avoided during rehabilitation | The authors found no scientific basis for clinicians' common reluctance to let patients exercise into pain, concluding that pain during therapeutic exercise for chronic musculoskeletal pain need not be a barrier to successful outcomes. | High |
| Higher load may explain the benefit | Exercises into pain typically involved higher loads, resistance and dose than the comparison exercises, so the authors note it is unknown whether the short-term advantage reflects pain itself or a dose/load response; no trial recorded the actual level of pain patients experienced during exercise. | Medium |
Strengths
- Prospectively registered (PROSPERO) and conducted per PRISMA, with extensive multi-database and grey-literature searching
- Two independent reviewers for study selection, risk of bias and partial data extraction, with substantial inter-rater agreement on risk of bias (κ=0.76)
- Used validated appraisal tools (Cochrane risk of bias and GRADE) and pre-specified Cohen effect-size interpretations
- Low statistical heterogeneity at short-term follow-up (I²=0%) despite pooling different body regions, and sensitivity analyses to test robustness
Limitations
- Small evidence base: only 7 trials and 385 participants, with low participant numbers downgrading GRADE quality for most pooled estimates
- No trial blinded participants (100% high risk for performance bias), and some trials had high attrition (e.g., 51% dropout at 12 months in one trial)
- No included trial recorded the actual level of pain patients experienced during exercise, preventing analysis of mechanism
- Exercises into pain typically used higher loads/dose, so benefit may reflect load rather than pain; only English-language, US-spelling search terms were used and one reviewer screened titles/abstracts
- Long-term results showed high heterogeneity (I²=70%) and very low quality; fibromyalgia and other widespread pain disorders were excluded, limiting generalisability
Key Takeaways for Patients
What This Means for You
For long-standing musculoskeletal pain, exercises that temporarily reproduce or aggravate your pain appear to be safe and may help your pain slightly more than pain-free exercises in the short term (within about 3 months).
Over the medium and longer term, painful and pain-free exercise programmes worked about equally well, and both led to meaningful improvements.
Feeling some pain while doing prescribed rehabilitation exercises does not necessarily mean you are causing harm, and it need not stop you from continuing.
This conclusion is based on a small number of studies, so it should be discussed with your clinician and is not a reason to push into severe or worsening pain on your own.
Read the Full Paper
Access the complete peer-reviewed study from British Journal of Sports Medicine