Exercise for chronic low back pain: what evidence supports
Table of contents
- Why chronic low back pain matters
- First step: the red flags
- Natural course and prognosis
- What the WHO and the reviews say
- Which exercise to choose
- How much exercise to prescribe
- Keeping up the practice is the challenge
- How NEXO helps train this reasoning
- Frequently asked questions
- Take-home messages
- References
A 47-year-old woman, an administrative assistant, presents with low back pain for five months. The pain started after carrying boxes during a move, improved a little and came back. She spends the day sitting, sleeps poorly and avoids the gym because she thinks "her spine is worn out". She brings an MRI with mild disc protrusions and asks whether she needs rest, a brace or surgery.
The consultation has two tasks, in this order. The first is to look for signs of serious disease, which are rare, but cannot be missed. The second, after ruling out that possibility, is to build an active treatment in which exercise is the base. The World Health Organization guideline for chronic primary low back pain analyzed 37 non-surgical interventions and issued 24 recommendations (Briggs, 2025).
This text goes through those two tasks: how to screen for red flags with what the evidence actually supports, what the reviews show about exercise, which modality to choose, how much exercise to prescribe and how to help the patient keep up the practice.
Key points
- Low back pain affected 619 million people worldwide in 2020, with a projection of 843 million in 2050 (GBD 2021 Low Back Pain Collaborators, 2023).
- In a large cohort of people who sought care for low back pain, fewer than 3 in 100 people received a red-flag diagnosis within 90 days (Rhon, 2026).
- Many red flags from the guidelines change the probability of fracture or cancer very little; a history of cancer is the most useful for malignancy (Downie, 2013).
- Structured exercise compared with no intervention reduced pain and functional limitation, with moderate certainty, in a review done for the WHO (Verville, 2023).
- Pilates, motor control and strengthening rank well in network meta-analyses, with low-certainty evidence (Owen, 2020).
- Sessions of 20 to 30 minutes, 3 to 5 times a week, had the largest effect in stabilization exercises (Mueller, 2020).
Summary
- Why chronic low back pain matters
- First step: the red flags
- The natural course and prognosis
- What the WHO and the reviews say
- Which exercise to choose
- How much exercise to prescribe
- Keeping up the practice is the challenge
- How NEXO helps train this reasoning
- Frequently asked questions
- Take-home messages
Why chronic low back pain matters
Low back pain is the leading cause of years lived with disability worldwide (GBD 2021 Low Back Pain Collaborators, 2023). The Global Burden of Disease study estimated 619 million people with low back pain in 2020 (GBD 2021 Low Back Pain Collaborators, 2023).
A problem that will grow
The same analysis projects 843 million people with low back pain in 2050 (GBD 2021 Low Back Pain Collaborators, 2023). Age-standardized rates fell slightly between 1990 and 2020, but the absolute number of people affected keeps rising (GBD 2021 Low Back Pain Collaborators, 2023).
About 39 of every 100 years lived with disability from low back pain were attributed to occupational factors, smoking and high body mass index (GBD 2021 Low Back Pain Collaborators, 2023). Part of the burden, therefore, is linked to factors that the person and the health system can change.
What chronic primary low back pain is
The WHO guideline addresses chronic primary low back pain, with or without pain radiating to the leg, that lasts more than three months and for which no specific disease is identified as the cause (Briggs, 2025). It is the picture of most adults who come to the office with persistent low back pain.
This label requires a prior step: looking for and ruling out specific causes. Only after this screening does active treatment take the center of the consultation.
MRI rarely decides
Disc disease, protrusions and degenerative changes appear frequently in people who feel pain and in people who feel nothing. The finding becomes a label, and the label pulls toward rest, fear of moving and procedures. It is the same mechanism described in the text about the drivers of overdiagnosis.
The reasoning of using clinical rules to decide when imaging helps also appears in the text about how the cervical spine imaging rules work.
First step: the red flags
Serious disease behind low back pain is rare, but it exists. Screening must be done carefully and with what the evidence supports, without turning every consultation into a cascade of tests.
How rare serious disease is
A study of 1,214,604 people seen for low back pain in a large United States military health system tracked red flag diagnoses in the following 90 days (Rhon, 2026). Fewer than 3 in 100 patients received any of these diagnoses (Rhon, 2026).
Infection was the most frequent red flag diagnosis, in just over 1 in 100 patients (Rhon, 2026). Cauda equina syndrome, vertebral fracture, abdominal aortic aneurysm and ankylosing spondylitis each came in at about 2 in 10,000 patients (Rhon, 2026).
Red flag diagnoses were more common in those first seen in the emergency department, in women and with increasing age (Rhon, 2026). The care setting therefore changes the pre-test probability.
What each red flag is worth
A systematic review published in the BMJ assessed 53 red flags for fracture and cancer across 14 studies (Downie, 2013). Many signs recommended in guidelines barely changed the probability of serious disease or had accuracy that was never tested (Downie, 2013).
For vertebral fracture, prolonged corticosteroid use raised the post-test probability to about 33 in 100 (Downie, 2013). The presence of contusion or abrasion brought the probability to about 62 in 100 (Downie, 2013).
Major trauma raised the probability of fracture to about 11 in 100, and older age to about 9 in 100 (Downie, 2013). When several fracture signs were present at the same time, the probability reached about 90 in 100 (Downie, 2013).
For spinal cancer, a history of cancer was the most useful sign, with a post-test probability of about 33 in 100 (Downie, 2013). The authors conclude that only a small subset of signs has evidence of usefulness (Downie, 2013).
How to use this in the consultation
Ask actively about a history of cancer, prolonged corticosteroid use, relevant trauma, fever, unexplained weight loss, progressive neurological deficit and changes in bladder or bowel function. Examine strength, sensation and reflexes in the lower limbs.
A single weak sign on its own rarely justifies tests. Strong signs, or several signs together, call for targeted investigation. This approach protects against what the text on cascades of care describes.
In places with few resources, an organized history and physical examination do almost all the work, as the text on the clinical history as the first instrument shows.
Natural course and prognosis
Explaining the prognosis is part of treatment. The patient who believes the spine is "finished" tends to protect it too much and move less.
What happens in one year
An Australian cohort followed 406 people whose acute low back pain persisted for three months (Costa, 2009). At 12 months, about 42 in 100 were free of pain and about 41 in 100 had fully recovered (Costa, 2009).
Among those who still had pain-related disability when they entered the study, about 47 in 100 recovered within 12 months (Costa, 2009). More than a third of people with recent-onset chronic low back pain recover within a year (Costa, 2009).
Who takes longer
Previous absence from work due to low back pain, severe disability or pain at the onset of chronicity and lower schooling were associated with slower recovery (Costa, 2009). The perception of a high risk of persistent pain also delayed recovery (Costa, 2009).
That last factor matters because it can change through conversation. Pessimistic beliefs about the spine are a target of treatment, just as much as muscle strength.
What the WHO and the reviews say
The 2023 WHO guideline was the organization's first for primary chronic low back pain in adults (Briggs, 2025). It relied on systematic reviews of randomized trials published up to mid-2022 (Briggs, 2025).
Conditional recommendations
All the guideline's recommendations were classified as conditional, because the certainty of the evidence ranged from moderate to very low (Briggs, 2025). For 12 interventions, the WHO issued no recommendation due to lack of evidence or an uncertain balance between benefits and harms (Briggs, 2025).
The recommendations also apply to older people, with one exception: the use of nonsteroidal anti-inflammatory drugs received a favorable recommendation that does not extend to older adults (Briggs, 2025). This detail matters when reviewing the prescription of someone who already takes an anti-inflammatory every day.
Structured exercise programs
The systematic review carried out to inform the guideline included 13 trials with low or uncertain risk of bias (Verville, 2023). Compared with no intervention, exercise reduced pain with a standardized mean difference of 0.33, with moderate certainty (Verville, 2023).
Functional limitation also fell, with a standardized mean difference of 0.31 in adults and older adults, with moderate certainty (Verville, 2023). The authors conclude that structured exercise programs probably reduce pain and functional limitation in primary chronic low back pain (Verville, 2023).
The broader picture
A 2025 umbrella review gathered 70 systematic reviews on exercise and physical activity in low back pain (Comachio, 2025). In 35 of 36 reviews on treatment, Pilates, motor control, mixed exercise, tai chi, water exercise and yoga brought a small benefit for pain and disability, especially in the short term (Comachio, 2025).
To prevent new episodes, 6 of 10 reviews with meta-analysis indicated a small benefit of general exercise and leisure-time physical activity (Comachio, 2025). The certainty of this evidence was low to moderate (Comachio, 2025).
The size of the effect in perspective
A sequential analysis of the trials in the Cochrane review showed that the superiority of exercise over usual care or no treatment was already established in 2004, after four or five trials (Chamberlain, 2025). Another 43 trials with 2,626 participants were published after that (Chamberlain, 2025).
When trials with fewer than 70 participants were excluded, the reduction in pain was 8.8 points on a scale from 0 to 100 (Chamberlain, 2025). The authors suggest that small trials inflated the effects and that the benefit may fall below what is considered clinically important for many patients (Chamberlain, 2025).
The honest reading is that exercise helps, with a modest effect, and remains the foundation because it is safe, cheap to offer and improves other health outcomes. Selling exercise as a cure tends to generate frustration.
Which exercise to choose
The question every patient asks is which exercise works best. The short answer is that several modalities work and the difference between them is small.
What network meta-analyses show
A network meta-analysis of 89 studies with 5,578 patients compared exercise modalities in nonspecific chronic low back pain (Owen, 2020). Pilates had the highest probability of being the best for pain, with a standardized mean difference of 1.86 compared with the no-treatment control (Owen, 2020).
For physical function, strengthening and motor control stabilization came out on top, with differences of 1.14 and 1.13 (Owen, 2020). For mental health, strengthening and aerobic exercise had the best positions (Owen, 2020).
Stretching and McKenzie exercises were similar to the control for pain and function in this analysis (Owen, 2020). The quality of the evidence was low, and the journal published an expression of concern about the article (Owen, 2020).
Convergence between the reviews
Seven network meta-analyses gathered in the umbrella review favored motor control and Pilates for reducing pain, with low certainty (Comachio, 2025). In the review carried out for the WHO, aerobic exercise and Pilates were associated with less pain compared with no intervention, with very low certainty evidence (Verville, 2023).
The practical message is that it is worth choosing among options with some support, such as Pilates, motor control, strengthening, aerobic exercise and yoga, according to the person's preference, access and history.
Yoga at a distance
A randomized trial tested 12 weekly 60-minute hatha yoga classes, streamed live over the internet for adults with chronic low back pain (Tankha, 2024). At 12 weeks, pain fell 1.5 points more than in the waiting-list group, on a scale from 0 to 10 (Tankha, 2024).
Function measured by the Roland Morris questionnaire improved 2.8 points more in the yoga group (Tankha, 2024). At 24 weeks, the reduction in pain reached 2.3 points (Tankha, 2024).
Use of any analgesic in the previous week was about 21 in every 100 participants lower in the yoga group at 12 weeks (Tankha, 2024). No serious adverse events were reported (Tankha, 2024).
The evidence side by side
| Study | Design and population | Main result |
|---|---|---|
| GBD, 2023 | Global burden of low back pain in 204 countries (GBD 2021 Low Back Pain Collaborators, 2023) | 619 million people in 2020 and 843 million projected for 2050 (GBD 2021 Low Back Pain Collaborators, 2023) |
| Rhon, 2026 | Cohort of 1,214,604 people seen for low back pain (Rhon, 2026) | Fewer than 3 in every 100 with a red-flag diagnosis within 90 days (Rhon, 2026) |
| Downie, 2013 | Review of 14 studies on 53 red flags (Downie, 2013) | History of cancer raised the probability of malignancy to about 33 in every 100 (Downie, 2013) |
| Costa, 2009 | Cohort of 406 people with recent-onset chronic low back pain (Costa, 2009) | About 41 in every 100 recovered within 12 months (Costa, 2009) |
| Verville, 2023 | Review of 13 trials carried out for the WHO (Verville, 2023) | Exercise reduced pain and functional limitation, moderate certainty (Verville, 2023) |
| Owen, 2020 | Network meta-analysis of 89 studies, 5,578 patients (Owen, 2020) | Pilates at the top for pain; strengthening and motor control for function (Owen, 2020) |
| Mueller, 2020 | Meta-regression of 50 stabilization studies (Mueller, 2020) | Greater effect with 20 to 30 minutes, 3 to 5 times a week (Mueller, 2020) |
| Tankha, 2024 | Trial of live yoga over the internet (Tankha, 2024) | Pain 1.5 points lower at 12 weeks (Tankha, 2024) |
How much exercise to prescribe
Knowing the modality helps little if the amount of training is insufficient or excessive. The evidence on volume is scarcer, but it exists.
What the meta-regression shows
A meta-regression of 50 studies with 2,786 patients assessed the relationship between training volume and the effect of stabilization exercises (Mueller, 2020). The programs lasted on average 7 weeks, with about 3 sessions per week of about 45 minutes (Mueller, 2020).
Sessions of 20 to 30 minutes had the greatest effect on pain and disability, with moderate-quality evidence (Mueller, 2020). Training 3 to 5 times a week brought the greatest effect, with low-quality evidence (Mueller, 2020).
How to translate this into the consultation
A reasonable starting point is to propose short sessions of 20 to 30 minutes, several times a week, instead of a long and rare session. Progression should follow the person's tolerance, with gradual increases in load and complexity.
Some discomfort during and shortly after exercise is expected. A useful rule is to accept mild pain that returns to the usual level within a day and to adjust the program if the worsening lasts longer.
Safety
Fewer than one third of the reviews gathered in the umbrella review reported adverse events (Comachio, 2025). The events described were muscle pain after exercise and temporary increase in pain, especially in yoga studies (Comachio, 2025).
Serious adverse events were absent, but the authors warn that harms are still poorly studied (Comachio, 2025). In the review carried out for the WHO, temporary increase in mild pain appeared with mixed exercise and with yoga (Verville, 2023).
Keeping up the practice is the challenge
Most of the effect depends on the person continuing to exercise. Adherence is usually the weakest link in treatment.
Barriers and enablers
A review of qualitative studies gathered 23 articles with 677 participants with chronic low back pain (Gilanyi, 2024). Four major themes influenced adherence: the relationship between exercise, pain and the body; psychological factors; social factors; and external factors (Gilanyi, 2024).
The same factor can act as a barrier or as a help, depending on the person and the moment, before, during or after exercise (Gilanyi, 2024). Fear of making the pain worse, for example, keeps some patients away, while noticing relief with training motivates others.
What to do in practice
Explain that pain is not a synonym for damage and that the spine is a strong structure. Choose with the person an activity that fits the routine and that they enjoy. Agree on small goals and review the program at follow-up visits.
Motivational interviewing techniques such as the OARS skills help turn the recommendation into the patient's own decision. The reasoning behind prescribing an exercise the person can keep up also appears in the text on how to prescribe exercise in depression.
Medications and other measures
Medications can help during phases of worsening, but chronic treatment supported only by pills carries risks. Reviewing the continuous use of analgesics is an opportunity to decide with the patient what to stop.
In chronic pain, knowing the difference between physical dependence, tolerance and substance use disorder helps before thinking about opioids. For those who need mobility support, the WHO recommends offering quality assistive products after a person-centered assessment (Briggs, 2025).
Back to the case
For the 47-year-old woman, the history and the examination show no red flags. The conversation explains that protrusions are common and that the pain has a good chance of improving. The proposal is to start Pilates or strengthening in short sessions, three to five times a week, add walks and schedule a follow-up in a few weeks to adjust the program and reinforce adherence.
How NEXO helps train this reasoning
In NEXO, the case of the day is fictional, with a real clinical presentation and up to six progressive clues, and you propose the diagnosis by looking for the code in ICD-10.
The limits of using the game are described on the NEXO economy page. If you research the teaching of clinical reasoning, the page for researchers explains how to get in touch. Professionals who write or review cases find information under curation and under creators, and institutions can learn about the proposal on the partnerships page. For the most frequent reasons for consultation, see the most common codes in primary care.
Frequently asked questions
Does every chronic low back pain need an MRI?
Rarely. Serious disease appeared in fewer than 3 in every 100 people seen for low back pain in a large cohort (Rhon, 2026). Imaging is reserved for those with strong red flags or neurological deficit.
Which red flags matter most?
For fracture, prolonged corticosteroid use, severe trauma, contusion and several signs together were the most useful (Downie, 2013). For cancer, a history of cancer was the most informative sign (Downie, 2013).
Does exercise really work?
It works, with a modest effect. In the review done for the WHO, structured exercise reduced pain and functional limitation with moderate certainty (Verville, 2023).
What is the best type of exercise?
Pilates, motor control and strengthening ranked well in the network meta-analyses, but with low certainty (Owen, 2020). The best modality is the one the person can keep up.
How long and how often?
In stabilization exercises, sessions of 20 to 30 minutes, 3 to 5 times a week, had the greatest effect (Mueller, 2020).
Can exercise make the pain worse?
There may be muscle soreness and a temporary increase in pain, especially with yoga, and no serious events were identified in the reviews (Comachio, 2025).
Does chronic low back pain get better on its own?
In some cases, yes. About 41 in every 100 people with recent-onset chronic low back pain recovered completely within 12 months (Costa, 2009).
Take-home messages
- Start with red flag screening, using the few signs with evidence of usefulness.
- Serious disease behind low back pain is rare, especially outside the emergency department.
- Structured exercise is the basis of treatment for primary chronic low back pain, with a modest and safe effect.
- Pilates, motor control, strengthening, aerobic exercise and yoga are reasonable options; the person's preference decides.
- Short, frequent sessions seem to work better than long, rare sessions.
- Explaining the prognosis and dispelling pessimistic beliefs is part of the treatment.
- Adherence is the weakest link; small goals, enjoyable activities and follow-ups help.
Play the case of the day on NEXO.
References
- Briggs AM, Sumi Y, Banerjee A. The World Health Organization guideline for non-surgical management of chronic primary low back pain in adults: implications for equitable care and strengthening health systems globally. Glob Health Res Policy. 2025. PMID 40624581. DOI 10.1186/s41256-025-00426-w.
- GBD 2021 Low Back Pain Collaborators. Global, regional, and national burden of low back pain, 1990-2020, its attributable risk factors, and projections to 2050: a systematic analysis of the Global Burden of Disease Study 2021. Lancet Rheumatol. 2023. PMID 37273833. DOI 10.1016/S2665-9913(23)00098-X.
- Rhon DI, Parsons NA, Boeth R, Yuan X. Prevalence of non-spinal pathology and red flag diagnoses associated with care seeking for low back pain in the Military Health System. Sci Rep. 2026. PMID 42604939. DOI 10.1038/s41598-026-51658-w.
- Downie A, Williams CM, Henschke N, et al. Red flags to screen for malignancy and fracture in patients with low back pain: systematic review. BMJ. 2013. PMID 24335669. DOI 10.1136/bmj.f7095.
- Costa LdaC, Maher CG, McAuley JH, et al. Prognosis for patients with chronic low back pain: inception cohort study. BMJ. 2009. PMID 19808766. DOI 10.1136/bmj.b3829.
- Verville L, Ogilvie R, Hincapié CA, et al. Systematic Review to Inform a World Health Organization (WHO) Clinical Practice Guideline: Benefits and Harms of Structured Exercise Programs for Chronic Primary Low Back Pain in Adults. J Occup Rehabil. 2023. PMID 37991647. DOI 10.1007/s10926-023-10124-4.
- Comachio J, Beckenkamp PR, Ho EK, et al. Benefits and harms of exercise therapy and physical activity for low back pain: An umbrella review. J Sport Health Sci. 2025. PMID 40180212. DOI 10.1016/j.jshs.2025.101038.
- Chamberlain F, Elliott L, Yarahmadi A, Underwood M. Research waste in randomised trials of exercise treatments for chronic low back pain: trial sequential and cumulative meta-analyses by publication date and size. BMJ Open. 2025. PMID 40132821. DOI 10.1136/bmjopen-2024-087787.
- Owen PJ, Miller CT, Mundell NL, et al. Which specific modes of exercise training are most effective for treating low back pain? Network meta-analysis. Br J Sports Med. 2020. PMID 31666220. DOI 10.1136/bjsports-2019-100886.
- Tankha H, Gaskins D, Shallcross A, et al. Effectiveness of Virtual Yoga for Chronic Low Back Pain: A Randomized Clinical Trial. JAMA Netw Open. 2024. PMID 39485352. DOI 10.1001/jamanetworkopen.2024.42339.
- Mueller J, Niederer D. Dose-response-relationship of stabilisation exercises in patients with chronic non-specific low back pain: a systematic review with meta-regression. Sci Rep. 2020. PMID 33037280. DOI 10.1038/s41598-020-73954-9.
- Gilanyi YL, Shah B, Cashin AG, et al. Barriers and enablers to exercise adherence in people with nonspecific chronic low back pain: a systematic review of qualitative evidence. Pain. 2024. PMID 38635470. DOI 10.1097/j.pain.0000000000003234.