Two respected organizations have published verdicts on yoga for long-term low back pain, and they sound like they are describing different treatments.
The American College of Physicians, in 2017, told clinicians and patients to start with non-drug options for chronic low back pain and listed yoga among them, under what its grading system calls a strong recommendation. Cochrane, in November 2022, pooled 21 trials and 2,223 participants and reported that compared with no exercise, yoga produced improvements in back function and pain that were small and, by the review’s own pre-set yardsticks, clinically unimportant. Against ordinary back-focused exercise such as physical therapy, it found probably little or no difference.
Both statements are accurate. The physicians’ group was answering “what should we try first?” and the reviewers were answering “how big is the effect?” Those questions have different answers, and a reader handed only one of them has been shortchanged. Notice also what sits inside the physicians’ own sentence: the parenthesis grading the evidence for yoga is the low-quality one, not the moderate-quality one.
This article holds both findings up at once, which is the only way to report them fairly, and then turns to the part most coverage skips: yoga has the best-documented harm record of the three movement practices in this section, and the most useful finding in it is about a posture almost every beginner is taught.
A strong recommendation resting on low-quality evidence
Guideline grading systems separate two things that readers tend to merge: how firmly the committee advises something, and how good the underlying studies are. A panel can advise firmly on weak evidence when the option is cheap, safe and the alternatives are worse. That is roughly what happened here.
The physicians’ guideline groups its chronic-back-pain options and tags each group with an evidence rating. Exercise, multidisciplinary rehabilitation, acupuncture and mindfulness-based stress reduction carry the moderate-quality tag. Yoga sits in the next bracket with tai chi, motor control exercise, progressive relaxation, biofeedback, low-level laser therapy, cognitive behavioral therapy and spinal manipulation, and that bracket is labelled low-quality evidence. The recommendation covering all of them is strong.
Two further details matter for how much weight to give it. The searches behind the guideline ran through April 2015 with an update through November 2016, so its evidence base is nine years old, and a search of the medical literature for a replacement found none. And for acute or recently started back pain, yoga is not on the recommended list at all; that shorter list is superficial heat, massage, acupuncture and spinal manipulation.
What happens when somebody measures the size of it
The Cochrane review is the measurement, and its details are worth knowing. Twenty-one trials from the United States, India, the United Kingdom, Croatia, Germany, Sweden and Turkey. Most participants were women in their forties or fifties. Most trials used Iyengar, Hatha or Viniyoga. The last search was August 31, 2021.
Against no exercise at three months, back-specific function improved by 1.69 points on a 24-point questionnaire where the reviewers had set 5 points as the smallest difference a patient would notice. Pain improved by 4.53 points on a 100-point scale where that threshold was 15. Those are the two headline numbers, and both fall well short of their own marks. The function result was rated low certainty, the pain result moderate.
Not everything was null. Four trials asked participants directly whether their back pain had improved or resolved, and the proportion saying yes was more than twice as high with yoga, a risk ratio of 2.33, at low certainty. Quality of life improved slightly on both the physical and mental subscales. Depression scores barely moved.
Then the comparison that decides the practical question. Set against back-focused exercise such as physical therapy, function differed by 0.38 points, with a confidence interval straddling zero, at moderate certainty. Pain differed by 2.68 points in the direction favouring the other exercise, at very low certainty. The proportion of people reporting improvement was essentially identical. Cochrane’s own framing of what to do with that is refreshingly plain: the choice between yoga, other exercise or nothing may come down to availability, cost and what the patient or provider prefers.
Two structural facts complete the picture. Every trial was at high risk of performance and detection bias, because nobody can be blinded to whether they are doing yoga and the outcomes are self-reported. And no trial has ever compared yoga with a fake version of yoga. The companion article on acupuncture takes up the broader question of what sham controls can and cannot settle; here the point is narrower. The reviewers end by saying that, given the bias, a blinded comparison would be unlikely to show a clinically important benefit.
It is also worth watching this review change its own mind. The 2017 edition, with 12 trials and 1,080 participants, described small to moderate improvements in back-related function. The 2022 edition, same team, nine more trials, describes small and clinically unimportant improvements. The characterisation got less favourable as the evidence base grew, which is the opposite of what usually happens to a popular therapy. For what it is worth, one of the trials in the earlier edition was funded by a yoga institution.
Government framing sat between the two all along. The clinical digest on yoga from the National Center for Complementary and Integrative Health reports that effects on back pain were similar to exercise, citing a 2018 report from the federal Agency for Healthcare Research and Quality covering eight trials and 1,466 participants, and notes that the earlier Cochrane review’s pain effects did not reach the minimum level of clinical importance. One warning about that page: its footer reads “Site Last Updated: October 19, 2026,” which is a site-wide stamp. The content itself is dated February 2020 and its studies end around 2019. Citing the footer would present six-year-old material as this month’s.
The injuries, the denominator, and one surprising posture
Yoga sends people to emergency rooms. A 2016 study in the Orthopaedic Journal of Sports Medicine used the federal injury surveillance system that samples hospital emergency departments and counted 29,590 yoga-related injuries treated between 2001 and 2014. Nearly half, 46.6 percent, involved the trunk. Sprains and strains accounted for 45.0 percent of diagnoses. The injury rate rose across those years.
The total means little without a denominator, and the same paper supplies one. In 2014 the rate was 11.9 per 100,000 participants aged 18 to 44, 17.7 for ages 45 to 64, and 57.9 for people 65 and over. For a physical activity, those are low numbers.
The important finding is neither the total nor the reassurance. It is that the rate among people 65 and over runs roughly five times the rate in younger adults, and was rising. That is exactly the group most often steered toward gentle or chair yoga as the safe option.
A second line of evidence shows which events turn up when something goes seriously wrong. A 2013 systematic review in PLOS One collected the published case reports and case series through February 2013: 76 cases in all. Ten of those people had a pre-existing condition, most often glaucoma or thinning bone. The practices named most often were pranayama breathing, hatha yoga and Bikram yoga; the techniques named most often were headstand, shoulder stand, lotus position and forceful breathing. Twenty-seven cases involved the musculoskeletal system, 14 the nervous system and 9 the eyes. Fifteen people recovered fully, 9 partially, 1 not at all, and 1 died.
Handle that collection for what it is. Case reports have no population behind them and are published precisely because they are unusual, so they cannot support any statement about how often anything happens. What they are good for is identifying which practices and which pre-existing conditions keep reappearing.
One widely repeated claim does not appear in that record in usable form. Popular coverage often asserts a specific link between yoga postures and a torn artery in the neck. That claim could not be traced to indexed medical literature for this article, and the case-report collection does not itemise its 14 nervous-system events. No rate can be stated, and none should be inferred from silence in either direction.
The harm the trials measured is more mundane and more useful. In the Cochrane review, yoga raised the risk of an adverse event compared with no exercise, at roughly 43 per 1,000 against 9 per 1,000 over six to twelve months, and the events were primarily increased back pain. Compared against other exercise, there was no excess at all: about 84 per 1,000 with yoga, 91 per 1,000 without. Doing yoga is riskier than doing nothing and no riskier than doing something else.
The posture the standard warning misses
Advice for people with glaucoma has settled for years on a single instruction: avoid inversions, meaning headstand and shoulder stand. A 2024 study in a German ophthalmology journal suggests that advice points at the wrong poses.
Researchers fitted 25 healthy volunteers and 25 patients with glaucoma with a handheld self-tonometer and had each hold four postures for 90 seconds, with a two-minute break between, measuring eye pressure before, during and after. Pressure rose significantly in every eye in every position. Legs-up-the-wall produced the smallest change, 1.6 mmHg in healthy eyes and 2.8 in glaucoma patients. Plough pose produced 7.5 and 6.0. A standing forward fold produced 14.4 and 11.6.
The largest rise came in downward-facing dog: 16.5 mmHg in healthy eyes and 15.1 in glaucoma patients, with mean pressure climbing above 31 mmHg in both groups. That pose is in essentially every beginner class and every gentle sequence. The spike appeared the instant the posture was assumed, did not grow over the following 90 seconds, and returned to baseline afterward in every participant.
Read the limits honestly. This measured pressure, not vision loss. The authors say plainly that whether yoga practice drives glaucoma to progress remains unclear and needs more research, while noting that pressure spikes are a major risk factor for the optic nerve damage that defines the disease. Their advice is that people with glaucoma choose their postures carefully. That is a different instruction from avoiding inversions, and it is a better one, because the pose that produced the biggest spike is not an inversion at all. Anyone with glaucoma or raised eye pressure should take this question to their ophthalmologist rather than to a class.
Yoga is at least three different things
Much of the confusion in this subject comes from one word covering three distinct objects.
There is yoga as exercise, a class at a gym. There is yoga as a therapeutic intervention, meaning a named, written-down protocol delivered for a diagnosis. And there is yoga as a philosophical and religious tradition with roots in ancient Indian thought. The rheumatology guideline goes only as far as philosophy, and the federal consumer page acknowledges the rest only by noting that classical yoga includes elements such as a spiritual practice. That tradition belongs to the people who carry it, and treating it as a wellness accessory is inaccurate as well as discourteous.
The trial literature makes the middle category concrete. The anxiety trial below used Kundalini yoga, a specific lineage with an explicit spiritual framework, delivered as a 24-hour clinical course. The cancer sleep trial used a purpose-built four-week program of 16 gentle Hatha and restorative postures. The back pain trials used Iyengar, Hatha and Viniyoga. Whatever a local studio is teaching this week, it is probably none of these.
Where it works, and what it loses to
The single best-designed yoga trial in the literature is also the clearest illustration of how to report one.
Published in JAMA Psychiatry in 2021 and funded by the National Institutes of Health, it enrolled 226 adults with a primary diagnosis of generalized anxiety disorder at two academic centres, and randomized them to Kundalini yoga, cognitive behavioral therapy, or stress education as an active control. All three ran as twelve 120-minute group sessions with 20 minutes of daily homework, and independent raters who did not know the assignments judged the outcomes. Response rates: 54.2 percent for yoga, 70.8 percent for the therapy, 33.0 percent for the control. Yoga beat the control with a number needed to treat of about five. It also failed the test of being as good as cognitive behavioral therapy, by a gap of 16.6 percentage points. The authors’ conclusion was that yoga worked and that the therapy should stay the first choice.
Both halves are the finding. Reporting only the first oversells it; reporting only the second buries a real result.
For cancer-related sleep problems, a 410-person multicentre trial published in the Journal of Clinical Oncology in 2013 found that adding a four-week yoga program to standard care improved overall sleep quality and, notably, cut the use of sleep medication. Ninety-six percent of participants were women and three-quarters had breast cancer. The comparison was standard care rather than another exercise program, so the trial says nothing about whether yoga specifically was the active part, and the federal digest makes exactly that point, observing that other forms of exercise may deliver similar benefits.
For knee osteoarthritis, the 2019 osteoarthritis guideline from the American College of Rheumatology and the Arthritis Foundation rates yoga as conditionally recommended, a weaker standing than tai chi receives in the same document, and makes no recommendation at all for hip osteoarthritis because the data were absent. The tai chi article covers that guideline’s reasoning.
Two credentials, and neither one is a licence
A “yoga therapist” and a “registered yoga teacher” come from different organizations and mean different things, and neither is issued by any government.
Certification from the International Association of Yoga Therapists, written C-IAYT, is the therapy credential. The association built educational competencies starting in 2007, began accrediting training programs in 2014, and began certifying individuals in 2016. Graduates of an accredited program of 800 hours or more may apply; a portfolio route exists for experienced practitioners from elsewhere. An examination became mandatory for portfolio applicants on March 1, 2023, and for accredited-program students enrolling from July 1, 2023. A grandfathering window closed in June 2017 and will not reopen. The association’s own word for the whole arrangement is a self-regulatory initiative.
The RYT designations come from Yoga Alliance and are a teaching registry. Its standards set hour thresholds: 200 or 500 training hours for the basic designations, 1,000 teaching hours plus two years for an experienced 200-hour teacher, 2,000 hours plus four years at the 500-hour level. No examination appears anywhere in the document, and neither does yoga therapy.
Sort both against the four categories that matter. Defined in regulation: no. A government certification program: no. A private third-party credential: yes for both, and that is the correct description. Undefined: the title itself, because nothing prevents a person holding neither credential from calling themselves a yoga therapist. No state licensure of yoga therapists was identified for this article.
Pregnancy, older bodies and thin bones
Population-specific advice is where practitioner lore and professional guidance get mixed together, so it helps to say where each line comes from.
For pregnancy, the relevant American professional guidance is Committee Opinion Number 804 from the American College of Obstetricians and Gynecologists, issued in April 2020, which holds that in the absence of obstetric or medical complications physical activity in pregnancy is safe and desirable, while some modification may be needed for the anatomical and physiological changes of pregnancy. That document, as published, does not single out yoga. The specific precaution people repeat, avoiding long stretches lying flat on the back, comes from the federal consumer page rather than from the obstetricians, and the threshold question of whether exercise is safe at all in a given pregnancy belongs to the clinician providing prenatal care.
For older adults, federal advice is to discuss needs with both a health care provider and the instructor, and to modify or skip some postures. Set that beside the emergency department rate in people 65 and over and the advice stops sounding like boilerplate.
For thin or fragile bone, the recommendation to avoid forceful practices comes from the authors of the 2013 case-report review, in their own recommendations section. That is peer-reviewed literature rather than studio folklore, which is worth saying because no professional body’s guidance on yoga for osteoporosis was located for this article. The same review is the origin of the glaucoma caution, which the 2024 pressure study has now narrowed and partly redirected.
Call a clinician rather than an instructor if a posture produces numbness, tingling or weakness down an arm or leg, if back pain is worse a week into practising than it was at the start, if you have unexplained eye pain or visual change after class, or if you have osteoporosis, a recent fracture, uncontrolled glaucoma or recent spinal surgery and nobody has reviewed which postures are appropriate. Nothing in this article is a reason to start, stop, substitute or delay any treatment.
The useful takeaway
Yoga for chronic back pain is better than nothing and about the same as other exercise, and the gap between it and nothing is smaller than the thresholds researchers set for a difference patients would notice. That is a modest claim, and it is compatible with a strong guideline recommendation, because the options it competes against are not impressive either. For a diagnosed anxiety disorder yoga has real, measured efficacy and still loses to talk therapy. If you enjoy it, the enjoyment is a legitimate reason to choose it, since adherence is the one variable none of these comparisons controls. If you are over 65, have glaucoma or have fragile bones, raise it with your clinician first, and tell your instructor exactly what you were told.
Sources
- Annals of Internal Medicine (American College of Physicians). “Qaseem A, Wilt TJ, McLean RM, Forciea MA. Noninvasive Treatments for Acute, Subacute, and Chronic Low Back Pain: A Clinical Practice Guideline From the American College of Physicians (strong recommendation; yoga in the low-quality-evidence bracket). 2017. DOI: 10.7326/M16-2367.” pubmed.ncbi.nlm.nih.gov
- Cochrane Database of Systematic Reviews. “Wieland LS, Skoetz N, Pilkington K, Harbin S, Vempati R, Berman BM. Yoga for chronic non-specific low back pain (21 trials, 2,223 participants; no sham-yoga comparison exists). 2022. DOI: 10.1002/14651858.CD010671.pub3.” pubmed.ncbi.nlm.nih.gov
- Cochrane Database of Systematic Reviews. “Wieland LS, Skoetz N, Pilkington K, Vempati R, D'Adamo CR, Berman BM. Yoga treatment for chronic non-specific low back pain (the 2017 edition, 12 trials, 1,080 participants). DOI: 10.1002/14651858.CD010671.pub2.” pubmed.ncbi.nlm.nih.gov
- National Center for Complementary and Integrative Health (NIH). “Yoga for Health: What the Science Says (Clinical Digest; content dated February 2020, under a site-wide footer reading Site Last Updated: October 19, 2026).” nccih.nih.gov
- National Center for Complementary and Integrative Health (NIH). “5 Things You Should Know About Yoga (no publication or last-updated date appears on the page).” nccih.nih.gov
- Orthopaedic Journal of Sports Medicine. “Swain TA, McGwin G. Yoga-Related Injuries in the United States From 2001 to 2014 (National Electronic Injury Surveillance System; rates by age group). 2016. DOI: 10.1177/2325967116671703.” pubmed.ncbi.nlm.nih.gov
- PLOS One. “Cramer H, Krucoff C, Dobos G. Adverse events associated with yoga: a systematic review of published case reports and case series (76 cases; no denominator). 2013. DOI: 10.1371/journal.pone.0075515.” pubmed.ncbi.nlm.nih.gov
- Klinische Monatsblatter fur Augenheilkunde. “Shajiei TD, Wachtl J, Schuknecht A, Bachmann L, Kniestedt C. The Effect of Yoga on Intraocular Pressure Using the iCare HOME2 Tonometer (largest rise in downward-facing dog). 2024. DOI: 10.1055/a-2200-3970.” pubmed.ncbi.nlm.nih.gov
- JAMA Psychiatry. “Simon NM, Hofmann SG, Rosenfield D, et al. Efficacy of Yoga vs Cognitive Behavioral Therapy vs Stress Education for the Treatment of Generalized Anxiety Disorder: A Randomized Clinical Trial. 2021. DOI: 10.1001/jamapsychiatry.2020.2496.” pubmed.ncbi.nlm.nih.gov
- Journal of Clinical Oncology. “Mustian KM, Sprod LK, Janelsins M, et al. Multicenter, randomized controlled trial of yoga for sleep quality among cancer survivors. 2013. DOI: 10.1200/JCO.2012.43.7707.” pubmed.ncbi.nlm.nih.gov
- Arthritis & Rheumatology (American College of Rheumatology / Arthritis Foundation). “2019 American College of Rheumatology/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee (yoga conditionally recommended for knee OA; no recommendation for hip OA). 2020. DOI: 10.1002/art.41142.” pubmed.ncbi.nlm.nih.gov
- Obstetrics & Gynecology (American College of Obstetricians and Gynecologists). “Physical Activity and Exercise During Pregnancy and the Postpartum Period: ACOG Committee Opinion, Number 804, April 2020 (the published document does not name yoga). DOI: 10.1097/AOG.0000000000003772.” pubmed.ncbi.nlm.nih.gov
- International Association of Yoga Therapists. “Certification Background and Future (C-IAYT; 800-hour accredited-program threshold, portfolio pathway, examination phased in during 2023, described by IAYT as a self-regulatory initiative).” iayt.org
- Yoga Alliance. “Standards for RYT Credentials (training and teaching-hour thresholds; no examination and no mention of yoga therapy).” yogaalliance.org