The stroke comparison that explains the chiropractic debate

A tear in the wall of an artery in the neck is one of the leading causes of stroke in adults under 50. It is called a cervical artery dissection, and the symptoms it starts with are neck pain and headache.

Hold that sentence for a second, because everything difficult about this subject follows from it. Neck pain and headache are also the two most common reasons people book an appointment to have their neck adjusted.

So when researchers compared people who had a stroke in the back of the brain against matched controls, and found that the stroke patients were more likely to have seen a chiropractor in the preceding days, there were two possible readings. Either the adjustment tore the artery, or the tear was already happening and sent the patient looking for someone to help with the neck pain it was causing.

One study design settled which reading the data support, and almost nobody explains it. Researchers ran the same comparison using primary care physicians instead of chiropractors. Family doctors do not manipulate necks. They found the same association, and in some analyses a stronger one.

The event, and why it is hard to study

The American Heart Association and American Stroke Association published a scientific statement on this question in the journal Stroke in August 2014. It is the single most-cited document in the argument, and it says three things at once. Reading only one of them is how both sides misreport it.

First, that current biomechanical evidence is not sufficient to establish that cervical manipulation causes dissection. Second, that case reports point to mechanical force as a contributor in a substantial share of dissections, and that most population-controlled studies have found an association between manipulation and vertebral artery dissection stroke in young patients. Third, the operative recommendation. Writing CD for cervical dissection and CMT for cervical manipulative therapy, the statement says practitioners should strongly consider dissection as a presenting symptom, and that “patients should be informed of the statistical association between CD and CMT prior to undergoing manipulation of the cervical spine.”

Those are not contradictory positions. They are what a careful body says when an association is real, a mechanism is unproven, and the stakes are a stroke. The statement also records that no evidence-based guidelines exist for managing a dissection once it happens.

The federal government declines to adjudicate. The National Center for Complementary and Integrative Health, whose spinal manipulation page carries a September 2022 content date, says neck manipulation has been linked to these tears, that the incidence among people receiving manipulation appears low, and then, using the abbreviation CAD for cervical artery dissection, states that “there is disagreement about whether manipulation can cause CAD.” A federal health agency printing the word disagreement is itself a finding, and an article that resolves the question is going further than the government will.

The comparison that changes the question

The decisive study is a population-based case-control and case-crossover analysis by David Cassidy and colleagues, published in Spine in 2008. It identified 818 vertebrobasilar strokes admitted to Ontario hospitals between April 1993 and March 2002, from a population of more than 100 million person-years. Each case was matched to four controls, and the case-crossover design used patients as their own controls.

Among patients under 45, cases were about three times more likely than controls to have seen a chiropractor before the stroke. They were also about three times more likely to have seen a primary care physician. Above 45 there was no increased association with chiropractic visits at all, while positive associations with physician visits appeared in every age group. Visits billed for headache and neck complaints were highly associated with the subsequent stroke, whoever provided them.

The authors concluded that the excess risk most likely reflects people whose dissection was already underway going to someone for the neck pain and headache it was causing, in the days before the stroke arrived. They found no excess risk attached to chiropractic care over primary care.

A 2015 US case-control study by Thomas Kosloff and colleagues reproduced the pattern in commercial and Medicare Advantage health plans, covering 1,829 stroke cases between 2011 and 2013. It found no significant association with chiropractic visits in either population or in any age stratum, and a significant association with primary care physician visits in both, at every hazard period. All four authors worked for a health insurer and it appeared in a chiropractic journal, which is worth knowing. Its secondary finding cuts against every claims-based study here, including the ones that found an association: in nearly a third of commercial-population stroke cases, the chiropractic visit record did not report that any manipulation had been performed. Counting visits is not the same as counting adjustments.

The older positive study is worth reading the same way. An Ontario case-control study published in Stroke in 2001 covered 582 cases and found that under 45, cases were five times more likely than controls to have visited a chiropractor within a week. The confidence interval around that estimate ran from 1.32 to 43.87. Writers quote the five and drop the interval. An upper bound of 44 is not a precise finding, and the authors flagged potential sources of bias themselves.

A 2016 systematic review and meta-analysis by Ephraim Church and a Penn State neurosurgery group pooled the available studies and landed in the middle. It found a small association, an odds ratio of 1.74 with a confidence interval that excludes no effect, and rated the whole body of evidence very low quality under the GRADE system. Its conclusion was that there is no convincing evidence of a causal link. Notice that the paper confirms an association while denying causation, and writers tend to cite one half.

Two boundaries matter. A 2015 Dartmouth analysis of 1.16 million Medicare beneficiaries aged 66 to 99 with neck pain found stroke incidence extremely low and the differences between chiropractic and physician patients probably not clinically significant. Every study that finds a signal finds it under 45, so an over-66 cohort cannot answer the question about young patients.

And the whole epidemiological literature is about one vessel. A systematic review by Chadwick Chung and colleagues searched from 1970 to November 2012 for any cohort study, case-control study or trial on cervical manipulation and internal carotid artery dissection. It found none. Not a weak study, not a conflicting one. None. The phrase cervical artery dissection covers the carotid and the vertebral arteries, and for the carotid there is only a pile of case reports with no denominator. Any article that refers to the cervical artery dissection studies is describing half the vessel territory and implying the other half.

What the harm looks like

A systematic review led by Bryden Leung at the University of Queensland, published in 2025, is the most comprehensive catalogue available. Searching to May 2023, it identified 233 studies describing 334 serious adverse event cases after physical procedures directed at the neck, 41 of them reported in the previous five years.

Fifty-eight percent were vascular, mainly arterial dissection or vertebral artery events, and 75 percent involved manipulation. The remainder rarely gets written about: 25 percent neurological, 12 percent both, and 5 percent other, a category including cerebrospinal fluid leaks, phrenic nerve palsies and retinal detachments. Some cases followed gentler things entirely, and the review names vestibular testing, gentle mobilization, exercises, acupuncture and massage.

Of the reported cases, 62 percent recovered favorably, 16 percent recovered with disability, and 6 percent died. That 6 percent is a share of reported serious events, not a share of patients treated. There is no denominator anywhere in this literature, which is why no rate can be computed from it in either direction.

The reviewers identified a consistent warning pattern that a patient can use. What came first was a sudden worsening of neck pain or headache, along with dizziness, nausea or vomiting and changes in sensation, appearing either during the treatment or inside 48 hours, and usually before any neurological sign showed up. Their recommendation is to monitor for early signs for up to 48 hours if a serious event is suspected.

That gives the one piece of practical guidance everyone in this argument agrees on. New, severe or unfamiliar neck pain or headache, especially with dizziness, visual change, slurred speech, facial droop or altered sensation, is a reason for urgent medical assessment, and is not a symptom to take to a manual therapist in the first instance. The heart association statement and the chiropractic profession’s own response both point there.

That response is on the record. The World Federation of Chiropractic published talking points on the heart association statement in August 2014. It accepts the statement’s causation language, notes that other professions perform the same procedure, and endorses the recommendation on informed consent. It disputes the singling out of manipulation, cites biomechanical work finding that the forces involved fall below those in ordinary passive movement, and puts incidence at roughly one in a million treatments. That last figure is the profession’s own, not independently established, and should be read as a position rather than a measured rate.

What manipulation does for back and neck pain

Effectiveness is a separate question from safety and the answers are more modest than the marketing on either side.

For acute low back pain, a 2012 Cochrane review by Sidney Rubinstein and colleagues covering 20 trials and 2,674 participants concluded that manipulation performs no better than a dummy procedure, no better than a sham version of itself, no better when bolted onto another treatment, and no better than the other therapies guidelines already recommend. Six of the 20 trials had low risk of bias. The reviewers suggested the referral decision should rest on cost, preference and relative safety rather than on superior effect.

For chronic low back pain, Rubinstein’s team published a larger review in the BMJ in 2019 covering 47 trials and 9,211 participants. The order in which its results are reported determines how it reads. High-quality evidence says manipulation produces small effects on pain and small to moderate improvements in function compared with therapies that are not recommended. Moderate-quality evidence says it produces similar effects to therapies that are recommended, with no significant difference in short-term pain. The evidence against sham manipulation was low to very low quality, and the authors said those effects should be considered uncertain. Reported in that sequence, the strongest evidence here is that manipulation beats things nobody recommends.

The American College of Physicians names spinal manipulation for both acute and chronic low back pain in its 2017 guideline, each time as a strong recommendation resting on low-quality evidence. In the chronic recommendation it sits in the low-quality group while acupuncture and exercise sit in the moderate-quality group. A strong recommendation here reflects a favorable balance of benefit against harm relative to drugs, not strong efficacy data.

For neck pain, a 2015 Cochrane review led by Anita Gross covered 51 trials and 2,920 participants. Its findings on cervical manipulation against an inactive control came from three trials and 154 participants, were rated very low to low quality, and were not pooled into a single estimate. Its best-evidenced finding, rated moderate to high quality, is an equivalence. Several sessions of cervical manipulation and several sessions of cervical mobilization, which is slower non-thrust movement of the joint, came out the same on pain, on function, on satisfaction and on quality of life. The review also found manipulation more effective than certain medications for acute and subacute pain, on moderate-quality evidence from single trials.

Then the authors wrote the sentence that makes this actionable. Because manipulation carries a risk of rare but serious adverse events, they called for more high-quality research focused on mobilization. They did not say mobilization is safe, and the catalogue above records serious events after gentle mobilization too. If the two perform the same, whether mobilization would do is still a fair question to raise.

Three professions, one technique

Spinal manipulation and chiropractic are not synonyms. Cochrane’s own definition of the intervention covers hands-on therapy directed at the spine from chiropractors, manual therapists and osteopaths alike.

Chiropractors are the largest group, with about 61,700 jobs in 2025 by the Bureau of Labor Statistics count. Practice requires a Doctor of Chiropractic degree, typically four years after at least three years of undergraduate study, accredited by the Council on Chiropractic Education, which the Secretary of Education recognizes as an accrediting agency. Candidates sit examinations from the National Board of Chiropractic Examiners, which states that all 50 states either accept or require Parts I through IV for licensure, with some states adding their own examinations. Every state requires a license and continuing education.

Osteopathic physicians hold a full medical degree and learn manipulative treatment inside it, a distinct tradition with its own training route and its own article in this library. Physical therapists also perform manipulation, and their national association has asserted since 2000 that thrust and non-thrust techniques are taught exclusively in physical therapist programs. That is a professional body’s position in a boundary dispute, not a statement of law. Credentials across these practices have their own article here.

What this means is simple. You are not choosing a technique. You are choosing a profession, with its own degree, accreditor, examination and scope, and scope varies enormously between states.

Claims beyond the spine, and a regulator that changed its mind twice

For conditions outside the musculoskeletal system, the authoritative sentence already exists. Reviewing studies on things like high blood pressure and menstrual cramps, the federal center reports that not one of them turned up a clear benefit from spinal manipulation.

A 2022 systematic scoping review in BMC Pediatrics by Nikki Milne and colleagues went through 87 articles on manipulation and mobilization in children. It found strong to very strong evidence of no significant effect for asthma as measured by lung function, for headache and for bedwetting, and inconclusive or insufficient evidence for everything else examined, from colic and ear infections to cerebral palsy. Its overall conclusion was that the collective findings do not support either practice in children for any condition, and that adverse event reporting is too thin to establish true risks.

Which makes the Australian regulatory chronology the best-documented thing in this subject. In 2019 the Chiropractic Board of Australia released an interim policy on spinal manipulation of children under 12 while evidence was gathered. It then commissioned Cochrane Australia for an updated systematic review. In November 2023, on the strength of that review, the board published updated guidelines allowing chiropractors to treat children according to current evidence and best practice, lifting the restriction. On 14 June 2024, Australia’s health ministers asked the board to think again. On 17 June 2024 the board reinstated the interim policy, now advising chiropractors not to use spinal manipulation on children under two, pending further consultation. A regulator restricted a practice, commissioned the review, changed its mind, and changed it back inside five years, all in public. There is no equivalent US restriction.

The most consistently omitted finding in the pediatric literature is also the most probable harm. A 2007 systematic review in Pediatrics by Sunita Vohra and colleagues identified 14 cases of direct adverse events in children, nine of them serious, including a subarachnoid haemorrhage and a case of paraplegia. It also identified 20 further cases of indirect harm. Some involved a diagnosis arriving late, with diabetes and neuroblastoma among the conditions missed. Others involved manipulation being given to children who turned out to have meningitis or a soft tissue cancer. The authors were careful that reports like these cannot establish either a cause or a rate.

A late diagnosis does not require the manipulation to injure anything. It requires only that a child with a serious illness spends weeks being treated for something else. That is the harm parents are least often warned about, and the one that needs no biomechanics to work.

A last oddity sits inside the federal government. Medicare’s coverage page for chiropractic services pays for spinal adjustments that correct a subluxation, which the page describes as joints of the spine that stop moving as they should while still touching one another, performed by hand or with a device called an activator. The federal health agency’s own chiropractic page never uses the word subluxation, and its manipulation page says no study shows clear benefit outside musculoskeletal conditions. Two parts of one government, two vocabularies.

The one question to raise before a neck adjustment

Every decision about starting, continuing or changing treatment belongs with a clinician who knows your history. What the record supports is a short list of things to raise.

1. Ask whether mobilization would do instead of a thrust. Cochrane rates the two as equivalent on its best evidence, and the reported serious events concentrate on manipulation.
2. Ask to be told the association before, not after. The heart association recommends it and the profession’s international federation accepts it. If nobody raises it, raise it yourself.
3. Describe the headache or neck pain precisely, including whether it is new, unusually severe, or unlike your usual pattern. That description is what should prompt assessment rather than treatment.
4. Ask which profession you are seeing and what its scope covers in your state, since scope for the same procedure varies widely.

If you take one thing from this, take the primary care comparison. An association between a practitioner visit and a later stroke says nothing about what the practitioner did, unless somebody checks whether visiting a practitioner who touches nothing shows the same pattern. Somebody checked. It does. That reasoning is missing from almost every article on this subject, and it cuts both ways: it is the best reason not to assume an adjustment caused a stroke, and the best reason to get new neck pain looked at first.

Sources

1. Stroke (American Heart Association / American Stroke Association). “Biller J, Sacco RL, Albuquerque FC, et al. "Cervical arterial dissections and association with cervical manipulative therapy: a statement for healthcare professionals from the American Heart Association/American Stroke Association." Stroke, 2014. PubMed. DOI: 10.1161/STR.0000000000000016.” pubmed.ncbi.nlm.nih.gov
2. Spine. “Cassidy JD, Boyle E, Cote P, He Y, Hogg-Johnson S, Silver FL, Bondy SJ. "Risk of vertebrobasilar stroke and chiropractic care: results of a population-based case-control and case-crossover study." Spine, 2008 (818 strokes, more than 100 million person-years). PubMed. DOI: 10.1097/BRS.0b013e3181644600.” pubmed.ncbi.nlm.nih.gov
3. Chiropractic & Manual Therapies. “Kosloff TM, Elton D, Tao J, Bannister WM. "Chiropractic care and the risk of vertebrobasilar stroke: results of a case-control study in U.S. commercial and Medicare Advantage populations." Chiropr Man Therap, 2015 (1,829 cases; authors employed by a health insurer). PubMed. DOI: 10.1186/s12998-015-0063-x.” pubmed.ncbi.nlm.nih.gov
4. Cureus. “Church EW, Sieg EP, Zalatimo O, Hussain NS, Glantz M, Harbaugh RE. "Systematic Review and Meta-analysis of Chiropractic Care and Cervical Artery Dissection: No Evidence for Causation." Cureus, 2016 (OR 1.74, GRADE very low). PubMed. DOI: 10.7759/cureus.498.” pubmed.ncbi.nlm.nih.gov
5. Journal of Manipulative and Physiological Therapeutics. “Chung CLR, Cote P, Stern P, L'Esperance G. "The Association Between Cervical Spine Manipulation and Carotid Artery Dissection: A Systematic Review of the Literature." J Manipulative Physiol Ther, 2015 (no epidemiologic studies found). PubMed. DOI: 10.1016/j.jmpt.2013.09.005.” pubmed.ncbi.nlm.nih.gov
6. Stroke. “Rothwell DM, Bondy SJ, Williams JI. "Chiropractic manipulation and stroke: a population-based case-control study." Stroke, 2001 (582 cases; under 45, 5 times more likely, 95% CI 1.32 to 43.87). PubMed. DOI: 10.1161/01.str.32.5.1054.” pubmed.ncbi.nlm.nih.gov
7. Journal of Bodywork and Movement Therapies. “Leung B, Treleaven J, Dinsdale A, Marsh L, Thomas L. "Serious adverse events associated with conservative physical procedures directed towards the cervical spine: A systematic review." J Bodyw Mov Ther, 2025 (233 studies, 334 cases). PubMed. DOI: 10.1016/j.jbmt.2024.10.018.” pubmed.ncbi.nlm.nih.gov
8. BMJ. “Rubinstein SM, de Zoete A, van Middelkoop M, Assendelft WJJ, de Boer MR, van Tulder MW. "Benefits and harms of spinal manipulative therapy for the treatment of chronic low back pain: systematic review and meta-analysis of randomised controlled trials." BMJ, 2019 (47 trials, 9,211 participants). PubMed. DOI: 10.1136/bmj.l689.” pubmed.ncbi.nlm.nih.gov
9. Cochrane Database of Systematic Reviews. “Rubinstein SM, Terwee CB, Assendelft WJJ, de Boer MR, van Tulder MW. "Spinal manipulative therapy for acute low-back pain." Cochrane Database Syst Rev, 2012 (20 trials, 2,674 participants). PubMed. DOI: 10.1002/14651858.CD008880.pub2.” pubmed.ncbi.nlm.nih.gov
10. Cochrane Database of Systematic Reviews. “Gross A, Langevin P, Burnie SJ, et al. "Manipulation and mobilisation for neck pain contrasted against an inactive control or another active treatment." Cochrane Database Syst Rev, 2015 (51 trials, 2,920 participants). PubMed. DOI: 10.1002/14651858.CD004249.pub4.” pubmed.ncbi.nlm.nih.gov
11. Journal of Manipulative and Physiological Therapeutics. “Whedon JM, Song Y, Mackenzie TA, Phillips RB, Lukovits TG, Lurie JD. "Risk of stroke after chiropractic spinal manipulation in Medicare B beneficiaries aged 66 to 99 years with neck pain." J Manipulative Physiol Ther, 2015 (1,157,475 beneficiaries; incidence extremely low; differences probably not clinically significant). PubMed. DOI: 10.1016/j.jmpt.2014.12.001.” pubmed.ncbi.nlm.nih.gov
12. Pediatrics. “Vohra S, Johnston BC, Cramer K, Humphreys K. "Adverse events associated with pediatric spinal manipulation: a systematic review." Pediatrics, 2007 (14 direct and 20 indirect adverse events). PubMed. DOI: 10.1542/peds.2006-1392.” pubmed.ncbi.nlm.nih.gov
13. BMC Pediatrics. “Milne N, Longeri L, Patel A, Pool J, Olson K, Basson A, Gross AR. "Spinal manipulation and mobilisation in the treatment of infants, children, and adolescents: a systematic scoping review." BMC Pediatr, 2022 (87 articles). PubMed. DOI: 10.1186/s12887-022-03781-6.” pubmed.ncbi.nlm.nih.gov
14. National Center for Complementary and Integrative Health, National Institutes of Health. “Spinal Manipulation: What You Need To Know (content Last Updated September 2022): evidence by condition, the statement that no studies showed a clear benefit for non-musculoskeletal conditions, and the statement that there is disagreement about whether manipulation can cause cervical artery dissection.” nccih.nih.gov
15. National Center for Complementary and Integrative Health, National Institutes of Health. “Chiropractic: In Depth (content Last Updated April 2019): degree, accreditation by the Council on Chiropractic Education, national board examination, state licensure and continuing education.” nccih.nih.gov
16. Chiropractic Board of Australia (Ahpra). “Chiropractic Board reinstates interim policy, 17 June 2024 (2019 interim policy for under 12s, Cochrane Australia review, November 2023 guidelines, reinstatement for under twos after the Health Ministers Meeting of 14 June).” chiropracticboard.gov.au
17. National Board of Chiropractic Examiners. “Links to NBCE Exams (Parts I to IV, Physiotherapy elective, SPEC; all 50 states accept or require Parts I through IV for licensure; additional state examinations in Florida, Missouri, Oklahoma and Oregon).” nbce.org
18. U.S. Bureau of Labor Statistics. “Occupational Outlook Handbook: Chiropractors (last modified August 27, 2026; about 61,700 jobs in 2025, median annual wage $79,200 in May 2025).” bls.gov
19. Centers for Medicare & Medicaid Services (Medicare.gov). “Chiropractic services coverage: Part B covers manual manipulation of the spine to correct a subluxation, by hand or with an activator device; X-rays, massage therapy and acupuncture ordered by a chiropractor are not covered.” medicare.gov
20. World Federation of Chiropractic. “WFC talking points on AHA scientific statement on CMT and cervical artery dissection (page dated 1 August 2014): accepts the causation language and the informed consent recommendation, disputes the focus on manipulation, puts incidence at about one in a million treatments.” wfc.org