Industries / Health / Healthcare / Chiropractic
Evidence and outcomes
This is the work of finding out, by trials and reviews, whether chiropractic care helps, for which conditions, and what harm it can do.
7
Milestones achieved, from 2004 to 2021.
3
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
2004
Manipulation tested in UK practice
A 1990 trial in Britain of 741 patients in 11 centres found chiropractic better than hospital outpatient care for low back pain, with about 29% more improvement on the Oswestry disability score at three years, but it was criticised for not separating the treatment from where it was given. The UK BEAM trial then assigned 1,334 patients from 181 general practices to "best care" from their doctor, with or without exercise classes, spinal manipulation in NHS or private premises, or both. Compared with best care, manipulation improved scores on the 24-point Roland Morris disability questionnaire by 1.6 points at three months and 1.0 at 12 months, and manipulation followed by exercise by 1.9 and 1.3; the premises made no difference, and no serious adverse events occurred.
Lessona fair test separates the treatment from the setting it is given in.
pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov
Credit
T. W. Meade and colleagues (BMJ, 1990 and 1995); the UK BEAM Trial Team (BMJ, 2004)
2007
Neck thrusts counted nationally
In a prospective national survey, UK chiropractors recorded what happened after 28,807 visits by 19,722 patients, which included 50,276 thrusts to the neck. No serious adverse event, such as a hospital referral or lasting disability, was reported, which puts the risk at worst at about 1 per 10,000 visits immediately after treatment and about 6 per 100,000 neck manipulations, by the authors' calculation. Minor effects were common: headache followed up to about 4 in 100 visits within 7 days.
Lessona survey of this size sets an upper limit on rare harms but cannot see harms rarer than that limit.
Credit
H. W. Thiel, J. E. Bolton, S. Docherty and J. C. Portlock, Anglo-European College of Chiropractic
2012
Colic results hinge on blinding
A Cochrane review of manipulation for infant colic found six trials with 325 babies under six months old. Pooled, parents of treated babies recorded about 1 hour 12 minutes less crying a day; in the two trials where parents did not know whether their baby had been treated, the difference, 34 minutes, was not statistically significant. The authors concluded that no definitive conclusion was possible and that there were too few data to judge safety; Bronfort's 2010 UK evidence report had rated manipulation not effective for colic against sham.
Lessonoutcomes reported by people who know which treatment was given can favour that treatment.
Credit
D. Dobson, P. L. Lucassen, J. J. Miller, A. M. Vlieger, P. Prescott and G. Lewith
2014
Heart association on neck tears
The American Heart Association and American Stroke Association reviewed cervical artery dissection, a tear in an artery of the neck and among the most common causes of stroke in young and middle-aged adults. Their statement says most population studies found a link between neck manipulation and vertebral artery tears in young patients, but that it is unclear whether manipulation causes tears or is given to people whose tear has already begun, and that biomechanical evidence is insufficient to show it causes them. It advises that patients be told of the statistical association before neck manipulation.
Credit
J. Biller, R. L. Sacco and colleagues, American Heart Association Stroke Council
2017
Guidelines list manipulation
For low back pain of up to six weeks, a review of 26 trials by N. M. Paige, P. G. Shekelle and colleagues found that manipulation improved pain by about 10 points on a 100-point scale in 15 trials with 1,711 patients, and function modestly, with large differences between trials. No trial reported a serious adverse event, while large case series found minor, short-lived effects such as more pain, stiffness or headache in 50% to 67% of patients. The American College of Physicians' 2017 guideline listed spinal manipulation, on low-quality evidence, among the first treatments to try for acute and chronic low back pain, and in December 2023 the World Health Organization's first guideline on chronic low back pain included it, with all of its recommendations conditional.
pmc.ncbi.nlm.nih.goveuropepmc.orgwho.intpmc.ncbi.nlm.nih.gov
Credit
N. M. Paige, P. G. Shekelle and colleagues, West Los Angeles VA Medical Center (JAMA, 2017); A. Qaseem and the American College of Physicians; WHO
2018
Chiropractic added in the military
At three US military hospitals, 750 active-duty service members with low back pain got usual medical care alone or with six weeks of chiropractic care. At six weeks the chiropractic group had 1.1 points less pain on a 0-to-10 scale and 2.2 points less disability on the 24-point Roland Morris scale, with no serious related adverse events. In a 2026 analysis of the last 154 people enrolled, who were asked to report for a year, the disability difference was 1.7 points at 52 weeks, too uncertain to rule out no difference.
Lessonthe gain was clear at six weeks and had narrowed by a year, so short trials alone cannot show lasting benefit.
Credit
C. M. Goertz, C. R. Long and colleagues, Palmer College of Chiropractic
2021
No effect beyond the spine
At a Global Summit in Toronto in September 2019, 50 researchers reviewed English-language trials of spinal manipulation for non-musculoskeletal disorders, conditions not of the bones, joints or muscles. All six acceptable or high-quality trials, with 534 participants, on infant colic, childhood asthma, high blood pressure, period pain and migraine, found manipulation no better than sham; 6 of the 50 did not approve the final report. A 1998 trial in 91 children with asthma had already found that real manipulation added nothing over simulated manipulation, and a second 2021 review found no clinical study at all of manipulation for preventing infections or improving immune function, though such claims had grown during COVID-19. Dana Lawrence of Parker University objected that the summit's assumption of little publication bias was untested in chiropractic, and Bronfort's 2010 report had rated manipulation effective for migraine.
pmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.goveuropepmc.orgpmc.ncbi.nlm.nih.govpmc.ncbi.nlm.nih.gov
Credit
Pierre Côté, J. Hartvigsen and colleagues; Ngai Chow, Pierre Côté and colleagues, Ontario Tech University; J. Balon and colleagues, Canadian Memorial Chiropractic College (NEJM, 1998)
Watch for
Fewer opioid prescriptions
A meta-analysis of 2 randomized trials with 838 participants and 18 cohort studies with about 6 million people found that adults with spine pain who saw a chiropractor had lower odds of an opioid prescription than those given medical care alone (odds ratio 0.36), and lower still when seen within 30 days. The authors rate the certainty of this evidence as very low and call for randomized trials. Separately, among 128,377 US veterans with new low back pain who had not been taking opioids, 13.0% of chiropractic users and 16.8% of others got an opioid prescription within a year.
Achieved whena randomized trial with opioid prescriptions as a planned main outcome finds fewer prescriptions with chiropractic care.
Claimed by
P. C. Emary, J. W. Busse and colleagues, McMaster University, December 2025; A. J. Lisi and colleagues, US Department of Veterans Affairs, May 2025
Adjustments and blood markers
In a trial in Rawalpindi, Pakistan, 106 adults with low-level spinal pain (subclinical, in the authors' term) got 12 weeks of chiropractic care or sham care. After 12 weeks the chiropractic group had higher blood levels of BDNF, a protein linked to the brain's ability to change, and of IL-6, a signal of inflammation, and higher saliva cortisol, a stress hormone; the authors take this as evidence that chiropractic care modulates the body's stress, inflammation and brain-plasticity markers. The trial enrolled fewer than the 150 people planned, and those who finished in the chiropractic group were older on average, 37.5 against 26.9 years.
Achieved whena team outside the college repeats the trial with a preregistered main outcome, finds the same changes and shows that they affect patients' health.
Claimed by
I. Amjad, H. Haavik and colleagues, New Zealand College of Chiropractic, December 2025
Brain networks after adjustments
In a pilot of 30 adults funded by Life University, one session of instrument-assisted chiropractic adjustment, compared with a sham, was followed by higher resting connectivity across several large brain networks on EEG (recordings of the brain's electrical activity), with no difference in heart rate variability. The authors call for a fully powered trial to confirm the effect and its clinical meaning.
Achieved whena fully powered randomized trial by a group without ties to the funder finds the same pattern and links it to a clinical outcome.
Claimed by
S. Sullivan and colleagues, Life University, October 2026
Open
01
Unreported trials, every registry
A public list, rerun every year, of every registered trial of spinal manipulation or chiropractic care in the registries that feed the WHO International Clinical Trials Registry Platform, not only ClinicalTrials.gov, and whoever gave the treatment. Each trial that finished at least two years earlier is marked as published, results posted on the registry, or unreported, with its condition (back or neck pain, other joint and muscle problems, or non-musculoskeletal), sponsor, country and whether the published main outcome matches the registered one.
The briefCredit · unclaimed02
Recheck non-spine claims, all languages
An update of the Global Summit review to 2026 that takes in trials in every language, not only English, of spinal manipulation for non-musculoskeletal conditions such as colic, asthma, high blood pressure and period pain, appraised with the same checklist, with a verdict for each condition.
The briefCredit · unclaimed
Open 01 · the brief
Unreported trials, every registry
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public list, rerun every year, of every registered trial of spinal manipulation or chiropractic care in the registries that feed the WHO International Clinical Trials Registry Platform, not only ClinicalTrials.gov, and whoever gave the treatment. Each trial that finished at least two years earlier is marked as published, results posted on the registry, or unreported, with its condition (back or neck pain, other joint and muscle problems, or non-musculoskeletal), sponsor, country and whether the published main outcome matches the registered one.
Why it is still open
Two counts exist, both made once and on ClinicalTrials.gov alone. Robert Côté, Stephen Perle and Derek Martin of the University of Bridgeport found 171 completed trials under the words "chiropractic" or "chiropractor", of which 102 (59.6%) had been published by 29 May 2020 and 25 (14.6%) had posted results on the registry. Breanne Wells and Dana Lawrence, searching only trials with US sites in 2018 and 2019, found that 35 of 56 published "chiropractic" trials and 49 of 72 "spinal manipulation" trials supported the treatment. Trials registered only in other registries, such as the Australian New Zealand Clinical Trials Registry or ISRCTN, were counted by neither. A 2024 review of 167 published manual therapy trials for low back pain found only 52.1% registered at all. A Europe PMC search in November 2026 found no count across registries and no repeat.
Built with what exists
the ClinicalTrials.gov API, the WHO International Clinical Trials Registry Platform's search portal and export, the Australian New Zealand Clinical Trials Registry and ISRCTN; the Europe PMC and Crossref APIs to search for each trial's papers by registration number, title and investigator; an AI model to propose matches, such as the open-weight Mistral Large 4 (see the Note "A trillion open parameters"); two people to confirm every "unreported" call.
First step
Rerun Côté's ClinicalTrials.gov search for "chiropractic" and "chiropractor" to today, then run the same words and "spinal manipulation" in the WHO platform and count how many trials appear there and not on ClinicalTrials.gov.
Done looks like
Every eligible registration entered from at least four registries; the share unreported after two years, overall and by registry, condition type, sponsor and treating profession; the Côté and Wells figures reproduced within 5 percentage points from the same ClinicalTrials.gov search; a second person checks 10% of entries with at least 95% agreement; the investigators of unreported trials contacted and their replies logged; the whole count rerun a year later.
Open 02 · the brief
Recheck non-spine claims, all languages
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An update of the Global Summit review to 2026 that takes in trials in every language, not only English, of spinal manipulation for non-musculoskeletal conditions such as colic, asthma, high blood pressure and period pain, appraised with the same checklist, with a verdict for each condition.
Why it is still open
The summit searched to 15 May 2019 and admitted only English-language trials, and the immune-function review searched to 15 April 2020. A Europe PMC search for reviews of manipulation for non-musculoskeletal conditions published since June 2021 found surveys, website studies and case reports but no update.
Built with what exists
the databases the summit named (MEDLINE, Embase, the Cochrane Central Register of Controlled Trials, CINAHL and the Index to Chiropractic Literature) with the language limit removed; an AI model to translate and screen non-English papers, such as Mistral Large 4; the SIGN checklist for randomized trials and the SWiM reporting guideline the summit used.
First step
Rerun the summit's MEDLINE search through Europe PMC from May 2019 to now with no language filter, and count the non-English trials it adds.
Done looks like
The search plan published before screening; every trial found, in any language, listed with its appraisal; two reviewers agree on at least 90% of eligibility calls before discussion; for each condition, a stated finding on whether any acceptable trial since 2019 or in another language changes the summit's verdict.