According to a new systematic review, serious complications of spinal manipulation are rare, but 33-60% of patients experience milder short-term adverse effects such as increased pain, radiation of pain, headaches, vertigo and even loss of consciousness. The study, published in the journal Spine, involved searching PubMed and the Cochrane Library for the years 1966 to 2007. They identified additional studies by hand searching. They looked for all articles that reported adverse effects associated with chiropractic irrespective of type of design. They omitted any reports where patients had underlying diseases (osteogenesis imperfecta, expansive vertebral hemangioma, osteoporotic fracture, etc.) that predisposed them to complications with manipulation.
They found 46 pertinent studies:
- One randomized controlled trial
- Two case-control studies
- Six prospective studies
- Twelve surveys
- Three retrospective studies
- 115 case reports
They recognized that “the heterogeneity of the study designs did not allow conducting a formal meta-analysis.” But they did the best they could to make sense out of what they found.
It is surprising that after over a century of chiropractic manipulation, only one randomized controlled trial has attempted to evaluate its safety. And that study was really designed to compare different chiropractic methods to each other: manipulation (high velocity low amplitude thrusts that cannot be resisted by the patient) vs mobilization (low velocity passive motion that can be stopped by the patient), with and without adjunctive measures of heat or electrical muscle stimulation. The study, published in 2005 by Hurwitz et al. in both a chiropractic journal and in Spine, was limited to patients with neck pain. Of 336 patients enrolled, 280 responded 2 weeks later to an adverse events questionnaire. 30% reported at least one adverse symptom, most frequently increased neck pain or stiffness. Adverse reactions were more frequent with manipulation than with mobilization, but the difference didn’t reach statistical significance. The study concluded
Given the possible higher risk of adverse reactions and lack of demonstrated effectiveness of manipulation over mobilization, chiropractors should consider a conservative approach for applying manipulation to their patients, especially those with severe neck pain.
Two case control studies
A study in Canada, published in Stroke in 2001 matched 582 cases of vertebrobasilar accidents (stroke) with controls. In patients <45 years old, those with VBA were 5 times as likely as controls to have visited a chiropractor in the previous week, and 5 times more likely to have made more than 3 visits for cervical treatment in the preceding month.
A second study published in Neurology in 2003 used a nested case-control design and found that vertebral artery dissections were independently associated with spinal manipulative therapy in the previous 30 days, even after controlling for neck pain. The authors advised,
Patients undergoing SMT should be consented for risk of stroke or vascular injury from the procedure. A significant increase in neck pain following spinal manipulative therapy warrants immediate medical evaluation.
6 prospective studies  involved having patients, chiropractors, or physiotherapists fill out questionnaires on a predetermined number of consecutive patients (usually between 10 and 15).
1. 20 New Zealand physiotherapists were approached; only 9 returned the forms, and only 1 participated for 3 weeks. Questionnaires were completed by physiotherapists. There was only one report of increased neck pain. This study was inconsistent with all the rest. It raises the question of whether manipulation by physiotherapists might be safer than manipulation by chiropractors, and whether they are actually using the same techniques.
2. 10 Norwegian chiropractors reported side effects in 1/3 of patients. 23% had local or radiating symptoms: 90% moderate or slight, 87% commencing on the day of treatment and 83% disappearing in 24 hours. Questionnaires were completed by the chiropractor.
3. 86 Swedish chiropractors were approached and 66 participated; 27% of patients did not participate and 5% were lost to follow-up. 44% reported adverse reactions, mostly local discomfort in the area treated. Questionnaires were completed by the chiropractor.
4. 146 Norwegian chiropractors were approached; 102 participated. Adverse reactions were reported in 55% of patients, with 64% occurring within 4 hours of treatment and 74% disappearing by 24 hours. Questionnaires were completed by the chiropractor.
5. 11 British chiropractors were approached; 9 participated. 74% of patients responded of which only 63% of the forms were complete. 53% reported adverse reactions; 44% of these occurred within an hour of manipulation. Questionnaires were completed by patients.
6. 59 Belgian manipulative therapists; questionnaires filled out by patients. 60.9% reported adverse reactions, mainly headache, stiffness, and local discomfort, most appearing within 4 hours of treatment and resolving in 24 hours.
It is interesting that not one of these studies was from the US, where chiropractic originated and where it is most popular.
12 retrospective surveys mainly involved asking chiropractors, neurologists, or other physicians to fill out questionnaires. A couple of studies were based on medical or insurance records. In all, 308 serious adverse effects were reported: 163 strokes, 26 myelopathies, 100 radiculopathies, 3 transient ischemic accidents, 1 acute subdural hematoma and 29 other cases not specified. Minor adverse reactions totaled 1337 cases, most of them vertigo (1218 cases) and diminished or lost consciousness. Most of these occurred within 24 hours of manipulation; 5 patients died and 80 were left with permanent neurologic deficits.
A retrospective study of medicolegal cases suggested that chiropractors may have failed to recognize a stroke in progress. “The sudden onset of acute and unusual neck and/or head pain may represent a dissection in progress and be the reason a patient seeks manipulative therapy that then serves as the final insult to the vessel leading to ischemia.” Strokes occurred at any point during the course of treatment and there was no dose-response relationship.
115 case reports included strokes (66), spinal fluid leak (5), spinal epidural hematoma (7), cauda equina syndrome (2), herniated disc (20), radiculopathy (7), myelopathy (3), diaphragmatic palsy (3) and pathologic fractures of vertebra (2).
Conclusions: Adverse reactions are common after spinal manipulation
The results of this new systematic review are consistent with previous reviews. Adverse reactions are common after spinal manipulation, but they are usually benign and transitory. The true risk of serious injury is not known. Estimates of the risk of stroke vary wildly from 1 in 20,000 manipulations to 1.46 in 10,000,000 manipulations. The authors point out that these numbers are speculative because they are based on assumptions about the total number of manipulations performed, and because the degree of underreporting is likely to be high. In one survey, 35 cases of neurologic complications were identified, none of which had been previously published. Another study  by the Stroke Council of the American Heart Association identified 360 unpublished case of arterial damage. Chiropractors and neurologists are likely to have different perceptions of the risk, because without systematic followup, chiropractors may not know when their patient has suffered a stroke, while neurologists see only patients with stroke. One interesting but not unexpected finding of this study was that questionnaires completed by patients reported a higher incidence of adverse reactions than questionnaires completed by chiropractors.
This study concluded that the “data are inconclusive in terms of incidence, but the risk of occurrence of serious adverse reactions should be assumed.” There is no way to know if one type of treatment is more dangerous than another because the studies almost never specify what chiropractic maneuver was performed. But there is a hint that mobilization might be safer than manipulation, and there is a plausible rationale.
Screening protocols have been developed that attempt to identify patients at higher risk of stroke, but they have not proven to be useful. Their sensitivity and specificity are low. An intriguing study using PET scans concluded that cerebellar hypoperfusion (decreased blood flow to the base of the brain) may occur after cervical spinal manipulation. This could explain reports of headache, dizziness, vertigo, nausea, blurred vision, etc. It also raises the possibility that some of the minor reactions were strokes that almost happened but didn’t.
One might ask, if half of patients have adverse effects, why do they keep coming back? There are plenty of psychological factors that could explain that. Many alternative providers tell patients that increased symptoms mean the treatment is working and they have to feel worse before they can get better. And one practice-building technique taught to chiropractors is not to ask whether the original symptom is better, but to ask “What’s better today?” If they can find anything positive, like sleeping better or improved appetite, they emphasize that improvement and distract the patient’s attention from the fact that their original complaint has not improved.
We simply don’t have enough good data to quantify the risks of chiropractic treatment. It’s hard to understand why we don’t have good data after all this time. It would seem to be in the best interests of chiropractors and their insurance companies to get the facts. Good data would be easy to obtain by establishing a database specifying the exact intervention and contacting patients a day or two later and also a month after the treatment to ask about adverse effects or subsequent diagnoses of stroke. It could be compiled electronically and data pooled for a large number of chiropractors. It would cost next to nothing and could be carried out by office assistants as part of their routine duties.
Patients have the right to know. Apart from the risk of stroke and other serious outcomes, if there is a 50% chance I will feel worse after a treatment, I would like to be warned.
And as the authors point out, the question of risk is all that more important because we don’t have a “robust demonstration” of the effectiveness of these treatments. Risk alone is meaningless: it must be balanced against benefits to make a risk/benefit assessment. They say “Although the list of indications for which chiropractic is recommended is enormous, there is insufficient published evidence to support or refuse the efficacy of this treatment modality.”
The bottom line: chiropractic manipulations, especially neck manipulations, carry a small risk of serious consequences, a large risk of minor adverse effects; and, depending on the indication, there is little or no evidence that they are effective.
 Rivett DA, Milburn P. A prospective study of complications of cervical spine
manipulation. J Manipulative Physiol Ther 1996;4:166–70.
Senstad O, Leboeuf-Yde C, Borchgrevink CF. Side-effects of chiropractic
spinal manipulation: types, frequency, discomfort and course. Scand J Prim
Health Care 1996;14:50–53.
Leboeuf-Yde C, Hennius B, Rudberg E, et al. Side effects of chiropractic
treatment: a prospective study. J Manipulative Physiol Ther 1997;20:
Senstad O, Leboeuf-Yde C, Borchgrevink CF. Frequency and characteristics
of side effects of spinal manipulative therapy. Spine 1997;22:435–41.
Barrett AJ, Breen AC. Adverse effects of spinal manipulation. J R Soc Med
Cagnie B, Vinck E, Beernaert A, et al. How common are side effects of spinal
manipulation and can these side effects be predicted? Man Ther 2004;9:
 Robertson JT. Neck manipulations as a cause for stroke. Stroke 1981;12:1
This article was originally published in the Science-Based Medicine Blog.