A 2012 study published in the International Journal Of Clinical Practice “Assessing the risk of stroke from neck manipulation: a systematic review” concluded:
Conclusive evidence is lacking for a strong association between neck manipulation and stroke, but is also absent for no association.
Despite the uncertainty, they thought the association was strong enough to recommend informed consent be obtained and patients be warned that neck manipulation “may” increase the risk of a rare type of stroke.
A new study in the same journal, “Chiropractic and Stroke: Association or Causation?” applies Hill’s criteria of causation to the evidence and concludes that causality has not been determined. The author is Peter Tuchin, a senior lecturer in chiropractic at Macquarie University in Australia, and a known apologist for chiropractic. I agree with him that the existing evidence is inadequate to conclusively determine causality, but I think it supports a high probability of causality, and the alternate explanations he offers to exonerate chiropractors are questionable. And other factors should be considered, like the many “smoking gun” cases and whether there is any conclusive evidence of benefit to set against the possibility of risk.
Hill’s criteria of causation
Mark Crislip has explained Hill’s criteria here.
Hill’s criteria were applied to the chiropractic subluxation construct with disastrous results for the basic premise of chiropractic:
No supportive evidence is found for the chiropractic subluxation being associated with any disease process or of creating suboptimal health conditions requiring intervention. Regardless of popular appeal this leaves the subluxation construct in the realm of unsupported speculation. This lack of supportive evidence suggests the subluxation construct has no valid clinical applicability.
Tuchin’s study begins by critiquing published reports of chiropractic strokes, objecting that details are not furnished and other causes of stroke are not ruled out. Those are valid criticisms. Indeed, other causes of stroke can never be completely ruled out in any individual case, since vertebral artery dissection (VAD) is known to occur after any hyperextension of the neck, such as painting a ceiling or getting a shampoo at the beauty parlor, and it can even occur spontaneously without hyperextension. Delayed effects cloud the issue: in some of the reported cases, stroke occurred as long as 15 days after spinal manipulation therapy (SMT), postulating a clot that temporarily sealed the tear and later broke loose. That’s why we have to rely on controlled studies, such as the 2001 study that showed VAD was more common after SMT: patients under the age of 45 were 5 times more likely than controls to have visited a chiropractor in the preceding week.
Tuchin addresses each of Hill’s criteria. His explanation is confusing, since he includes comments that are not relevant to the criterion being considered, and he presents the same arguments under more than one criterion.
- Strength of association. He acknowledges that the association is there, but he argues that it is not very strong. Here he quibbles that some of the reports were of manipulations done by non-chiropractors, but that is only self-serving damage control that is irrelevant to the question of manipulation causing stroke. He misquotes a German paper as saying “there was a clear evidence that the dissection was present prior to the SMT” whereas it actually says “there was clear evidence or high probability.” [emphasis added] And he doesn’t mention the strong associations in the “smoking gun” cases where a stroke occurred on the chiropractor’s table immediately or shortly after manipulation.
- Consistency. He claims the studies are inconsistent, with some showing “a relationship” and others showing only “an association.” He doesn’t explain how a “relationship” is different from an “association;” the dictionary says they are synonyms. The only study he mentions is the Cassidy study (more about that later), which he misrepresents: it actually showed a strong association of SMT and stroke in those under 45. He doesn’t present any studies showing that there is no association. He admits that there is consistency with neck movement causing stroke, but not with SMT. However, SMT involves neck movement; so SMT is necessarily consistent as a cause.
- Dose-response relationship. He concludes that a dose-response relationship doesn’t exist. He bases this conclusion on the fact that patients can have many manipulations before the VAD occurs. And the fact that many of the manipulations associated with VAD were done by other non-chiropractic practitioners. This doesn’t make sense to me. I would think dose-response would have to be tested by other means. Perhaps the “dose” relates to the force used or the degree of extension and rotation rather than to who performed the manipulation or how many previous manipulations had been done without incident.
- Temporality. Exposure must precede outcome. Here he argues that a clear time-line has not always been established, and he questions whether some patients might have had stroke symptoms before visiting the chiropractor. That is speculation not supported by any evidence.
- Plausibility. He argues that the force exerted is not sufficient to cause a tear, and that modern manipulation techniques do not require full cervical spine rotation or extension. It is inconsistent to argue this after he has argued that VAD can occur with any neck movement. I think it is perfectly plausible that even “modern techniques” could result in the same degree of neck movement that is associated with painting ceilings or having a shampoo at a beauty parlor, both acknowledged causes of stroke. I suspect that some chiropractors are using more rotation, extension, and force than the author would like to believe. He doesn’t present any data to support his speculations.
- Other explanations ruled out. He argues that VAD can occur spontaneously, and that patients presenting to a chiropractor may have had prior trauma. He argues that prior manipulations without incident suggest that an adverse effect of the last manipulation means something had changed in the patient. (He doesn’t consider that something might have been different about the last manipulation itself, or that perhaps a weakened artery finally gave way after repeated stretching.)
- Experimental confirmation. Here he doesn’t address experimental confirmation at all. He only talks about possible precipitating factors. Admittedly, there is no experimental confirmation; but it could be argued that experimental confirmation is next to impossible and is not necessary to establish causality. Hill himself said that his criteria were aids to thinking about causality, not a list of requirements.
- Specificity. One in every hundred thousand people has a VAD each year, so he thinks chiropractors could expect to see 10 VADs for every million patients manipulated, even if manipulation didn’t cause the VAD. (I would argue that patients with stroke symptoms are more likely to go to an ER than to a chiropractor.) He also argues that many of the documented tears were not at the location that he thinks would be expected from manipulation, but that doesn’t necessarily mean that SMT was not a factor. We don’t have any data to compare tears after SMT to tears in patients who were not manipulated.
- Coherence. He doesn’t think a causal explanation coheres with existing theory and knowledge, in contrast to many others (including other chiropractors) who think it does. He offers an alternative hypothesis: that patients who were already symptomatic from a VAD sought chiropractic care, and manipulation may have dislodged an already existing clot. This is pure speculation not supported by any data. And it points out that chiropractors are not able to judge when manipulation is contraindicated.
His arguments boil down to these:
- Chiropractors didn’t do it: it was other practitioners who were less well trained who did the suspect manipulations. (But there are plenty of reports involving chiropractors.)
- SMT didn’t cause it: it was due to one of those other things that can cause VAD. (No evidence to support that claim.)
- Patients went to a chiropractor because they were already having symptoms of a stroke. (No supporting evidence, and a good reason to avoid neck manipulation.)
He places considerable emphasis on the Cassidy study. He says it “concluded that patients present to either a chiropractor or GP with neck pain because of their stroke already being present.” The Cassidy study does not say what he thinks it says. See Mark Crislip’s critique of that study. The idea that patients were already experiencing symptoms of a stroke was not a conclusion of the study, but merely a hypothesis of the authors that was not supported by their data.
The symptoms of a vertebral artery stroke are headache and focal neurological signs. Facial pain with numbness is the most common neurologic consequence. Other common neurologic signs are dizziness or vertigo, dysarthria or hoarseness, loss of pain and temperature sensation in the trunk and limbs, loss of taste, hiccups, nausea and vomiting, problems with vision, difficulty swallowing, and unilateral hearing loss.
Note that that neck pain is not even on the list. And although headache usually precedes the neurologic signs, many patients don’t seek care until the neurologic signs have developed; and I would guess that those signs would be more likely to send them to the ER than to the chiropractor. In fact, the Cassidy study found a strong association between visiting a chiropractor and having a stroke in the next 24 hours for patients under the age of 45 (odds ratio = 12).
Is neck manipulation beneficial?
A 2007 study purported to show that the benefits outweighed the risks for patients undergoing chiropractic care for neck pain. It was a prospective study, but there was no control group. They found that 2/3 of patients were improved at 3 and 12 months, but how many untreated patients would also have improved by then? 56% of patients reported adverse events and 13% reported these to be severe.
74-79% of patients with non-specific neck pain typically recover in a year, and even patients with cervical radiculopathy improve over time with only conservative treatments such as use of a collar and physical therapy.
Another study found a benefit of exercise combined with manipulation/mobilization but also of exercise alone.
A 2010 Cochrane systematic review found that mobilization was as effective as manipulation.
A 2004 Cochrane systematic review found that mobilization and manipulation were not beneficial alone, but were equally beneficial when used in conjunction with exercise.
It seems plausible that gentle mobilization techniques would be less likely to cause a vertebral artery tear than high velocity low amplitude (HVLA) manipulation techniques. This study showed that mobilization is less likely to cause adverse reactions in general.
In short, there is no evidence-based reason to prefer manipulation techniques to mobilization with exercise. In the absence of proven benefit, even a hypothetical risk is unacceptable.
“Smoking gun” cases
There are plenty of reported cases where patients suddenly developed stroke symptoms when a practitioner manipulated their neck. Until these cases are compiled, investigated, and explained otherwise, it is reasonable to assume a cause/effect relationship. Tuchin argues that it seems impossible for a thrombus to instantly form, dislodge, and travel to the cerebral cortex to cause a stroke within seconds of receiving SMT; but I would argue that if an existing clot were dislodged, symptoms would be almost immediate. Also, a tear can cause bleeding between the layers of the arterial wall, creating a hematoma that rapidly expands to directly occlude the artery.
What is most worrisome is that some of those patients got neck manipulations for symptoms in other parts of the body unrelated to the neck, or even for no symptoms at all (maintenance adjustments as a preventive health measure). A majority of patients consulting a chiropractor for any reason will be subjected to neck manipulation. Children get neck manipulations for problems like ear infections. Upper cervical (NUCCA) practitioners only treat the top vertebra of the neck in every patient. The whole premise of upper cervical chiropractic is nonsensical.
Tuchin correctly concludes that the evidence that SMT causes strokes is not definitive and is not supported by all of Hill’s criteria of causation. But he tries to make his case with fallacious arguments and questionable claims rather than with scientific data, and the whole article smacks of apologetics for chiropractic. While the existing evidence is not definitive, it supports the strong probability of a causal relationship, especially considering the many “smoking gun” cases. Since neck manipulation has not been demonstrated to be effective, even the suspicion of a risk is reason enough to reject the treatment. Alternatives like gentle mobilization, physical therapy, and exercise are better choices. And while everyone agrees that SMT is contraindicated in the presence of a VAD, chiropractors have demonstrated their inability to predict which patients are at risk. The safest course is to avoid neck manipulation altogether.
This article was originally published in the Science- Based Medicine Blog.