Research Review By Dr. Michael Haneline©

Date Posted:

June 2009

Study Title:

Safety of Chiropractic Interventions: A Systematic Review


Gouveia L, et al.

Author's Affiliations:

Department of Neurology, Hospital de Santa Maria, Lisbon, Portugal

Publication Information:

Spine 2009; 34(11): E405-13.

Background Information:

The authors attempted to provide background information on chiropractic, including a definition of spinal manipulation and mobilization “… manipulation (high velocity, low amplitude thrusts that cannot be resisted by the patient) and mobilization (low-velocity passive motion that can be stopped by the patient).” They provided a reference to the American Chiropractic Association’s website in support of this definition. However, the official ACA Spinal Manipulation Policy Statement (1) defines manipulation and mobilization quite differently, as follows:

1. Manipulation:
The definition of manipulation is multi-faceted. A manipulation is a passive manual maneuver during which the three-joint complex may be carried beyond the normal voluntary physiological range of movement into the paraphysiological space without exceeding the boundaries of anatomical integrity. The essential characteristic is a thrust—a brief, sudden, and carefully administered “impulsion” that is given at the end of the normal passive range of movement.

The “dynamic thrust” is the defining factor which distinguishes manipulation from other forms of manual therapy. The thrust technique can be low or high velocity. The most common characteristics of the adjustive dynamic thrust are a controlled force delivered with high velocity, in a specific direction or line of drive, at a regulated magnitude and depth. In short, manipulation is a passive dynamic thrust that causes an audible release (cavitation) and attempts to increase the manipulated joint’s range of motion.

2. Mobilization:
Mobilization is a non-thrust, manual therapy. It involves passive movement of a joint within its physiologic range of motion. This is approximately equivalent of the normal range of motion a joint can be taken through by intrinsic musculature. Active range of motion is motion which patients can accomplish by themselves. Mobilization is passive movement within the physiologic joint space administered by a clinician for the purpose of increasing overall range of joint motion.

The authors’ definitions conflict with the ACA’s, especially regarding the supposed lack of control that patients have while being manipulated. Anyone experienced in manipulation knows that the patient can resist being manipulated or request the doctor to refrain from manipulating them if they notice pain during the setup. Furthermore, the ACA definitions were not described accurately in this paper. The authors’ definition seems to portray patients as being vulnerable during manipulation.

The authors indicated that recent reviews on the effectiveness of chiropractic have concluded that the efficacy of spinal manipulation was not demonstrated for the treatment of “any condition”, citing a chiropractic clinical practice guideline that was published in the Journal of the Canadian Chiropractic Association (JCCA) (2). However, there was no such statement in the JCCA article. The guidelines actually stated in “Treatment recommendation 2: Based on all the evidence…we also recommend manipulation…for patients with acute or chronic pain…”

It is interesting that when providing background information on chiropractic, negative reviews by a known chiropractic detractor (namely Edzard Ernst) were cited, while positive reviews were ignored. This prejudicial handling of the evidence sets the tone for this entire review.

The purpose of the review was based on the following supposition: Since chiropractic care has not been subjected to formal efficacy and safety evaluations like pharmacological interventions have been, and which are demanded by national drug agencies, an investigation of its safety was called for.

Pertinent Results:

A literature search identified 151 potentially relevant articles, although 110 of them were discarded because the patients had an underlying disease that predisposed them to adverse reactions, there were double publications, and other reasons. An additional 28 articles were identified via hand searching. The authors ultimately found 46 articles on adverse events related to chiropractic care that were mostly case series.

Randomized controlled trials (RCT)
Only one RCT was included and the authors referred to it as “…the only randomized controlled trial published.” This may give the impression to an uninformed reader that only one RCT has ever been done in chiropractic. On the contrary, many RCTs have been conducted regarding chiropractic modalities, and most of them have commented on the number of adverse events that occurred and therefore should have been included in this review.

At least 1 adverse reaction was reported by 30% of the subjects in the cited RCT, mostly involving increased pain and headache. Although not statistically significant, patients who received manipulation were more likely to report an adverse reaction than those who received mobilization. No serious adverse events were reported.

Case-control studies
Two case-control studies were included in this review, although after performing a cursory search of PubMed, I was able to locate another case-control study (3) that was published during the time-frame of the Gouveia et al. search. The study was in conflict with the two that were included by the authors, leading one to suspect that this may be further evidence of either reviewer bias or ineptitude.

One of the included studies reported that VBA cases were 5 times more likely than controls to have visited the chiropractor within 1 week of their VBA among those who were less than 45 years of age. The other study showed that VA dissections were independently associated with spinal manipulation performed within 30 days of the event (OR 6.62).

The case-control study that was not included (4) looked at 7 mild mechanical traumas thought to be potential trigger factors for cervical artery dissection (CAD), including cervical manipulation. They found no significant association between any of the risk factors independently and CAD, even though prior manipulation was more common among CAD cases than among controls (10 vs. 5). However, mechanical risk factors overall that occurred less than 24 hours prior to symptom onset were significantly associated.

Prospective studies
Six studies were included, although the manipulation was performed by physiotherapists, osteopaths and manipulative therapists in two of them. It is not appropriate to include adverse events attributed to other types of practitioners in a study reporting on chiropractic safety.

The authors reported the high-end of the range of adverse reactions in these studies, leaving out the studies with lower reported figures. The given range was from 44 to 60.9% and the reported symptoms were transient, mild and benign. No serious adverse events were reported in any of the studies.

Another important prospective study was omitted from this review (4) that was published within the time-frame of the literature search. It was a multicenter study that compared risk to benefits among patients receiving chiropractic care for neck pain. The conclusion was that “Adverse events may be common, but are rarely severe in intensity. Most of the patients report recovery, particularly in the long term. Therefore, the benefits of chiropractic care for neck pain seem to outweigh the potential risks.” No severe adverse events were reported.

Retrospective studies
These studies entailed surveys that were conducted by neurologists and chiropractors. Many serious and non-serious adverse events were reported, including 5 deaths and 80 permanent disabilities. The authors gave a range of serious adverse events from 5 strokes/100,000 manipulations to 1.46 events/10,000,000 manipulations. The 5 strokes/100,000 manipulations ratio was incorrect though (see the Study Strengths/Weaknesses section below).

The retrospective studies conducted by neurologists had serious limitations and have been criticized in the past, yet the authors did not, as one would expect in a systematic review, comment on these drawbacks.

Case reports
The search revealed 115 cases that were reported between 1925 and 2006. Most of the cases involved strokes, although there were 5 cases of spinal fluid leak, 7 cases of spinal epidural hematoma, 2 cases of cauda equina syndrome, 20 of cases of herniated disc, 7 cases of radiculopathy, 3 cases of myelopathy, 3 cases of diaphragmatic palsy, and 2 pathologic fractures of vertebra.

Of course, case reports cannot be used to show cause-and-effect and one must always wonder if the person would have had the condition even without manipulation. This is especially apparent regarding the manipulation-stroke issue where it seems many of the strokes that have occurred following manipulation were probably not related, other than temporally. This is because the symptoms of cervical artery dissection (headache and neck pain) are often what prompt patients to seek chiropractic care. In many of these cases, the stroke would have occurred whether the patient received manipulation or not.

To make this point clear, a study by Cassidy et al. (5) found that patients were actually more likely to have gone to a medical doctor prior to their vertebrobasilar stroke than to a chiropractor. Certainly there would have been more patients who went to the chiropractor before their stroke, if manipulation was indeed causing them. Unfortunately, the Cassidy et al. article was not published until 2008, which was after the Gouveia literature search was carried out.

Clinical Application & Conclusions:

Being so riddled with flaws, one cannot apply any of the findings of this article to clinical practice. However, because it was published in such a respected medical journal, it will no doubt be read by many spine care practitioners. As a result, potential chiropractic patients may be dissuaded from seeking care, especially for their neck pain, because of it.

Readers of the Research Review Service are encouraged to become familiar with this review and be prepared to discuss it with their medical colleagues. A pro-active approach may be best, wherein the reader engages them in a face-to-face discussion about the pros and cons of this article.

Study Methods:

A literature search of PubMed and the Cochrane Library was conducted encompassing the years 1966 to 2007. Search terms included: chiropractic, adverse reactions, adverse events, safety, and spinal manipulation.

All articles that reported adverse reactions associated with chiropractic irrespective of type of design were included. The definition of an adverse event and the causality relationship with the chiropractic intervention was based on the judgment of the authors of each study. Not included were articles on clinical efficacy, the burden of disease, prevalence, epidemiology, cost, and other variables not directly related to safety.

Two of the reviewers independently screened the titles and abstracts of studies for eligibility. One author performed the data extraction, although a second author confirmed the results as to type of design, type of intervention, number of patients, and type and number of adverse reactions.

Study Strengths/Weaknesses:

There are no strengths in this study!

The weaknesses are almost too numerous to mention, but to list a few:
  • The most egregious error is a misquote of an article by Michael Haynes (6) who actually reported that “…there were perhaps fewer than five cases of manipulation-related stroke per 100,000 patients who had received cervical manipulation from a chiropractor.” The authors of the Gouveia review reported “5 strokes/100,000 manipulations” … a huge difference from what Haynes reported. Furthermore, the inaccuracy was repeated 4 times in this article, including the abstract. It has been reported that the typical chiropractic patient in North America is seen 12.8 times on average (7) and other studies have reported even more. Using the 12.8 figure, the statistic becomes fewer than 5 strokes per 1,280,000 manipulations.
  • The article purports to report on the safety of chiropractic interventions, yet some of the included studies involved physiotherapists and other manual therapists (e.g., osteopaths and physiatrists).
  • The number of articles found during the search minus the number that were excluded for various reasons does not tally. They started with 151 and 110 were eliminated which leaves 41. Another 28 were added from articles found in hand searches, which should total 69. However, the final tally was only 46. This may be explained by 23 of the articles retrieved during the hand searches being excluded, but it was not at all clear.
  • The article claimed to be a systematic review, but it did not incorporate most of the components needed to qualify as a one (e.g., the quality of studies was not rated, their inclusion/exclusion criteria were not clear, there was no legitimate literature synthesis, etc.).
  • The Methods section was much shorter than that of a typical systematic review, as well as being incomplete and vague.
  • The title and aim of the review suggested that its focus would be on the safety of chiropractic interventions without qualification, but it quickly shifted to the cervical manipulation-stroke controversy.
  • In the Discussion section, an article by Cagnie et al. that reported on cerebellar hypoperfusion that occurred following cervical manipulation was referenced. However, there was no mention of the Letters to the Editor that were written in response to the article which pointed out that the association was implausible and that their methods were not capable of ruling out confounding variables.
  • As a result of all of its flaws, the resulting Gouveia review was biased in support of the argument that since there is no evidence pointing to the effectiveness of chiropractic care yet there is some evidence pointing to its lack of safety, patients should not be referred to chiropractors for treatment until further studies have been carried out.

Additional References:

  1. ACA Spinal Manipulation Policy Statement. Updated 2003. Available from:
  2. Anderson-Peacock E, Blouin JS, Bryans R, Danis N, Furlan A, Marcoux H, et al. Chiropractic clinical practice guideline: evidence-based treatment of adult neck pain not due to whiplash. JCCA J Can Chiropr Assoc 2005;49(3):158-209.
  3. Dittrich R, et al. Mild mechanical traumas are possible risk factors for cervical artery dissection. Cerebrovasc Dis. 2007;23(4):275-81.
  4. Rubinstein SM, et al. The benefits outweigh the risks for patients undergoing chiropractic care for neck pain: a prospective, multicenter, cohort study. J Manipulative Physiol Ther. 2007 Jul-Aug;30(6):408-18.
  5. Cassidy J et al. Risk of Vertebrobasilar Stroke and Chiropractic Care: Results of a Population-Based Case-Control and Case-Crossover Study. Spine, 2008;33(4S):S176–83.
  6. Haynes MJ. Stroke following cervical manipulation in Perth. Chiropractic J Aust 1994 Mar;24(2):42-6.
  7. Hurwitz EL, Coulter ID, Adams AH, Genovese BJ, Shekelle PG. Use of chiropractic services from 1985 through 1991 in the United States and Canada. Am J Public Health. 1998;88:771-6.