Focus on Alternative and Complementary Therapies
www.pharmpress.com/fact
Focus Alternat Complement Ther©2005 Pharmaceutical Press
Focus Altern Complement Ther 2004; 9: 107–10
Dana J Lawrence
While a large amount of interest and resources have been devoted to studying the use of spinal manipulation for low back pain, neck pain, although a common occurrence in clinical practice, has received far less attention. This must change; around 15–40% of men and 20–65% of women will suffer such pain at some point in their lives.1 Thus, neck pain is quite common, can be disabling and may accrue significant cost.2,3 Spinal manipulation is one common treatment, offered by chiropractors, physical therapists, physical medicine specialists and others.
The problem confronting those who use manipulation as a primary treatment is that this area has received little study. However, some evidence does exist and can be gleaned both from specific trials and from systematic reviews. The Australian National Health and Medical Research Council defined levels of evidence for use in assessing the strength of evidence.4 Level I evidence, the highest level, is ‘that obtained from systematic reviews of relevant randomized trials (with meta-analysis where possible)’, while level II evidence is defined as ‘that obtained from one or more well designed randomized controlled trials.’
Both levels of evidence exist to support the use of spinal manipulation for treating mechanical neck pain. Of three systematic reviews,5–7 one had a negative outcome7 and two found manipulation to be more effective than rest and analgesia.5,6 The assessment by Koes included 35 trials comparing manipulation to other treatments for neck and back pain, although far more studies looked at back pain (N of 5 for neck pain).7 The majority of these papers were found to have significant flaws in methodology and although the results showed promise they were not conclusive. However, Hurwitz found that in patients with subacute and chronic neck pain, five RCTs reported findings.6 It was possible to collapse the data from three of the studies to examine outcomes 3 weeks after initial treatment. Although Hurwitz was unable to reach statistical significance in his interpretation, he concluded that it was ‘more than 90% probable that the difference in means favours the groups treated with manipulation.’ Furthermore, manipulation, along with mobilisation and physiotherapy, was more effective than usual medical care or the use of muscle relaxants in achieving short-term relief of pain. There was not enough information concerning acute pain or long-term relief for Hurwitz to draw a conclusion.
In the paper by Gross et al., the condition of interest was neck pain without neurological deficit.5 The paper is a meta-analysis in which manipulation was one of the treatments studied. Four papers were retrieved that looked at manual therapy alone, while two looked at manipulation. Here, there was no way to combine the data, which in some studies did show significant results but in others did not. When the authors turned their attention to the use of manual therapy combined with other forms of therapy, six trials were retrieved; five of these could be combined to draw conclusions. The pooled effect size was found to be –0.6, indicating a reduction in the outcome of interest, pain. When three of these studies were then combined at the 8-week period, the effect size was –0.5, still showing a reduction in pain. Thus, the authors concluded that there was evidence for a short-term benefit for the use of manipulation for neck pain.
Gross and colleagues’ work at Cochrane found that with 33 trials in their study, a single session of manipulation and mobilisation showed a non-significant benefit in pain relief compared to placebo or control or other treatments, and that multimodal care showed greater benefit than waiting list control for a variety of outcomes (pain reduction, improvement in function and global perceived effort), with the common elements being manipulation and mobilisation plus exercise; however, there was some evidence that compared to other treatments the multi-modal care offered a similar but not significantly better benefit.8
Hoving et al. examined 25 review articles that evaluated conservative care for neck pain.9 Of these, 12 were systematic reviews, and not all examined the use of spinal manipulation. However, what was interesting was that their study showed that the concordance among the reviews varied, some used poor methodology and there was little evidence available from primary studies from which one could draw conclusions. They also point out one significant challenge in looking at such studies, i.e. ‘unfortunately, current classifications for neck disorders have no clinimetric standards,’ making any type of analysis that much more difficult, as one has few means of comparing groups from different studies.
One cannot escape the fact that there is a paucity of evidence drawn from systematic reviews and meta-analysis, since so few studies exist. Using a best-practice approach combining the data from those reviews that do exist, those studies that have been completed and the clinical judgement from various professions that treat neck pain, one can support the use of manipulation for neck pain. While all this evidence may be inconclusive, there is some short-term evidence that the effect may be similar or even superior to other commonly used treatments for chronic neck pain. There is currently no convincing evidence to suggest that spinal manipulation would not be of benefit; it is a viable treatment option far likelier to provide benefit than harm.
That harm, however, must be taken into account. As Rosner has noted, during the past 10 years ‘the issues of cerebrovascular accidents (CVAs) and spinal manipulation have become linked in a debate of ever-increasing intensity.’10 It is well beyond the scope of this short article to delve into the literature and debate on this issue, but its importance to those practising manual procedures cannot be overstated. The number of CVAs related to manipulation have been estimated as ranging from 1 case in 400 00011 to no cases in 5 000 000.12,13 The meta-analysis by Hurwitz et al., one of the most thorough examinations of this subject, estimated the risk of serious complication at 0.64 per 1 000 000 and the risk of death at 0.27 per 1 000 000.6 Certainly, the risk estimates for cervical manipulation are significantly less than those associated with more common medical procedures and lifestyle activities.14 Again, the scope of this short paper precludes discussions related to involvement of homocysteine,15 spontaneous dissection,16 arterial fragility and biomechanical factors,17 as well as reporting irregularities.18
In two papers trying to look for risk factors for spinal manipulation, Haldeman et al. were unable to determine discrete risk factors.19,20 What Haldeman did note was that: vertebral artery dissections were extremely rare, there were insufficient cases to make a detailed analysis possible, spinal manipulation did not appear to be the precipitating factor in most dissections and that the only way to test the hypothesis that a given disorder poses a risk for dissection would be to review a large group of patients who suffered a dissection and match them to a control, which is vastly difficult to do.
Edzard Ernst
This question is important for at least two reasons: neck pain is common and (chiropractic) spinal manipulation is often used to treat it. But also let us remember, neck pain is not a life-threatening condition, it is often a self-limiting complaint, and treatments exist that are devoid of risk. These considerations should, I think, influence our clinical decisions.
The most recent systematic review had narrow inclusion criteria (e.g. trials including interventions other than chiropractic spinal manipulation were excluded) and thus included only four RCTs.1 None of them convincingly demonstrated the superiority of spinal manipulation over a range of control interventions (e.g. mobilisation and exercise). My conclusion was that chiropractic spinal manipulation is not more effective than conventional exercise treatment.
There is general agreement (supported by data) that Cochrane reviews are of the highest quality and thus contribute the most definitive evidence. The Cochrane review on the present topic had very broad inclusion criteria (e.g. including trials with multi-modal interventions),2 thus we have to be very careful when citing it specifically in relation to spinal manipulation. In its results section, the reviewers dedicate several paragraphs to the heading ‘manipulation alone’. These make very clear statements: ‘Four RCTs assessed the effect of a single session of manipulation. When compared to a control there was moderate evidence that single sessions did not result in short-term pain relief … Five trials assessed the effect of 6–20 sessions of manipulation … against various comparisons … In every case, the results were negative … Three trials found no difference in short- and intermediate-term pain relief when manipulation was compared to mobilisation …’ This important piece of evidence is often misquoted. It yields positive results for ‘multimodal therapy’, e.g. manipulation plus mobilisation or exercise or physical therapies such as heart application. Proponents of spinal manipulation therefore tend to represent this as positive evidence for their therapy. This is not correct!
Proponents of spinal manipulation will rightly point out that the absence of evidence is not the same as evidence of absence of an effect. On the other hand, critical thinkers would argue that, even if an effect can one day be demonstrated, it could be due to a placebo response.3 Spinal manipulation has several of the characteristics that make a ‘powerful placebo’. But who cares? Does it really matter whether a patient gets better because of a specific or a non-specific response? Perhaps not from a practical, clinical point of view – provided the placebo is safe!
In particular, rotational manipulation of the cervical spine has been repeatedly associated with disastrous consequences: arterial dissection often followed by stroke, sometimes followed by significant and persistent neurological deficits or death.4 Table 1 summarises such cases published in the last two years.5
Chiropractors and other spinal manipulators are well aware of these facts but argue that such tragedies are extreme rarities: millions of spinal manipulations performed every year are not followed by adverse effects. Thus the incidence, they believe, is lower than that for aspirin, for example. I beg to differ. First, the argument ignores mild, transient adverse effects which, according to the best evidence available to date, occur in about 50% of patients.6 If the benefit of spinal manipulation is small or uncertain (see above) even minor but frequent adverse effects must shift the balance in the area of negativity. Second, the rate of under-reporting of severe adverse effects, such as stroke, could be huge simply because patients thus affected are unlikely to come to their next chiropractic appointment. In our own case series, under-reporting was precisely 100%.7 Safety comparisons between treatments such as aspirin for which adverse effect reporting schemes exist and those for which this is not the case (e.g. spinal manipulation) are nonsensical. The estimates of incidence figures which chiropractors regularly produce are therefore unreliable to a degree of being misleading. This means that the risk of serious harm is simply not known.
Table 1. Cases of rotational manipulation of the cervical spine associated with serious consequences published in the last two years
| Patient | Therapist | Intervention | Adverse event/complication/symptoms | Outcome |
|---|---|---|---|---|
| 31-year-old woman with neck pain | Chiropractor | ‘Rapid rotary manipulation’ | Left vertebral artery dissection Extensive cerebellar infarction | Full recovery |
| 64-year-old man | Chiropractor | ‘Gentle manipulation of neck’ | Dissection of left internal carotid artery Parietal stroke | ndp |
| 51-year-old man | Chiropractor | ‘Rotary neck manipulation’ | Right internal carotid artery dissection Subcortical stroke Horner’s syndrome | ‘Significant recovery’ after bypass surgery |
| 32-year-old woman | ‘Native American healer’ (bone setter) | Cervical SM | Acute infarction in the middle left cerebellar hemisphere and vermis (dissection left vertebra artery) Stroke with vertigo, ataxia and vomiting | Minimal neurological deficit at 5 months’ follow-up |
| 46-year-old man | Traditional Chinese ‘bone setter’ | Cervical SM | Incomplete cervical cord injury | Minimal neurological deficit at 28 months |
| 67-year-old woman with neck pain | Not stated | Cervical SM | Epidural haematoma | Full recovery after surgery |
| 34-year-old man with ‘neck cramping’ | Chiropractor | Cervical SM | Bilateral internal carotid and vertebral artery dissections | Persistent left-sided hemianaesthesia |
| 46-year-old woman with neck pain | Chiropractor | ndp | Left-sided tinnitus Left internal carotid artery dissection | ndp |
| 40-year-old woman | Chiropractor | Cervical SM with external axial tension and rotation | Instant pain followed by headache, nausea, vomiting, double vision, dural tear | Full recovery |
| ‘Young patient’ | Chiropractor | Cervical SM | Incomplete Wallenberg syndrome | ndp |
ndp = no details provided; SM = spinal manipulation.
Spinal manipulation for neck pain is a treatment with unknown benefits and unknown harm. Because of this and the fact that serious risks are on record, a responsible risk–benefit assessment cannot ignore the risks and cannot come out in favour of spinal manipulation. Remember the supreme law in medicine: first do no harm. Other therapies for neck pain exist, e.g. exercise, which are supported by at least as good evidence for benefit and which are at the same time free of significant risks. The inescapable recommendation based on the best evidence available today is to use exercise rather than spinal manipulation as a treatment for neck pain.
