Focus on Alternative and Complementary Therapies
www.pharmpress.com/fact
Focus Alternat Complement Ther©2005 Pharmaceutical Press
Focus Altern Complement Ther 2008; 13: 41–2
To estimate the risk of serious and minor adverse events following chiropractic cervical manipulations.
Prospective national survey.
UK clinical practice of chiropractic.
A total of 377 UK chiropractors treating 19 722 patients with at least one cervical manipulation.
Manipulation of the upper spine, i.e. high-velocity/lowamplitude or mechanically assisted thrusts.
Serious or minor adverse events.
No serious adverse events were noted; minor adverse events were more frequent — the most common was immediate worsening of neck pain, which occurred in 1.7% of all patients.
‘Although minor side effects following cervical spine manipulation were relatively common, the risk of a serious adverse event, immediately or up to 7 days after treatment, was low to very low.’
E-mail: hthiel@aecc.ac.uk
This is a most laudable attempt to define how often serious adverse effects of cervical manipulation happen. To generate certainty, such studies need to be very large, must not lose patients to follow-up and have to rely on accurate reporting. The sample of this survey was sizeable but not large enough to exclude rare events. Thus the authors can only state that, at worst, the risk of serious adverse events within 7 days after manipulation is 2 per 10 000 treatment consultations. If the average patient has a series of 10 treatments, this risk could therefore be as high as 1 in 500 per patient. Given the nature of the risk, i.e. stroke or death, this is by no means negligible. The picture gets more complicated when considering the 698 treatment consultations of patients who failed to return for their next treatment. Theoretically some or most or all of these patients could have died of a stroke. Overlooking even one single serious adverse event would change the estimated incidence rates from this study quite dramatically.
In my view, the most confusing aspect about the results of this survey is the fact that the incidence of minor adverse events is so low. Previous studies have repeatedly shown it to be around 50%. The discrepancy requires an explanation. There could be several but mine goes as follows: the participating chiropractors were highly self-selected. Thus they were sufficiently experienced to select low-risk patients (in violation of the protocol). This explains the low rate of minor adverse events and begs the question whether the incidence of serious adverse events is reliable. Studies of this nature are very difficult to conduct such that we can trust the results. One of the problems is that one has to rely on the honesty of the participating therapists who could have a very strong interest in generating a reassuring yet unreliable picture about the safety of their intervention.
None declared.
Ernst talks about ‘certainty’ in research studies and a sample size that was not large enough to exclude rare events. Just what is ‘large enough’ to achieve certainty? Choosing the worst risk estimate (˜2 per 10 000 treatment consultations), Ernst infers that this is a cumulative risk as well as for a patient instead of a single treatment. There is no evidence to suggest that the risk is cumulative and increases with repeated exposure. Even if this was the case, our design would have accounted for this by including not only the first but subsequent treatment consultations. Of course, Ernst is correct in saying that an adverse event could have occurred in those patients lost to follow-up. However, by any reckoning 1.4%, particularly in a sample of 50 000 treatments, is not only acceptable but, in our view, strengthens rather than weakens the credibility of the risk estimates. Ernst uses the difference in this study’s rate of a single minor side-effect with those reported in others to argue that the recruitment of chiropractors and patients was biased and therefore could have resulted in an underestimation of risk.
When taking into account all minor side-effects observed in this study, this difference diminishes considerably. In endeavouring to provide research evidence and in debating its limitations, we should not lose sight of the bigger picture. In the UK alone, there is an estimated 4 million manipulations of the neck carried out by chiropractors each year. In the absence of any significant numbers of reported serious events (i.e. stroke or death), including those in which a causal link is based on extremely weak evidence, there can be little doubt that the risk is very low. In failing to find a serious event, the risk was estimated based on sample size. It can therefore be argued that this risk estimate is a conservative one, and that had it been possible to achieve a ‘large enough’ sample, we might then have been ‘certain’ of a very rare event.
