Chiropractic Care: Attempting a Risk–Benefit Analysis

There are about 5000 chiropractors in active practice in North America, predominantly treating spinal problems, particularly neck and back pain.1 Worldwide, single-year prevalence rates for use of chiropractors among the general population vary from 10% in Austria to 33% in the United Kingdom.2 This level of popularity renders a critical risk–benefit analysis of chiropractic care a desirable exercise.

EFFECTIVENESS

Several evaluations of the effectiveness of chiropractic treatment have been published (e.g., Shekelle et al.,3 Meeker and Haldeman4). Invariably, they include studies of manual therapies other than chiropractic treatments. Because the chiropractic approach is distinct from that of other professions,4 it is important to evaluate this approach separately.

In 1996, Assendelft and colleagues5 published a rigorous assessment of treatments performed by chiropractors. Their review included 8 randomized clinical trials of chiropractic for treatment of low back pain. All of these studies had serious flaws in their design, execution, and reporting. The authors found “no convincing evidence for the effectiveness of chiropractic for acute or chronic low back pain.”5

Since the publication of the Assendelft et al. article, several additional randomized controlled trials of chiropractic spinal manipulation for back pain have emerged.6–9 Results of these trials, which included a total of 741 patients suffering from acute, subacute, or chronic back pain, did not show an advantage of chiropractic over control treatments (i.e., physiotherapy, educational booklets, oral medications, acupuncture, or steroid injections). Thus, the conclusions drawn by Assendelft et al.5 6 years ago seem to be still valid today.

Of course, use of these data to evaluate the effectiveness of chiropractic presents problems for several reasons. First, few of the studies available to date have controlled for a potentially powerful placebo effect. A recent systematic review revealed no compelling evidence to suggest that chiropractic yields clinical effects that are distinct from those of placebo manipulation.10

Second, chiropractic treatment often comprises more than spinal manipulation.4 It is therefore difficult to judge the relative contribution of any element in this “chiropractic treatment package.” Most of the trial data just described refer to chiropractic spinal manipulation.

Third, chiropractors do not treat only back pain. Yet, back pain is the indication that has been researched more thoroughly than any other condition, and the evidence for other complaints is (even) less encouraging.4 Thus, a preliminary conclusion as to the effectiveness of chiropractic is that its benefits are by no means certain.

SAFETY

And what about the risks? Prospective investigations into the risks of chiropractic are scarce. The methodologically best studies show that mild, transient adverse effects such as localized pain are experienced by about 50% of all chiropractic patients.11 In addition to such minor events, dramatic complications have been noted with some degree of regularity.12 These complications typically involve upper spinal manipulation, which has been associated with cerebrovascular accidents. To date, it has not been possible to identify risk factors.13 As a result, essentially everyone receiving chiropractic treatment is at risk.

The question is, How large is that risk? Estimates by chiropractors of the risk of serious events vary from 1 in 400 000 to 1 in 3.85 million cervical spine manipulations.13 These estimates are based on certain assumptions, such as that complications are underreported. However, rates of underreporting can be as high as 100%14; this, in turn, implies that the existing estimates are meaningless.

The most compelling data in this respect originate from a recent Canadian population-based, nested case–control study conducted by Rothwell et al.12 That study matched 582 patients who had experienced vertebrobasilar accidents with control subjects (i.e., nonpatients) who did not have a history of stroke. Patients younger than 45 years who had experienced a vertebrobasilar accident were 5 times more likely than control subjects to have visited a chiropractor in the preceding week and 5 times more likely to have made more than 3 cervical treatment visits in the preceding month. No significant associations with past-week receipt of chiropractic services were found among older patients.

The Rothwell et al. study also indicated that for every 100 000 chiropractic patients below the age of 45 years, approximately 1.3 cases of vertebrobasilar accidents attributable to chiropractic treatment would be observed within 1 week of treatment.12 It seems to follow that the frequency of serious complications of chiropractic has been underestimated in the past.

One could also argue that serious complications mostly involve upper spinal manipulation and that these data should not be associated with the preceding evidence on effectiveness, which relates to low back pain. However, chiropractors view the spine as a functional entity; thus, they frequently manipulate the spine at levels in which subluxations, or partial dislocations, are detected regardless of the location of pain. In other words, patients with low back pain often receive upper spinal manipulation.

CONCLUSIONS

Where does all of this leave us when attempting a risk–benefit analysis? On one side we have uncertain benefits; on the other we must consider common nonserious adverse effects as well as serious complications that occur at an unknown rate. Given this situation, a tentative risk–benefit analysis cannot produce a positive result. The conclusion must therefore be that, according to the evidence to date, chiropractic spinal manipulation does not demonstrably do more good than harm. In view of the incompleteness of our current knowledge and the popularity of chiropractic, research into this complex area should be intensified.