International Journal of Clinical Practice

Review Criteria

Randomised studies testing the effectiveness of spinal manipulation for infant colic were evaluated.

Infant colic occurs in about 10–30% of all infants with a peak around the age of 6 weeks (1). The syndrome is characterised by paroxysms of inconsolable crying, often accompanied by flushing of the face, meteorism, drawing‐up of the legs and flatulence (2). The aetiology of infant colic is not clear and a range of causes are being discussed: lactose intolerance, dysmotility, gastro‐oesophageal reflux, hormonal imbalances, aberrant microflora of the gut, feeding disorders, food hypersensitivity and psychological factors (2). The optimal therapy is not yet known and many experts believe that a multifactorial management strategy might be the best option (3).

Chiropractic spinal manipulation is being recommended as a treatment of infant colic by professional bodies of chiropractic. Case reports and case series also seem to suggest that spinal manipulation is an effective therapeutic option (4, 5). One retrospective survey suggested that 91% of patients show a positive response (6). However, a health technology assessment concluded that ‘there is no convincing evidence that spinal manipulation alone can affect the duration of infantile colic symptoms’ (7). This report included studies up to April 2003. As new data have become available since then, it requires updating. The aim of this systematic review was to critically evaluate the data from randomised clinical trials of chiropractic spinal manipulation as a treatment of infant colic.

Methods

Electronic searches were carried out (January 2009) in the following databases: Amed, Embase, Medline (using the Ovid interface) and Cinahl (using the Ebsco interface). The search terms were constructed over two concepts: spinal manipulation and infant colic. Our own, extensive department files were hand‐searched and the bibliographies of review articles and other relevant literature were also screened. No language or time limitations were imposed. The abstracts of the articles thus located were screened in End Note to remove duplicates and overtly irrelevant studies.

To get included, a clinical trial had to be randomised, test the effectiveness of spinal manipulation, focus on infants with the clinical diagnosis of colic and include outcome measures which are of clinical importance. Any type of control intervention was admissible. Studies of chiropractic or osteopathic treatments not including spinal manipulation were excluded. Trials which failed to include clinical end‐points or which were not fully published (e.g. available as an abstract only) were also excluded (8).

Key data of the included trials were extracted according to prespecified criteria (Table 1). Data extraction was performed by two independent reviewers. The methodological quality of all reviewed studies was estimated using the Jadad score (9). Again this was performed by two independent reviewers. A meta‐analytical approach was considered because of the clinical heterogeneity of the primary studies.

Results

Three randomised clinical trials (RCTs) met the above criteria and were included for further analysis (Figure 1) (10-12). Key data from these studies are summarised in Table 1. Two (10, 12) of the four RCTs imply that chiropractic spinal manipulation might be effective. However, these data require careful scrutiny. The methodological quality of all RCTs was low (Table 1). Apart from one study (11), all trials had very low sample sizes. Two of the studies (10, 12) were designed such that placebo‐effects were not controlled for. One RCT was an equivalence trial comparing two interventions of unknown effectiveness (12). None of the studies employed validated outcome measures to quantify the therapeutic success.

Discussion

Collectively these RCTs fail to demonstrate that chiropractic spinal manipulation is an effective therapy for infant colic. The largest and best reported study failed to show effectiveness (11). Numerous weaknesses of the primary data would prevent firm conclusions, even if the results of all RCTs had been unanimously positive.

The trial by Wiberg et al. (10) did not attempt to blind the infants’ parents who acted as the evaluators of the therapeutic success. The paper provides little details about the recruitment process, but it is fair to assume that patients were asked to participate in a trial of spinal manipulation. Thus one might expect a degree of disappointment in parents of the control group whose children did not receive this treatment. This, in turn, could have impacted on the parents’ subjective judgements. In any case, there is no control for placebo effects which can be very different for a physical intervention compared with an oral placebo – dimethicone was administered as a placebo and the authors stress that it is ‘no better than placebo treatment’.

The RCT by Olafsdottir et al. (11) is by far the best‐reported study of all the included RCTs. In many ways, it is a replication of Wiberg’s investigation (10) but on a larger scale with twice the sample size. It is the only study where a serious attempt was made to control for the placebo effects of spinal manipulations. For these reasons, its results seem more reliable than those of the other RCTs.

The RCT by Browning and Miller (12) is a comparison of two manual techniques both of which are assumed by the authors to be effective. Thus it is essentially a non‐inferiority trial. Yet, it is woefully underpowered for such a design. Even if it had the necessary power, its results would be difficult to interpret because none of the two interventions have been proven to be effective. Thus, one would still be uncertain whether both interventions are similarly ineffective or effective. As it stands, the result simply seems to demonstrate that symptoms of infant colic lessen over time possibly as a result of non‐specific therapeutic effects, the natural history of the disease, concomittant treatments, social desirability or a combination of these factors.

Other reviewers have arrived at similar conclusions. The above‐mentioned Canadian health technology assessment found that there is no convincing evidence (7). Given that spinal manipulation is not risk‐free (13, 14) some experts even warn: ‘physicians should be cautious about spinal manipulation in infants and should discourage families from treating infantile colic with spinal manipulation’ (15). In addition, we might also consider the costs. Three to six sessions of spinal manipulation, the amount of treatments used in the included RCTs (10-12), would cost £100 to £300. These are considerable expenses, and one would need comparative cost‐evaluations to see how they compare with conventional care for infant colic.

This review has several limitations that deserve to be mentioned. Even though the search strategy was thorough, there is no guarantee that all the RCTs ever conducted were retrieved. The field of ‘alternative’ medicine is plagued by negative publication bias (16). If it applies also to a specific area, negative studies might have remained unpublished and the overall picture painted here might be too positive. The paucity and the low quality of the primary studies are further, serious drawbacks. They limit the conclusions that can be drawn from this review.

Future studies should aim at minimising the bias that may have tainted the evidence available to date. In particular, trials need to have sufficient power, control for non‐specific effects, blind parents or other evaluators of therapeutic success and employ validated outcome measures. None of this may be easy to achieve but all of it is feasible.

In conclusion, the current evidence from RCTs does not show that chiropractic spinal manipulation is an effective treatment for infant colic. Further rigorous studies may be warranted.

Acknowledgements

I am grateful to Kate Boddy for the literature searches and to Katherine Hunt for help with data extraction and quality assessments.