Review: Evidence of the effectiveness of spinal manipulation for back and neck pain is limited
ACP J Club. 1992 Mar-April;116:48. doi:10.7326/ACPJC-1992-116-2-048
Koes BW, Assendelft WJ, van der Heijden GJ, Bouter LM, Knipschild PG. Spinal manipulation and mobilisation for back and neck pain: a blinded review. BMJ. 1991 Nov 23;303:1298-303.
To evaluate the efficacy of spinal manipulation and mobilization for back and neck pain by means of a review of randomized controlled trials.
A MEDLINE search for the period 1966 to 1990 (using keywords backache, musculoskeletal diseases, joint diseases, manipulation, osteopathy, chiropractic, evaluation studies, and outcome and process assessment) was supplemented by references given in relevant articles.
Randomized controlled trials including patients who had been treated with spinal manipulation for back or neck pain were eligible.
The methodologic quality of 30 studies of back pain and 5 of neck pain was scored by 2 reviewers blinded to the source and outcome of each trial. Disagreements between the reviewers (276 instances in 1400 items; 20%) were resolved by consensus. The results of each study were classified as "positive" or "negative" based on the original investigators' conclusions about spinal manipulation.
Studies compared osteopathic, rotational, Cyriax, Maitland, Janda, and other methods of manipulation with placebo manipulation, detuned short-wave diathermy, physiotherapy, medical treatment, or bed rest. 6 trials had > 50 patients in the smallest group (range 6 to 384 patients). On quality ratings, no trial scored ≥ 60 points of a possible 100. Of 4 trials scoring ≥ 50 points, 1 reported positive results for Bourdillon manipulation compared with placebo for chronic low-back pain; 1 reported no difference between 3 types of manipulation and diazepam for neck pain; and 2 found improvement in a subgroup with acute or chronic low-back pain of 2 to 4 weeks duration, in comparisons of osteopathic or rotational manipulation with physiotherapy. 9 trials scored between 40 and 50 points; 2 of these reported positive results; 6, negative; and 1, positive results in a patient subgroup. 14 studies reported effects at ≥ 3 months, and 4 of these trials showed a benefit of manipulation. Overall, 18 investigations found positive effects of manipulation for acute or chronic back or neck pain; 3 of these (17%) scored ≥ 40 points compared with 7 of 11 trials (64%) that found no positive effects.
Although some randomized controlled trials show results that favor manipulation for back and neck pain, the poor methodologic quality of the trials and their heterogeneity limit the strength of the conclusions that can be drawn.
Source of funding: Not stated.
Address for article reprint: Mr. B.W. Koes, Department of Epidemiology and Biostatistics, University of Limburg, P.O. Box 616, 6200 MD Maastricht, The Netherlands.
Methodologic studies such as this are important to the researcher and have a positive influence on the quality of clinical research. The practicing clinician, who may be unfamiliar with some of the nuances of the methods, may have difficulty interpreting the findings. For example, whether a study is classified as positive or negative is an assessment of statistical significance based on the conclusion of each study's author. Only 6 trials compared more than 50 persons in each group (studies likely to be of sufficient size to avoid false-negative findings). Of these, 5 had a positive result (83%), compared with 19 of 29 trials (66%) with fewer than 50 subjects in at least one group. It would be useful to the clinician to include a study rating factor for the power to identify a clinically important result.
In some cases, the weighting factors may not reflect the true validity of the study results. The maximum score for adequacy of follow-up period, for example, was only 5 out of 100. Of the 10 studies scoring 5, 8 were "positive," suggesting that this important factor may have been underweighted.
A major difficulty for the clinician is the heterogeneity of the studies. There is little commonality in patient groups, control groups, interventions, and outcome measures. It is difficult to apply the findings to one's own clinical practice.
Some trials have had promising results for the efficacy of manipulation. This review identifies inadequate research methods as a major factor limiting the strength of the conclusions that can be drawn from the studies. Some aspects of the review methods, however, may make the results appear to be less positive than they really are.
Edward S. Gibson, MD
Dofasco Medical CentreHamilton, Ontario, Canada
Edward S. Gibson, MD
Dofasco Medical Centre
Hamilton, Ontario, Canada