Research Review By Dr. Michael Haneline ©

Date Posted:

September 2010

Study Title:

Adverse events and manual therapy: A systematic review


Carnes D, et al.

Author's Affiliations:

Barts and the London School of Medicine and Dentistry, Centre for Health Sciences, London, UK.

Publication Information:

Manual Therapy 2010; 15: 355-63.

Background Information:

Manual therapies comprise a number of hands-on techniques, including massage, mobilization and manipulation. A great deal of research has been generated in support of these interventions for a variety of musculoskeletal disorders, which has resulted in favorable endorsement in recent treatment guidelines (e.g. NICE Guidelines [1]).

However, some authors have expressed concerns about the risk of adverse events associated with manual therapies, especially with regard to cervical spine manipulation. Reported adverse events have typically involved mild, temporary, self-limiting symptoms, although some reports have implicated severe permanent symptoms and even death.

The purpose of this systematic review was to retrieve and synthesize published prospective studies that involved manual therapy to determine the rate of adverse events and the relative risk of various manual therapies.

Pertinent Results:

After considering hundreds of possible articles, the reported data were derived from 8 prospective cohort studies (reported in 9 articles) and 31 randomized controlled trials (RCTs) (including 5 articles that presented data from the same studies).

The cohort studies represented data on approximately 36,949 manual therapy treatments, which included manipulation in 22,898 patients. The upper 95% confidence interval (CI) incidence risk rate of major adverse events was very low (0.007% [0/42, 451] after treatment or 0.01% [0/22, 833] per patient).

The estimated of incidence of minor or moderate adverse events was approximately 41% (95% CI - 17 to 68%), with the majority of these adverse events occurring within 24 hours of treatment.

Approximately 5060 participants were included in the RCTs, with 2,228 of them receiving manual therapy and 2,779 other therapies. However, no major adverse events were reported in any of the trials. The upper incidence rate of major adverse events was also very low in the RCTs (0.13% [0/2, 301]) after treatment.

The overall estimate of minor or moderate adverse events among the RCTs was 22% (95% CI - 11.1 to 36.2%).

Manual therapy interventions, mainly represented by manipulation, were associated with:
  • more adverse events than general practitioner care RR [relative risk] 1.91 (95% CI - 1.39 to 2.64);
  • about the same number as exercise RR 1.04 (95% CI - 0.83 to 1.31);
  • fewer than drug therapy RR 0.05 (95% CI - 0.0 to 0.20); and
  • a trend that was not statistically significant for more adverse events than sham, passive or control interventions RR 1.84 (95% CI - 0.93 to 3.62).

Clinical Application & Conclusions:

The majority of adverse events that occurred in the included studies happened within 24 hours of treatment. Moreover, Rubinstein et al. (2) reported that 72% of their reported adverse events occurred after the first treatment. Given this information, one would think that the best practice for manual therapists would be to advise their patients that they might feel a temporary increase of symptoms following their initial session of manipulation. The informed patient will be happy if they feel the same or better following treatment, and if they experience an exacerbation, they will likely not be as upset as they would have been without forewarning.

Some chiropractors that I know have had good success by routinely contacting their patients by telephone following their first manipulation (EDITOR’S NOTE from Dr. Thistle: I routinely explain this to patients, and then email [most patients prefer this] or call them soon after a first treatment – I ask which method they would prefer. Patients are always very appreciative for the contact, as it allows them to ask any further questions they may have. This practice takes only seconds, but builds trust and rapport that is extremely valuable).

One thing that should be pointed out regarding the conclusion of this study is that some prospective studies, including randomized clinical trials (the gold-standard of therapeutic effectiveness), are not very useful in establishing the rate of adverse events for a given intervention. In the testing of new drugs, for instance, Phase IV clinical trials are the best means of determining safety. These studies are carried out after the drug has been fully approved and released to the market.

Prospective studies often involve less than 100 subjects, and this is especially true in the manual medicine domain. While it is good if no adverse events occur during a study of this size, what happens when the intervention is administered to millions of people? Given a 1:1000 adverse event rate, there would only be a 10% chance that an adverse event would occur during a 100 subject clinical trial, but many would occur after the intervention starts being used more extensively.

Thus, even though the Carnes et al. review incorporated numerous prospective studies, there are still questions about the safety of manual methods. Practitioners can, however, use the provided information to help inform their patients about the relative incidence of adverse events to the manual therapy.

Study Methods:

Manual therapy was defined by the authors as: “any techniques administered manually, using touch, by a trained practitioner for therapeutic purposes”.

The following terms were used to classify the severity of adverse events:
  • Major: medium to long term; moderate or severe intensity
  • Moderate: medium to long term; moderate intensity
  • Minor: short term and mild intensity
Eight biomedical databases were searched, including the Index of Chiropractic literature, but not MANTIS (Manual, Alternative and Natural Therapy Index System).

The following search terms were used in the queries: chiropractic, osteopathy, orthopaedic, physiotherapy, manual therapist, manipulation, cavitation, mobilisation, articulation, adjustment. These terms were combined using the Boolean operator “AND” with the following terms: adverse event, effect, reaction, outcome, complication, response, side effects, spine, vertebra, muscle, disc, body, vascular, neurological.

Criteria for including articles:
  • RCTs and prospective cohort studies that contained original data about adverse events from manual therapy that was delivered by registered manual therapy professionals;
  • the intervention or therapy involved physical and/or manual contact with therapeutic intent, administered without the use of mechanical, automated, electronic, computer or pharmacological aides/products;
  • patients were conscious during the intervention.
Articles that were excluded included studies that involved:
  • mixed and multi-disciplinary interventions where the manual therapy effects would be unclear/undeterminable, and
  • self-administered interventions, including exercise programs.
The authors only reviewed and extracted data from RCTs that were published after the CONSORT statement was initially published in 1996 because, prior to its introduction, the reporting of adverse events in manual therapy efficacy trials was considered to be inadequate.

The quality of the included articles was judged using a modified CASP quality appraisal template for the cohort studies or a modified musculoskeletal appraisal template to assess the quality of the RCTs. However; only 10% of the articles were jointly reviewed by both reviewers to check the quality of the appraisal process.

In the statistical analysis, the authors estimated the incidence of minor, moderate and major adverse events and then determined the relative risk (RR) of reported adverse events that were related to manual therapy as compared to exercise, drug therapy, usual medical care, sham, passive and control interventions.

Study Strengths / Weaknesses:

This review was carefully conducted in many respects and the reader has good reason to be confident in its conclusions.

The authors admitted that it was difficult to properly classify the manual therapies involved in this review because they were often multipart interventions that made it impossible to rigorously ascribe causality.

Cohort studies and RCTs are not the best research methods for estimating the frequency of very rare events, such as vertebral artery dissection (VAD). It is also possible that adverse events with long latency periods, as has been suggested to occur in some cases of VAD, may not have had the chance to manifest during the trial period.

The main weakness of this study is the vague description of who assessed the quality of the articles that were included. In higher quality reviews, all articles are assessed by at least 2 reviewers, but the Carnes et al. review mentioned that only 10% of the articles were jointly reviewed.

The severity of some of the adverse events in this study may have been misclassified because of an overlap of definitions. Major adverse events were defined as: “medium to long term; moderate or severe intensity” and moderate adverse events defined as: “medium to long term; moderate intensity”. Thus, an adverse event that was “medium to long term; moderate intensity” could have been classified as either major or moderate.

The authors chose not to search the MANTIS database, which I consider to be one of the best on the topic of manual therapy.

Additional References:

  1. National Institute for Health and Clinical Excellence. Low back pain: Early management of persistent non-specific low back pain. NICE clinical guideline 88. May 2009.
  2. Rubinstein SM, Leboeuf-Yde C, Knol DL, de Koekkoek TE, Pfeifle CE, van Tulder ME. The benefits outweigh the risks for patients undergoing chiropractic care for neck pain: a prospective, multicenter, cohort study. J Manipulative Physiol Ther. 30(6):408-18.