J Manipulative Physiol Ther 2007 (Mar); 30 (3): 215–227 ~ FULL TEXT
Howard Vernon, DC, PhD, Kim Humphreys, DC, PhD, Carol Hagino, MBA
Canadian Memorial Chiropractic College,
Toronto, ON, Canada.
Neck pain is a common ailment, with approximately 10-15% of the population suffering from it at any given time. The researchers in this study reviewed 1980 citations and found 13 high quality trials utilizing manipulation or manual therapy. Their review found that the benefit from manipulation was greater. The long-term data regarding spinal manipulation for chronic neck pain was not as conclusive. However, the results demonstrated a great benefit with treatment up to 104 weeks. The spinal mobilization trials also showed very good benefit at a 6 to 7 week outcome point, with 70% of patients having full recovery or important improvement at that point. The researchers attempted to determine whether the benefit could be the natural history and placebo effect and compared these trials with a separate group of controlled no-treatment patients. The researchers concluded that the benefit noted in this review exceeded the placebo effect and natural history of the no treatment group.
OBJECTIVE: This study provides a systematic analysis of group change scores in randomized clinical trials of chronic neck pain not due to whiplash and not including headache or arm pain treated with manual therapy.
METHODS: A comprehensive literature search of clinical trials of chronic neck pain treated with manual therapies up to December 2005. Only clinical trials scoring above 11.5 (Amsterdam-Maastricht Scale) were included in the analysis.
RESULTS: From 1980 citations, 19 publications were selected. Of the 16 trials analyzed (3 were rejected for poor quality), 9 involved spinal manipulation (12 groups), 5 trials (5 groups) were for spinal mobilization or nonmanipulative manual therapy (1 trial overlapped), and 2 trials (2 groups) involved massage therapy. No trials included trigger point therapy or manual traction of the neck. For manipulation studies, the mean effect size (ES) at 6 weeks for 7 trials (10 groups) was 1.63 (95% confidence interval [CI], 1.13-2.13); 1.56 (95% CI, 0.73-2.39) at 12 weeks for 4 trials (5 groups); 1.22 (95% CI, 0.38-2.06) from 52 to 104 weeks for 2 trials (2 groups). For mobilization studies, 1 trial reported an ES of 2.5 at 6 weeks, 2 trials reported full recovery in 63.8% to 71.7% of subjects at 7 to 52 weeks, and 1 trial reported greater than 2/10 point pain score reduction in 78.3% of subjects at 4 weeks. For massage studies, 1 reported an ES of 0.03 at 6 weeks, whereas the other reported mean change scores of 7.89/100 and 14.4/100 at 1 and 12 weeks of, respectively.
CONCLUSIONS: There is moderate- to high-quality evidence that subjects with chronic neck pain not due to whiplash and without arm pain and headaches show clinically important improvements from a course of spinal manipulation or mobilization at 6, 12, and up to 104 weeks posttreatment. The current evidence does not support a similar level of benefit from massage.
From the Full-Text Article:
Currently, the Cochrane Review by Gross et al [36, 37] and the work of Bronfort et al  form the standard for evaluating the evidence for the treatment of neck pain by manipulation or mobilization. Our review differed from these works in several ways. With respect to the studies included, our review included not only studies of manipulation and mobilization but also of massage and other manual therapies as well. Our review included several studies that Gross et al and Bronfort et al had excluded because they were not studies comparing manipulation or mobilization to another form of therapy. Rather, these studies compared one form of these therapies with another form. In our review, each of these study groups was appropriate because they included selected, randomized subjects receiving one of the therapies of interest.
With respect to exclusions, we did not include studies involving subjects with acute neck pain, neck and arm pain, neck pain due to whiplash injury, or those with headache, whether clearly cervicogenic in nature or not. Thus, our review has remained within the boundaries of studies of chronic neck pain treated with one or more forms of manual therapy.
Our review did not include several studies that reported on subjects with neck pain that had been included in larger spine pain groups but did not clearly separate the results of the subjects with neck pain nor did they provide separate results for those with chronic neck pain. [65-69]
The primary difference between these reviews and our review lies in the analysis of change scores within groups so as to identify levels of improvement as opposed to determining whether differences between groups occurred as a measure of the “effectiveness” of the experimental (in this case, manual therapy) treatment. Interestingly, Bronfort et al  specifically endorse this line of inquiry (p 351); however, they do not pursue it in their review. In fact, they reported only the percentage differences between groups in their review of studies of manipulation and mobilization for spinal pain (including chronic neck pain). They do not even provide the outcome data for the study groups (as was done here) so that the reader might make these intragroup determinations (as a form of subgroup analysis within the larger review).
The recent Clinical Practice Guideline published by the Canadian Chiropractic Association  also explicitly distinguishes between the improvement obtained within groups and the effect of a treatment versus other comparative treatments (between-group effects) and focuses on the former in its evidence synthesis.
Gross et al [36, 37] did provide the mean values pre- and postintervention for all their study groups. However, they did not provide intragroup variability measures, and they did not analyze the degree of intragroup change at all. In other words, no summary of the change scores either as percentage difference, absolute difference, or effect size was provided. The sole thrust of their analysis, as sophisticated as it was, was the intergroup comparisons. In this, they provided intergroup differences as mean values and CIs that, when appropriate, were pooled to provide a summary measure of these differences. Additional analyses, such as “number needed to treat” were performed with the same intergroup theme in mind. Their conclusions were that, “The evidence did not favor manipulation and/or mobilization done alone or in combination with various other physical medicine agents; when compared to one another, neither was superior” (36, p 1).
As noted above, they did find supportive evidence (“for short-term and long-term maintained benefits”) for a multimodal approach of manipulation and/or mobilization combined with exercises for subacute/chronic mechanical neck disorders (as defined above).
With respect to our approach to subgroup analysis, it could be asked if it is appropriate to conduct intragroup analyses from a set of published RCTs. In none of the manipulation or mobilization trials included in this review was there a comparison between a form of manual therapy and a placebo control procedure. These trials are more properly seen as randomized comparative trials in which none of the subjects in these trials were blinded as to the form of treatment they received. Interestingly, both trials of massage are placebo-controlled clinical trials.
We maintain that once the intergroup outcomes are analyzed in standard systematic reviews, it then becomes appropriate to assess the magnitudes of change within each treatment group randomized to receive the therapy of interest and, if possible, summarize these results among studies. In fact, several studies in this review only report change scores. After hypothesis testing has been conducted, it is only sensible to assess these scores on their own for their clinical relevance. Our subgroup analysis only extends this exercise to the collective body of trials in this area.
Results from All Trials
From the baseline pain scores, it is evident that this body of trials involves patients with chronic neck pain, with mild to moderately severe neck pain. Most studies included outcome assessments up to 6 to 10 weeks. Several studies provided long-term outcomes up to 52 weeks, with one  providing outcomes to 104 weeks. There was considerable variance in the format of reporting the outcomes in these trials. Most studies reported pre- to posttreatment changes in primary outcomes. Some trials, reported only change scores, [60, 61, 63] whereas others only reported the percentage of subjects achieving a criterion level of outcome. [57, 60, 61]
The largest number of trial reports is available for manipulation (n = 9). All groups showed positive changes. Effect sizes could be calculated from 7 of 9 trials of a course of manipulation. Table 2 shows these effect sizes ranging from 0.56 to 3.2, most of which would be characterized as “large.” [43, 44] These effect sizes are maintained up to 12 weeks posttreatment. For long-term outcome, the data from 2 trials are less conclusive but still shows large effect sizes for up to 104 weeks.
The other 2 trials of a course of manipulative therapy [50, 56] reported change scores differently. In the first trial of Giles and Muller,  4-week mean reductions of scores on a 10-point VAS were reported as statistically significant for only the manipulation group (mean reduction, 1.5 [3.0] out of 10) as compared with the groups receiving non-steroidal anti-inflammatory drugs or acupuncture. Hurwitz et al  did not report change scores per se and only indicated that none of their contrasts between manipulation and mobilization achieved statistical significance at any outcome point. In all, 8 of 9 trials of a course of spinal manipulation reported statistically significant or clinically important changes in the group receiving manipulation. No trial group was reported to remain unchanged, and no trial group was reported to have worsened. In none of these trials were any major adverse reactions reported.
Five studies are available to determine the outcome of a course of mobilization therapy, one of which did not provide pre- and posttreatment pain scores.  All groups showed positive changes. Two studies [59, 60] provide data up to the 6- to 7-week outcome point. Only one of these  permits the calculation of an effect size, which was found to be large and at the upper end of the range found in the manipulation studies for the same period. Two studies provided data on the percentage of subjects achieving a clinically important improvement  or full recovery. [60, 61] From these, it appears that approximately 70% of patients achieve this level of improvement at the 6- to 7-week point. Only 1 study provided long-term data,  showing full recovery in approximately 70% of subjects at 13 and 52 weeks.
Only 2 trials of massage for chronic neck pain were retrieved. An effect size was calculated from Gam et al  for a group receiving massage and exercises of 0.03 at the 6-week outcome point. Irnich et al  reported the change scores in 100 mm VAS points at 1 week (7.89) and at 12 weeks (14.4), neither of which exceeds the 20 mm (2 of 10 points) level established by Brodin  and others [70, 71] as a clinically important difference in chronic pain patients.
There are several ways to assess the clinical relevance of change scores. They can be compared with what is known as the “minimum clinically important change.”  However, this value is properly derived from an analysis of patients' minimum expectations of change on a specific instrument as compared with a global or objective standard of change. To our knowledge, this has not specifically been done for pain scores for chronic neck pain patients.
More generally, Farrar et al  have reviewed the change scores on the 11-point pain scale in 10 clinical trials for a variety of chronic pain complaints (2724 subjects) and have determined that a 2-point or 20 of 100 mm change is clinically relevant for patients with chronic pain.
It could be argued that these change scores represent the natural history of chronic neck pain or the placebo effect within a trial and therefore do not reflect the influence of the treatments provided. We have investigated the average change scores in a separate group of controlled clinical trials of conservative treatments for chronic neck pain  and found that these are not generally greater than 15 mm on a 100 mm VAS (around 25% improvement). In several of these studies, there was no change at all in the control groups over up to 10 weeks posttreatment. Given these findings, the changes obtained in this review would appear to exceed what could be ascribed to either the natural history or the placebo affect.
Notwithstanding these comparisons with published benchmarks for clinical change, there is an urgent need for placebo- or sham-controlled clinical trials of manual therapies for chronic neck pain. Until such trials are performed, it will not be possible to accurately determine the attributable effect of these therapies over and above the nonspecific effects that are generally present in all clinical trials but even more strongly present during manual therapies in particular.
There is moderate- to high-quality evidence that subjects with chronic neck pain not due to whiplash and without arm pain and headaches who are randomized to receive a course of spinal manipulation or mobilization show clinically important improvements at 6, 12, and up to 104 weeks posttreatment. The current evidence does not support a similar level of benefit from massage therapy. There is a need for controlled studies of these therapies for chronic neck pain.