|Home | About | Journals | Submit | Contact Us | Français|
We report a systematic analysis of group change scores of subjects with chronic neck pain not due to whiplash and without headache or arm pain, in randomized clinical trials of a single session of manual therapy. A comprehensive literature search of clinical trials of chronic neck pain treated with manual therapies up to December 2006 was conducted. Trials that scored above 60% on the PEDro Scale were included. Change scores were analyzed for absolute, percentage change and effect size (ES) whenever possible. Nine trials were identified: 6 for spinal manipulation, 4 for spinal mobilization or non-manipulative manual therapy (2 overlapping trials), and 1 trial using ischemic compression. No trials were identified for massage therapy or manual traction. Four manipulation trials (five groups) reported mean immediate changes in 100-mm VAS of −18.94 (9.28) mm. ES for these changes ranged from .33 to 2.3. Two mobilization trials reported immediate VAS changes of −11.5 and −4 mm (ES of .36 and .22, respectively); one trial reported no difference in immediate pain scores versus sham mobilization. The ischemic compression study showed statistically significant immediate decreases in 100-mm pain VAS (average = −14.6 mm). There is moderate-to-high quality evidence that immediate clinically important improvements are obtained from a single session of spinal manipulation. The evidence for mobilization is less substantial, with fewer studies reporting smaller immediate changes. There is insufficient evidence for ischemic compression to draw conclusions. There is no evidence for a single session of massage or manual traction for chronic neck pain.
Neck pain is a very common problem, second only to low back pain in its frequency in the general population1,2,3,4 and in musculoskeletal practice5. Approximately 15% of females and 10% of males suffer with chronic neck pain at any one time6,7,8. Chronic neck pain produces a high level of morbidity by affecting occupational and avocational activities of daily living and by affecting quality of life9,10,11,12.
Manual therapies are commonly used in the treatment of chronic neck pain, and there are numerous systematic reviews of the treatment of neck pain by manual therapy13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39. In a recent report40, we reviewed all studies of manual therapies for chronic neck pain, focusing on the change scores obtained after courses of treatments of various manual therapies. However, the issue of change scores in studies of single sessions of these treatments has received little attention.
The importance of single-session trials of a treatment can be viewed in two ways. First, the trials provide a form of proof of principle in that the studies indicate whether a single dose of the treatment achieves an appropriate level of the intended outcome. Indeed, dose-dependence can be directly evaluated in these types of studies. Conversely, such studies can provide data on the utility of potential outcome measures for subsequent studies. Second, as most treatments are provided in a series or regimen, these studies provide clinicians and researchers with an indication of the expected outcome on each session of the treatment, at least at the outset of the treatment program. This indication must be tempered with the notion that the cumulative effects of several sessions may not accumulate in a linear fashion, and that the within-session changes may vary during the course of the treatment.
Recent interest in single session changes has grown among researchers. Tseng et al41 developed a preliminary prediction rule for cervical manipulation in a group of mostly chronic neck pain sufferers. Using the criterion of either 50% pain relief, 4 out of 7 on a global rating of change, or the rating of “very satisfied” with outcome to determine “responders,” 60% of their sample achieved this level after a single session. A prediction rule41 consisting of seven baseline criteria was developed. Subjects with 3 out of 7 criteria had a 74% chance of being a responder; subjects with 5–6 criteria were 100% likely to respond. Tuttle et al42,43 investigated the degree to which within-session changes in the first one42 or two43 treatments predict responses in subsequent treatments. They found a positive correlation at one week for only range of motion changes vs. changes on pain scores.
To date, there has been no summary of the magnitudes of change in pain scores reported in the clinical trials of manual therapies for chronic neck pain. In the present report, we focus our review on change in pain scores in those clinical trials of a single session of manual therapies for adults with chronic neck pain (without headache, whiplash, or arm pain).
A comprehensive literature search was performed in MEDLINE and CINHAHL using the strategy illustrated in Table Table1.1. Targeted text word searches were conducted in AMED, MANTIS, and Index to Chiropractic Literature (ICL) using the terms neck pain, randomized clinical trial, manual therapy, manipulation, and mobilization. Several recent systematic reviews18,36,37,38,39 were searched directly. Hand searches were also conducted of reference lists. Searches were conducted up to late 2006.
The following inclusion criteria were then applied: RCT: The study design had to be a randomized clinical trial in which at least one treatment group of adults ages 18–50 was provided with a course of one of the manual therapies (as defined below) for chronic mechanical neck pain. Chronicity: Chronic neck pain has been variously defined as to its duration. Some authors require at least three months of continuous symptoms while, for others, chronicity can develop after only one month of symptoms44. We defined chronic neck pain as being of a minimum of eight weeks duration. Neck pain: This review included only studies with subjects with neck pain without arm pain or headache and not due to whiplash injury. Treatment: For this review, we employed the separate therapy categories of manipulation, mobilization, manual traction, massage, and pressure techniques. Outcome: The primary outcome was pain (function-related outcomes such as measures of self-reported disability or measures of ranges of motion were also included if a pain-related outcome was present).
Once selected, the method of each study was scored using the PEDro Scale45. This scale has good reliability46 and was selected for its appropriateness to the task of evaluating studies of a single intervention46. The PEDro Scale provides a score out of 10; we identified 6/10 as the cut-off for acceptable quality. Two assessors (HV and BKH) scored studies separately, and disagreements were resolved by consensus. Evidence tables were compiled from extracted data by the primary author and a research assistant. Data were obtained only from the published works and not from follow-up with authors.
When continuous data were reported, as means and standard deviation (SD) for baseline and outcome intervals, absolute and relative changes were calculated. Intra-group effect sizes were calculated according to the method of Cohen47,48. Where median scores were reported, the confidence intervals were used to calculate proxy standard deviations and the median was treated as the mean. Where only change scores were reported (as opposed to both pre- and post-intervention scores), the effect size was not calculated. Where possible, change scores were averaged (mean [95% Confidence Interval]). Given that this is a secondary analysis, no further analysis such as weighting of effects was undertaken.
Eight trials were identified by this process. The AMED, MANTIS, and ICL searches generated no additional studies. Hand searches identified one additional trial49. Nine trials were ultimately selected for inclusion49,50,51,52,53,54,55,56,57, six of which involved spinal manipulation, four of which involved spinal mobilization or non-manipulative manual therapy (2 trials overlapped with 2 manipulation trials), and one that involved manual trigger point therapy. No trials included massage therapy or manual traction of the neck.
The PEDro scores for the nine included studies are shown in Table Table3.3. All nine studies scored at least 6 out of 10. The most frequently deficient items were concealment of allocation, blinding of all subjects, and blinding of all therapists.
A total of 223 subjects with chronic neck pain only were included in the manipulation studies (Table (Table4).4). Baseline pain levels were obtained from 3 studies50,51,52, ranging from 37 to 57 mm/ 100. Four manipulation studies (five groups)50,51,52,53 were judged to be sufficiently similar and homogeneous to permit summarizing of their change scores. The mean immediate change in these four trials in a 100-mm VAS was −18.94 [7.42,30.16] mm. A medium effect size (ES) for immediate change in 100-mm VAS was obtained in two studies at .3351 and .5250 while one study reported a large ES at 2.352. One study also reported mean change scores at 5 hours post-manipulation of −10.4 mm53. One study reported change scores only at 3 weeks post-manipulation, obtaining a mean of 18 mm improvement54. The other study reported on immediate changes in paraspinal tenderness scores, obtaining an average increase of 45% post-manipulation55. Mild, temporary pain-related adverse effects were reported in 6–17% of subjects in three studies50,53,54. No major adverse reactions (defined as any reaction requiring additional medical intervention at any time) were reported in any of these studies.
A total of 115 subjects were included in 4 studies employing a spinal mobilization group. Baseline pain levels were obtained from two trials: 31 and 55/100 mm50,52. Two trials reported immediate changes in a 100-mm VAS of −11.5 and −4 mm, respectively50,52. These changes amounted to an ES of .3650 and .2252, respectively. One trial reported no difference in pain change scores of the mobilization group as compared to the group receiving sham mobilization56. Two trials reported on immediate changes in local pressure pain thresholds, obtaining 19%57 and 22.5%56 increases, respectively. One study reported mild, temporary pain-related adverse effects in 6% of subjects50. No major adverse reactions were reported in any of these studies.
For the single trial of trigger point therapy57 (Table (Table5),5), 6 ischemic compression protocols were investigated, with several showing statistically significant increases in local pressure pain threshold (average = 26%) and all showing a statistically significant immediate decrease in 100-mm VAS (average = −14.6 mm).
To our knowledge, we are the first to report on the within-group changes of pain scores from clinical trials of a single session of a manual therapy for chronic neck pain in adults. In their recent review, Bronfort et al18 did not address this issue directly and included only the study by Sloop et al54. As that study acquired its outcomes three weeks after the intervention, they likely regarded it as a longitudinal study, even though only a single intervention was provided. Bronfort et al18 provided no summary of evidence nor any clinical recommendations regarding single-session studies.
In recent Cochrane reviews, Gross et al36,37,38 included 4 single-session studies, three of which50,54,55 were included in this review. Those authors also included the trial by Howe et al58; however, this study did not include a group receiving only manipulation, thus we excluded it from our review. Gross et al36,37,38 conducted a pooled analysis of Vernon et al55 and Sloop et al54, given that they were clinically comparable and not statistically heterogeneous. Outcome measures were pooled despite the fact that Vernon et al55 obtained only pressure algometry readings, not pain scores, at 5 minutes post-intervention and included a mobilization group, while Sloop et al54 obtained pain outcome only at 3 weeks post-intervention. Despite these considerable problems, Gross et al36,37,38 concluded from their analysis that “when compared to a control (other treatments deemed to be ineffective), there was moderate evidence that single sessions [of manipulation or mobilization] did not result in short-term pain relief [pooled SMD −0.51 (95% CI: −1.10 to 0.07)]” for acute, sub-acute, or chronic neck pain.
Our review differs from these two previous reviews in a number of ways. First, our primary objective was an analysis of the within-group, as opposed to between-group change scores in groups having received a single session of the treatments of interest. Second, our interest was exclusively in studies of a single session. Third, the scope of our review was larger, involving massage, pressure therapies, and manual traction, leading to the inclusion of additional studies such as Yurkiw et al51 for manipulation and Sterling et al56 and Hanten et al49 for mobilization (the more recent studies of Haas et al53 and Martinez-Segura et al52 were not available at the time for either the reviews of Bronfort et al18 or Gross et al36,37,38). Fourth, more sophisticated analyses such as pooling of effects were not conducted. Lastly, our review remained within the boundaries of studies of chronic neck pain treated with one or more forms of manual therapy.
The subjects included in these studies were relatively homogeneous, not only because of the application of our selection criteria but also because of the similarity of baseline pain severity. In addition, these studies appeared to be similar in regard to the types of treatments employed: manipulations were delivered as single-thrust procedures while mobilizations appeared to be relatively similar in application. Finally, most studies employed the same outcome measure of pain severity on a 100 mm scale (see below re: Farrar et al59), although a few reported on pressure pain thresholds measured by algometry. Given these clinical and statistical similarities, and given the sufficiently high quality of the studies reviewed, we felt that it was tenable to summarize the results of the 4 manipulation studies that reported pre- and post-intervention VAS scores and then derive a mean change score. This was not possible for the mobilization studies.
The question then arises as to how to interpret our findings. In a previous review60, we referred to the work of Farrar et al59 in establishing that the minimum clinically important difference in pain severity scores on the VAS as registered by a large variety of patients with chronic pain is 20 mm. This applies to participants in studies of courses of various treatments for their pain, courses extending over days and weeks. In the case of the present review, outcomes were obtained immediately following the interventions. We could not identify any study that provided a similar reference frame for studies of a single-session of treatment for chronic pain patients.
We did identify one study concerning single-session changes in acute pain patients. Bird and Dickson61 studied emergency-room pain patients and reported that with initial pain VAS scores between 34–66, the minimal clinically important in-session difference in pain score was 17 [95% CI:13, 21] mm; for achieving the outcome of a lot less pain, the mean score was 33 [24, 42] mm. The baseline pain scores of the 4 pre- and post-manipulation studies in our review50,51,52,53 lie in the same range as those of Bird and Dickson. Their level of minimal clinically important difference (17 mm) in a single emergency-room treatment session is similar to the average level of change (–18.94 [7.42,30.16] mm) that we found in these 4 studies. With Farrar et al59, Vernon et al60, and Bird and Dickson61 as references, we advocate that the mean change in pain score of −18.9 [7.42,30.16] mm immediately after a single manipulation in 4 similar studies50,51,52,53 represents a clinically important finding.
The findings with respect to manual pressure techniques and mobilization permit neither summarization nor simple interpretation. We believe that the results of the manual pressure techniques and mobilization studies are best left in the studies' descriptive mode for use by future clinical trial planners. The results provide some indication of the magnitudes of clinical change that can be expected in a single session of the various types of mobilizations included in these studies.
With respect to our approach to within-group analysis, it could be asked if it is appropriate to conduct such analyses from a set of published RCTs. Only one of the manipulation trials54 included a comparison between a form of manual therapy and a placebo control procedure (anamnestic valium only). Only one of the mobilization trials included a sham procedure56. The remaining majority of 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. Two studies compared manipulation to mobilization50,52 whereas two studies compared two different modes of manipulation51,53.
We maintain that once the inter-group outcomes of manual therapy trials have been analyzed in standard systematic reviews18,36,37,38, it then becomes appropriate to assess the magnitudes of change within each treatment group randomized to receive the therapy of interest and, if possible, to summarize the results among studies. In fact, several studies in this review only reported change scores53,54,55. After hypothesis testing has been conducted, it is reasonable to assess these scores individually for clinical relevance. Our sub-group analysis only extends this exercise to the collective body of trials in this area of interest.
Our study has limitations. The major limitation was the lack of high-quality RCTs that included clearly identifiable patient groups undergoing single-session manual treatments. Consequently, our review included only 4 trials of spinal manipulative therapy, only 4 trials of spinal mobilization and only 1 trial of trigger point therapy. No trials were identified for massage therapy or traction. In terms of quality, only one manipulation trial and one mobilization trial included a placebo control or sham group.
Our quality assessment suggests that weaknesses were associated with concealment of allocation and in blinding of subjects and therapists. In addition, most studies failed to report the presence or absence of adverse reactions. Only Sloop et al54 and Haas et al53 incorporated a double-blind design, while in none of the other studies were either therapists or subjects blinded to the interventions. This is a common problem in clinical trials of manual therapies40. There is an urgent need for a reliable sham procedure for future trials of these manual therapies, whether with single or multiple intervention sessions.
In addition, not all trials in this review used pain severity as measured on a 100-mm scale as the main outcome. Some used pressure algometry or a mix of outcomes. There were also variations in post-measurement for the main outcome measure. Some outcomes were measured immediately, others at 5 hours or 3 weeks after the single manual therapy intervention. Calculations such as effect sizes and comparisons of manual therapies were made more complicated as a result of differences in the outcomes measured, data reported (continuous versus ordinal; means or median scores; varied measures of variability; change scores versus pre- and post-intervention scores) and the timing of measurements (post-treatment).
Another problem was the differences in baseline pain severity scores for different trials, which made it difficult to pool patient treatment groups for the various manual therapies. In terms of inclusion criteria, chronic neck pain varied from one to three months depending on the trial.
Finally, as discussed previously, the minimal clinically important difference (MCID) in pain severity immediately following a single session of manual therapy has not been defined for manual therapies. For this review, work by Bird and Dickson61 in emergency-room pain patients was used. More studies are needed to identify the MCID for manual therapies after a single session.
Investigators have studied the effects of a single session of a variety of manual therapies for chronic neck pain by reporting on changes in subjective pain, local tenderness, and regional ranges of motion. The largest number of studies has involved spinal manipulation. 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 single session of spinal manipulation demonstrate immediate improvements which, in comparison to published benchmarks, could be considered clinically important. A minority of subjects have reported brief, mild pain-related side effects following this intervention. The evidence for mobilization is less substantial, with fewer studies reporting smaller immediate changes. The evidence for manual trigger point therapy is insufficient to draw conclusions. There is no evidence concerning the effects of a single session of massage or manual traction for chronic neck pain. Further high-quality studies are recommended to better determine the magnitudes of change in clinically important measures after a single session of any of the variety of manual therapies.
Howard Vernon, Professor, Division of Research, Canadian Memorial Chiropractic College, Toronto, ON, Canada.
Barry Kim Humphreys, Professor, Division of Research, Canadian Memorial Chiropractic College, Toronto, ON, Canada.