Chronic neck pain is a significant health problem affecting from 30% to 50% of the population annually. While there are relatively few evidence-based treatments for this condition, research studies have suggested that yoga may be a clinically effective option. A recently reported study helps to confirm that Iyengar yoga can be more efficacious than a home exercise program for relieving neck pain and improving quality of life; however, evidence for this is relatively sparse and the quality is weak.
As reported in the Clinical Journal of Pain, Holger Cramer PhD — at the University of Duisburg-Essen, Germany — and colleagues randomly assigned 51 patients with chronic neck pain (mean age ≈48 yrs; 82% female) to either weekly 90-minute Iyengar yoga classes during a 9-week period (n=25) or to home-based exercises for neck-pain relief (n=26) using a self-care manual [Cramer et al. 2013A]. The main outcome measure was current neck pain intensity on a 100mm visual analog scale (VAS). Secondary outcome measures included functional disability (Neck Disability Index), pain at motion (VAS), health-related quality of life (Short Form-36 questionnaire), cervical range of motion, proprioceptive acuity, and pressure-pain threshold.
After the study period, patients in the yoga group reported significantly less neck pain intensity compared with the home-exercise group [VAS mean difference = 13.9 mm (95% Confidence Interval or CI, 1.4 to 26.4), P=0.03]. The yoga group also reported less disability and better mental quality of life. Range of motion and proprioceptive acuity were improved and the pressure-pain threshold was elevated in the yoga group.
There is a movement in the pain field today to explore options besides systemic analgesic or other drug therapy for persisting neck pain conditions, especially when there is no explanatory musculoskeletal or neurological pathology as a cause. Multimodal, biopsychosocial approaches are of great interest, including clinical massage therapy, physical therapy, chiropractic, acupuncture, mind-body therapies, yoga, and others.
However, much of the research has been of low quality and has produced relatively weak evidence favoring and supporting the respective modalities. Additionally, most primary care physicians cannot provide those services themselves, and insurance coverage for referral to outside providers of complementary and alternative therapies can be spotty at best.
Neck pain has been variously discussed in other UPDATES articles here, and yoga was specifically considered in UPDATES here. Yoga is not entirely without a potential for harmful effects [eg, see UPDATES here and here].
The 9-week study by Cramer et al. [2013A] found an effect size favoring yoga of 0.40 (Cohen’s d, calculated from study data), which is only of moderate proportions. Furthermore, the decrease in VAS pain-intensity score of about 14mm in the yoga group had a wide CI (merely 1.4mm improvement at the low end); although, this was statistically significant, it is typically not considered to be clinically important for patients. According to generally accepted criteria, pre- versus posttreatment decreases of at least 20mm (or 30%–36%, on a 100mm VAS) usually denote subjects feeling “much better” or “meaningfully improved”; whereas, a decrease of ≥40mm or about ≥50% represents their feeling substantially (“very much”) improved.
The small group sizes, roughly 25 subjects per treatment arm, provide insufficient statistical power to portray valid, nonrandom outcomes. Additionally, there was no control group — eg, wait list, or “usual care” — to assess for comparison purposes the natural course of the neck-pain assessed in this study; so, it is possible that neither yoga nor exercise may have been truly efficacious.
Another paper by Cramer et al. on this same study, and appearing in the journal Pain Medicine [2013B], assessed outcomes at 12 months after study completion. The researchers reported that from baseline to the 12-month followup, at which time there were a total of only 36 patients assessed, there were statistically significant moderate-to-large effects (calculated from study data) favoring yoga: pain-intensity improvement (Cohen’s d=0.86; p< 0.001); decrease in neck-related disability (d=0.56; p = 0.001); and, improvement in bodily pain on the SF-36 (d=0.65; p = 0.005).
Improvements in pain intensity were predicted by weekly minutes of yoga practice at home during the 4 weeks prior to final assessment at 12 months, and improved neck-related disability and overall bodily pain were predicted by regular yoga practice during the past 12 months overall. Therefore, sustained yoga practice seems to be the most important predictor of long-term effectiveness. However, the small numbers of subjects available for followup and the lack of a control group preclude definitive and valid statements of longer-term yoga efficacy; as with the shorter-term results, improvements at one year could be related to the natural course of neck pain conditions involved in the study.
It is interesting that, in two separate journals, Cramer and colleagues published their 12-month results just one month after the original study article reporting 9-week outcomes. While this is not exactly a frowned-upon duplicate publication, the split presentation of short- and long-term data certainly fragments the total picture of what occurred among the patients. Perhaps, this was due to journal scheduling of article publication; but, it also demonstrates the frustrations faced by readers who want complete data on an investigation such as this.
Adding further context to the investigations by Cramer et al., approximately one year ago there was a very similar report of Iyengar yoga for neck pain published in the Journal of Pain by Michalsen and colleagues from Berlin, Germany [Michalsen et al. 2012, also discussed in UPDATE here]. For this study, 76 patients (mean age 48 years, 87% female) with chronic neck pain who scored initially >40mm on a 100mm VAS were randomized to a 9-week Iyengar yoga program with weekly 90-minute classes (n = 38) or to a self-care/exercise program (n = 38). There was no control group for comparison to assess either natural progression or remission of neck pain.
Twelve patients in the yoga group and 11 patients in the self-care/exercise group were lost to follow-up, leaving only 26 and 27 in each respective group. Adjusting for small sample sizes, we calculated the effect sizes (Cohen’s d) favoring yoga for pain at rest as 1.18, and 1.15 for pain during motion — both of which are significantly large and clinically important outcomes. At the same time, absolute mean reductions in VAS-pain scores favoring yoga were 20mm for pain at rest, and 19mm for pain during motion — both of which only attain minimal requirements for clinically “meaningful improvement.” Statistically significant positive effects of yoga were also found for pain-related apprehension, disability, QOL, and psychological outcomes.
From these studies of Iyengar yoga for chronic neck pain we might conclude that this is a modality with some promise, but better research evidence is needed for more definitive conclusions. And, from EBPM (Evidence-Based Pain Management) perspectives there are several “take away” points, as emphasized in other UPDATES articles [series here], illustrated by the research investigations:
- Studies of small groups — some say anything <50 subjects per arm is too small — usually do not provide unbiased, reliable, and/or valid results for clinical decision-making purposes.
- Multiple similar, but small research studies finding comparable outcomes favoring a therapy or intervention do not necessarily add up to a strong and convincing body of evidence.
- Even when outcomes — eg, group differences, effect sizes, etc. — are statistically significant in terms of P-value they may not be of clinical importance. Also, absolute improvements in scores (whether a VAS, numerical rating scale, or other measure) must meet established minimal levels of clinical significance for patients, no matter how statistically significant they may be.
- In pain research, data portraying group averages are often unhelpful or misleading, since they do not indicate the proportions of patients that greatly improved due to a therapy versus those that fared poorly. The distribution of patient responses is typically U-shaped (see Figure), with a relatively small proportion reacting to therapy in an “average” way, and even fewer at the median, or mid-point level, if that metric is used.
- From clinical perspectives, practitioners and patients could benefit from knowing NNT parameters; that is, how many patients must be treated for one additional patient to be helped or harmed in specific ways. However, as in the studies above by the Cramer and Michalsen teams, data often are not organized or reported in ways that permit calculating these metrics.
> Cramer H, Lauche R, Hohmann C, et al. Randomized-controlled trial comparing yoga and home-based exercise for chronic neck pain. Clin J Pain. 2013A;29(3):216-223 [abstract here].
> Cramer H, Lauche R, Hohmann C, et al. Yoga for Chronic Neck Pain: A 12-Month Follow-Up. Pain Medicine. 2013B;14(4):541-548 [abstract].
> Michalsen A, Traitteur H, Lüdtke R, et al. Yoga for chronic neck pain: a pilot randomized controlled clinical trial. J Pain. 2012 Nov;13(11):1122-1130 [abstract].