Study Design and Setting
We conducted a pilot randomized controlled trial for adults with chronic low back pain recruited from 2 community health centers (CHCs) that serve a racially diverse, low-income neighborhood of Boston, Massachusetts. Participants were randomly assigned to a standardized 12-week protocol of hatha yoga classes or a usual care waitlist control group. We used computer-generated permuted block randomization. Treatment assignments were placed in opaque, sequentially numbered envelopes prepared by a biostatistician (RBD) who had no contact with participants. The study was approved by the Boston University Medical Center Institutional Review Board and the CHCs’ research committees.
To recruit our target of 30 participants for this pilot study, we posted flyers in exam rooms and waiting room areas in the CHCs and surrounding community. A community newspaper and radio station carried study advertisements. We informed providers and staff members about the trial through presentations and e-mails. Using the CHCs’ electronic medical records, we identified patients seen in the last 2 years with a low back pain diagnosis and generated recruitment letters for their providers to sign and send.
Participants needed to be 18 to 64 years old and have current low back pain persisting ≥12 weeks. Mean pain intensity for the 2 weeks prior to enrollment needed to be ≥4 on a numerical rating scale of 0 to 10. Sufficient understanding of English to follow class instructions and complete surveys was required. Exclusion criteria included yoga use in the previous year; new pain medicine or other low back pain treatments started within the previous month or anticipated to begin in the next 6 months; pregnancy; back surgery in the previous 3 years; nonmuscular pathologies (eg, spinal canal stenosis, spondylolisthesis, infection, malignancy, fracture); severe or progressive neurological deficits; sciatica pain equal to or greater than back pain; active substance or alcohol abuse; serious systemic disease, medical, or psychiatric comorbidities precluding yoga practice; active or planned worker’s compensation, disability, or personal injury claims; and inability to attend classes at the times and location offered.
Interested individuals were initially screened for eligibility by telephone after oral informed consent was obtained. If individuals appeared eligible, they were invited to meet with research staff. At the first meeting, we obtained written informed consent to verify eligibility criteria and asked them to record their daily pain intensity for 2 weeks on an 11-point numerical rating scale. Two weeks later at the second meeting, we confirmed their average weekly pain intensity was ≥4, collected baseline data, and randomized them.
We developed a reproducible standardized hatha yoga intervention for CLBP intended for individuals with little or no yoga experience. We searched Medline, Alt HealthWatch, and Cochrane for papers on yoga and low back pain. Non-peer reviewed books, periodicals, and videos on yoga for low back pain were identified through searching an online bookstore (Amazon.com), an annotated bibliography,22
and websites on the first 2 pages of a Google Internet search using keywords “yoga” and “back pain.” We collected and distributed these materials to an expert panel consisting of 2 national yoga experts, a yoga instructor for the study (DCD), and the principal investigator (RBS). Panel members had expertise in several popular styles of hatha yoga including Anusara, Ashtanga, Iyengar, and Kripalu. One member had special expertise in yoga programs for minority women. Panel members reviewed the information before meeting in March 2006. At the meeting, they synthesized information from the literature with their experience to draft a protocol that was subsequently refined iteratively through discussion, consensus, and use in nonstudy yoga classes.
The final protocol consisted of 12 weekly 75-minute yoga classes divided into four 3-week segments ( and ). Each segment was given a theme, such as “Listening to the Wisdom of the Body” and “Engaging Your Power.” Each class began and ended with Svasana, a relaxation exercise. Classes included postures and breathing techniques. Each segment built upon the previous segment. The protocol provided variations and used various aids (eg, chair, strap, block) to accommodate different abilities. A variety of world music was used during the classes. Classes were limited to 8 participants, occurred at one CHC, and were taught by a team of 2 female yoga instructors, 1 white and 1 African American. Both were registered yoga teachers with Yoga Alliance, and each had approximately 4 years of teaching experience. Home practice for 30 minutes daily was strongly encouraged. We provided participants with an audio CD of the protocol; a portable CD player; a handbook describing and depicting the exercises; and a yoga mat, strap, and block. The 2 national yoga experts from the panel observed several classes in person to provide feedback to the instructors on accurate, effective, and safe protocol delivery.
12-week Standardized Hatha Yoga Protocol for the Treatment of Chronic Low Back Pain
APPENDIX FIGURE The yoga postures (asanas) shown were part of a standardized hatha yoga protocol developed for chronic low back pain in individuals with little or no yoga experience. To design the protocol, we performed a systematic search of the peer-reviewed and lay (more ...)
Usual Care Control Group
Both groups continued to receive their routine medical care and medications. The usual care control participants were offered the yoga intervention after 26 weeks. Both groups received an educational book used in previous low back pain studies23
that describes self-care management strategies for low back pain. Both groups were discouraged from starting any new back pain treatments during the study.
We collected baseline sociodemographic data and back pain history including duration, sciatica, and previous treatments such as physical therapy, epidural steroid injections, and CAM. Outcome data were collected at 6, 12, and 26 weeks.
Feasibility outcomes related to recruitment (time to complete enrollment, proportion of racial and ethnic minorities enrolled), participant retention, adherence to treatment allocation (class attendance, home practice, use of nonstudy treatments), and safety were measured through weekly adverse event logs.
There were 2 primary outcomes of efficacy measured at 12 weeks: (1) average pain level for the previous week using an 11-point numerical rating scale (0=no pain to 10=worst possible pain);24,25
and (2) back-related function using the modified Roland-Morris Disability Questionnaire,26
a 23-item reliable validated instrument measuring the number of activities of daily living limited due to back pain. Scores can range from 0 to 23 with higher scores reflecting poorer back-related function.
Secondary efficacy outcomes included use of pain medication during the preceding week; global improvement using a 7-point Likert scale (0=extremely worsened to 6=extremely improved); and health-related quality of life using the SF-36.27
Participants and yoga teachers could not be blinded to treatment allocation. All study participants met in person with unblinded research staff members to complete paper questionnaires at baseline, 6, and 12 weeks. We also attempted to collect follow-up data at 26 weeks. After completing each survey, participants received honoraria for study participation and defraying transportation costs: $25 at baseline, $25 at 6 weeks, $50 at 12 weeks, and $25 at 26 weeks. Blinded data-entry staff used double entry verification to minimize error.
Baseline data for the 2 groups were compared using Student’s t
-test for continuous variables, Fisher’s exact test for dichotomous variables, and chi-square test of independence for categorical variables. Feasibility outcomes were summarized with descriptive statistics. For the primary efficacy outcomes pain and function, we calculated change scores by subtracting 12-week data from baseline. Change scores for each group had a non-normal distribution and were therefore compared using the Wilcoxon Rank Sum test. We also completed a post-hoc analysis of the proportion of individuals experiencing a minimal clinically significant decrease in the primary outcomes at 12 weeks (≥2 points for pain28
and ≥30% decrease from baseline for the Roland29
) using Fisher’s exact test.
For the secondary efficacy outcomes, we compared change in pain medication use between groups at 6 and 12 weeks using exact logistic regression. Medication use was the dependent variable, and baseline medication use and group assignment were the independent variables. Global improvement at 12 weeks was dichotomized into improved vs no change or worse. Proportions of improved participants in the 2 groups were compared with Fisher’s exact test. Change scores for the SF-36 physical and mental health components were also compared with Wilcoxon Rank Sum.
All analyses used an intention-to-treat approach and a 2-sided P criteria of <.05 for statistical significance. Missing data were imputed using the last-value-carried-forward approach. We used LogXact software (Cytel, Cambridge, Massachusetts) for logistic regression analyses and SAS version 9.1 (SAS Institute, Cary, North Carolina) for all others.