Material and methods

Selection criteria

To determine their initial eligibility, all potential studies were retrieved in full text and reviewed for the inclusion criteria. All three authors independently screened all identified full-text articles, to determine whether the study met the inclusion criteria. To be included in this review, studies had to meet the following criteria: (1) enrollment of participants 21 years of age and older who met the standardized criteria for FM diagnosis, (2) study intervention met the definitions for CAM and land-based exercise, and (3) inclusion and reporting of pre- and post-Fibromyalgia Impact Questionnaire (FIQ) total score or numeric pain rating scale scores.51,52 This means that CAM studies with very low levels of exertion, such as studies of body awareness therapies, breathing studies, and mindfulness-based stress reduction (which generally offers very brief yoga in two out of eight sessions), were not included. Additionally, we excluded water-based CAM therapies, such as tai chi in water and yoga/breathing in water, as it is unclear whether the benefit derived from balneotherapeutic effects or physical movement and mindful techniques (Figure 1).53,54

Figure 1

Search strategy results.

Results

Meta-analyses

The standard difference in means (0.84), and standard errors (0.07) for CAM exercise, indicates a statistically significant (P = 0.00) and positive outcome. The individual forest plots for qigong, tai chi, yoga, and all others are reported in Figures 2–5 respectively. With the exception of two studies, all studies reported positive outcomes.71 Of those not reporting positive outcomes, Mannerkorpi and Arndorw71 found no difference between 14 weeks of qigong versus a sedentary “usual care” control in 36 FM patients. Likewise, Carbonell-Baeza et al61 concluded that a 4-month male tai chi study (n = 6) was negative for symptom improvement.

Figure 2

Qigong forest plot.

 

Figure 5

Other CAM forest plot.

Sensitivity analyses

A funnel plot of the Std diff in means indicates that if the standard difference in means (0.84) is true, there is little evidence for reporting bias. The majority of studies fell within two standard deviations, with an equal number of studies falling outside this on either side of the plot (Figure 6). A regression of the Jadad scores on the standard difference of means indicates more variation reported in the lower scores and less in the higher scores, with a slight increase in the slope seen with higher scores (Figure 7).

Figure 6

Funnel plot of standard error by standard difference in means.

 

 

Figure 7

Regression of Jadad score on standard difference in means (Jadad et al, 1996).57

Discussion

The data presented support the likely safety of CAM exercise interventions, as 832 participants were enrolled in 16 studies and 81% completed the studies without differential attrition by study arm or exercise modality.58–75 Only one study reported negative side effects (increased shoulder pain and plantar fasciitis) in two study participants.70 All studies reported that there were no serious adverse events.

The data also support some degree of efficacy of several of the CAM exercise interventions in improving overall FM symptoms and physical function. Taken as a whole, the data do not indicate that any one intervention type is consistently superior to another, though the strongest effects were for the exercise modalities with the largest number of trials: qigong (Std diff in means 0.47; P < 0.000) and tai chi (Std diff in means 1.14; P < 0.000). Further support for qigong comes from a trial excluded from this analysis. Stephens et al76 randomized 30 children (ages 8–18) with FM to aerobics or qigong. There was no difference between study arms, although qigong was intended to be the control group. The qigong group did demonstrate significant within-group changes, indicating that pediatric CAM exercise interventions may be a promising area of research. This is critical, as few interventions address pediatric FM even though its recognition as a valid diagnosis is increasing. This may indicate that mind–body exercises may be a feasible therapeutic option across a wider range of developmental states.76–80

Complete data were only available for three yoga trials (two were open labeled) but were positive (Std diff in means 0.45; P < 0.015).62,64,67 Additional support for the benefit of yoga is found in two studies that are not represented in the analyses due to missing data. da Silva et al65 randomized 40 women with FM into an individually delivered intervention of “relaxing” yoga with or without additional tui na (massage). Both groups experienced clinically and statistically significant reductions in FIQ total scores, but there was no difference between treatment arms. Another trial (n = 10) noted a 28% improvement in FIQ total scores following eight weeks of gentle Hatha yoga.81 Three other CAM-based exercise studies were positive, but the interventions have yet to be replicated (pilates, Biodanza, and Resseguier with body movement and perception).58,60,72 Pilates is a series of nonimpact strength, flexibility, and breathing exercises designed by Joseph Pilates to promote inner awareness and core stability. Biodanza, literally “life-dance,” is a program of exercise that seeks to optimize self-development and deepen self-awareness. It most often uses dance, music, singing, and other movements to promote positive feelings, emotional expression, and connection to others and to nature. Resseguier is based on selected soft, low-impact gymnastic movements integrated with postural self-control, body alignment, and relaxation.

The data are consistent with an earlier rigorous report on “mindful” meditative movement therapies that concluded that these therapies were safe and effective for reducing sleep disturbances, fatigue, and depression, and improving quality of life. In subanalyses, the authors found that only yoga also relieved pain.50 However, that meta-analysis was comprised of only RCTs (n = 7). The current data extend these findings by demonstrating efficacy in a combination of FM symptoms and physical function, in a larger number of studies (n = 16).

This meta-analysis shows that research in the area was at an early state (six open-labeled studies), although regressing by Jadad score showed that there was a nonsignificant difference by effect size. This may indicate that early-stage studies are not likely to overreport effect.

This study does have limitations, the greatest being that six of the studies were open labeled. Others may argue, however, that open-labeled trials are appropriate in early-phase research when safety, feasibility, and effect sizes are being determined.82 Further, because open-labeled studies were included, community-based exercise studios and instructors were tested, leading the field toward effectiveness rather than efficacy outcomes. The studies were generally small (average n = 51.5; range 6 to 128 participants), limiting their power and the ability to test multiple comparisons or to profile responders’ reliability.

Another limitation is that the studies were largely conducted in middle-aged women, in America and Europe. This is important, as many CAM exercise studies have been conducted in India and China, where many of these traditions are rooted (however, those studies were not available in English or from major database sources). Moreover, differential effects may be found in men, minorities, children and elders. Another limitation is that all the trials employed a single interventionist and did not rate treatment expectancy. It is possible, therefore, that a charismatic or caring instructor, rather than the intervention itself, may have been responsible for the beneficial outcome, although this is less likely in the trials that have been replicated (qigong, tai chi, and yoga). Future effectiveness studies could use multiple instructors.

A goal of future studies should include the standardization of exercise protocols, including scripting for mindfulness, posture sequences, and a range of modifications based on the patient’s physical ability. To date, studies have varied considerably in exercise dose and have not always provided adequate description of the intervention to allow replication. Publishing protocols in peer-reviewed journals would provide opportunities for a more detailed description of the actual intervention and for pictures of the less common poses or physical strategies. This is important because people with FM are likely to be in poor physical condition, have joint hypermobility, and poor balance, and may be at an increased risk for injury.42

Future studies should incorporate validated scales such as the revised FIQ (FIQ-R), Brief Pain Inventory, Multidimensional Fatigue Inventory, Medical Outcomes Study Sleep Scale, and Multiple Ability Self-Report Questionnaire.83,84 Full reporting of data and statistical tests will allow for more seamless comparison in future trials. Some FM exercise studies now report the number needed to treat, based on Patients’ Global Impression of change (PGIC) scale, which allows for direct comparison with pharmaceutical trials.85 Similarly, work is underway to develop a responder index.86

In conclusion, the CAM-based exercise therapies reviewed are likely safe, and no serious adverse events were reported in any study. These exercise programs were somewhat efficacious for FM pain, with many studies reporting improvement in overall FM symptoms and physical function. However, the majority of studies had lower methodological quality. There is a need for large, rigorous trials with active parallel arms–such as traditional aerobic exercise compared with CAM-based exercise – to extend this body of evidence.