3.2. Review Characteristics

Finally 25 reviews could be included; characteristics are presented in Table 1.

Table 1

Characteristics of included systematic reviews.

The trials were published in the years 2000 (N = 1), 2003 (N = 1), 2007 (N = 1), 2008 (N = 1), 2009 (N = 5), 2010 (N = 4), 2011 (N = 2), 2012 (N = 2), and 2013 (N = 8).

Reviews investigated the following therapies: complementary and alternative therapies in general (N = 4) [50–53], exercised-based therapies (N = 4) [54–57], manipulative therapies (N = 3) [58–60], Mind/Body therapies (N = 5) [61–65], acupuncture (N = 4) [66–69], balneotherapy (N = 3) [70–72], phytotherapy (N = 1) [73], and homeopathy (N = 1) [74].

RCTs and quasi-RCTs only were included by 17 reviews [50, 51, 53, 55, 56, 58, 62–64, 66–69, 73, 74]; the other reviews also included controlled clinical trials [52, 54, 61, 65], observational trials [70], or any kind of study/review [57, 59, 60]. Reviews included from 4 to 60 trials with a total patient sample from 163 to 2897. Of the 25 reviews only 11 conducted meta-analyses.

All but three reviews [59, 68, 70] assessed the risk of bias of included trials, nine used the Jadad score [51, 54, 57, 58, 63, 64, 69, 73, 74], three used the Cochrane risk of bias assessment [56, 65, 66], three used the van Tulder scale [67, 71, 72], four applied self-adapted instruments or partial instruments [53, 55, 62], one the Oxford rating scale [60], one the Consort rating scale [52] and one the Scale for rating of quality of psychological trials in pain [61]. Only two reviews explicitly reported methods for formulating recommendation [56, 65], and both used the GRADE approach [75].

Quality assessment revealed maximal differences between the reviews; while some were of highest quality, others did only receive 2 of 11 possible points; see Table 2. While all reviews conducted a systematic literature search (which explicitly was an inclusion criterion), some of the reviews did not state an a priori design (N = 3) [59, 64, 68], and in some reviews either data extraction was not described or data were not extracted independently by two people (N = 12) [50–54, 57, 59, 60, 64, 68]. Only seven reviews listed included and excluded studies [55, 56, 61, 62, 65, 71, 72], and comprehensive study characteristics were provided by 14 reviews only [50, 54–57, 59, 61–63, 65, 67, 71, 73]. Risk of bias assessment was conducted in all but two reviews [59, 68], but the scientific quality was not appropriately included in formulating conclusions in nine of those reviews using a risk of bias assessment [50, 53, 55, 57, 59, 60, 62, 64, 69]. Of those reviews conducting meta-analyses [53, 55–57, 61, 62, 65–68, 71] only one used nonappropriate methods to combine study findings [57].

Table 2

AMSTAR rating table of included systematic reviews.

Likelihood of publication bias was not assessed in the majority of reviews, and two reviews with meta-analysis lacked this information [67, 68]. Conflict of interests was stated in most reviews with only 3 exceptions [52, 63, 72].

3.3. Results of Reviews

Evidence for or against the respective interventions is summarized in Table 3.

Table 3

Overview of conclusions for investigated therapies from included systematic reviews.

4.1. Homogeneous Findings

Positive results were found for tai chi, yoga, Mind/Body interventions in general, meditation-based interventions, mindfulness-based stress reduction, hypnosis or guided imagery, and balneotherapy or hydrotherapy. Negative results were reported on autogenic training, and inconclusive findings were reported for phytotherapy, however only one review each was available for autogenic training and phytotherapy.

4.2. Heterogeneous Findings

Heterogeneous findings, that is, positive and negative findings, were reported for qigong, chiropractic interventions, biofeedback, acupuncture, and nutritional supplements. In the following we will try to determine the source of heterogeneity.

For qigong three reviews found positive evidence [53, 56, 57], and two found negative evidence [50, 55]. Positive reviews were associated with later publication date, higher methodological quality, and three times as many included trials as negative reviews. Therefore positive results appear to be more valid than negative results.

For chiropractic interventions there is one review with limited evidence [60] versus two reviews that do not find positive evidence [52, 58]. No association between time of publication, methodological quality, and number of included trials could be found. However the positive review included all study types limiting the validity of any conclusion. Altogether for chiropractic interventions there seems to be no reliable positive evidence.

For biofeedback two reviews conclude positive results [52, 62] and two do not [50, 62]. One reason might be that only one of the reviews distinguished between EEG and EMG biofeedback [62]. So while for EMG biofeedback positive results have been found, no evidence was present for EEG biofeedback.

For acupuncture the following was found: the later reviews were published, the more trials were included, the more frequent meta-analyses were conducted, and the more conclusive results became. Secondly, results for quality of reviews were more diverse, but a trend towards higher quality reviews coming to a positive conclusion can be found. Therefore one could assume that acupuncture might be effective for FMS.

For nutritional supplement one review from 2003 [52] found positive results while one review from 2011 [53] found negative evidence. One could assume that nutritional supplements are not effective in the treatment of FMS.

Results of the overview are mostly in line with a previous overview [45]; however there were also substantial differences. While Terry et al. [45] only included 5 reviews, we could include 25. Results on hydrotherapy are positive in both overviews; the results for massage however are more diverse, which might have been a result of the inclusion of massage into the category of manual therapies. According to Terry et al. [45] also the use of acupuncture appears promising. In our review we also found that especially high quality reviews supported positive effects of acupuncture, and this finding might however be influenced by other factors such as time which might be connected to better methodological training or more eligible trials. For homeopathy on the other hand the quality of evidence was not supportive of positive or promising results yet.

The evidence found in this review is not necessarily in line with current treatment guidelines; for example, there is conflicting or lacking evidence for qigong, tai chi, or yoga; nevertheless they are explicitly recommended in the German guideline [75] because physical activity and self-care are considered important therapeutic goals in fibromyalgia patients. Positive evidence on massage on the other hand did not lead to a positive recommendation per se; instead availability, costs, and its negative influence on self-efficacy and coping behavior led to downgrading of recommendation.

4.3. Limitations

Despite all previous findings and conclusion, there are several limitations that might limit the results of this overview.

The first limitation concerns the inclusion criteria of the included reviews. While some reviews explicitly included only certain CAM therapies, others conducted comprehensive reviews for CAM in general. Since the definition of CAM is not always consistent, this might lead to a different direction of each review. Two reviews aiming at the efficacy of CAM therapies in general might therefore end up in contrary conclusions merely by different definitions of CAM. Also reviews that focused on physiotherapeutic interventions or exercise in general might also have summarized evidence for single CAM interventions; however since their focus was not on CAM only, they were not included. Furthermore reviews and the trials they were based on might have used different diagnostic criteria or procedures, which also might have influenced the results.

The methodological quality of reviews is probably one of the major limitations. Quality ranged from lowest to highest possible, and quality of reviews was also associated with the direction of results. Particularly for topics with only low quality reviews available this might limit the validity of results. This might be problematic since reviews and meta-analysis are considered highest evidence standard for guideline developers.

And another limitation concerns the use of terminology when summarizing evidence, risk of bias, methodological quality, or recommendations. Most reviews used different terms that were not consistent with recommendations; this might result in misinterpretation of results and conclusions.

This overview of review was supported by a grant from the Rut and Klaus Bahlsen Foundation.