Summary of findings
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
AR: absolute risk; CI: confidence interval; EA: electroacupuncture; FIQ: Fibromyalgia Impact Questionnaire; MD : mean difference; NNT: number needed to treat; RR: risk ratio; VAS: visual analogue scale
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.
Fibromyalgia is a musculoskeletal disorder characterised by widespread chronic pain and any number of co-morbidities, such as sleep disturbance, fatigue, stiffness, irritable bowel syndrome, headaches and mood disorders. It affects over 2% of the population and occurs predominantly in females (Wallace 2005). There was, until recently, no pharmacotherapy that effectively addressed all the symptoms associated with fibromyalgia (Lawson 2006). The United States of America's Food and Drug Administration (FDA) has approved 'Lyrica' (pregabalin), 'Savella' (milnacipran HCl) and 'Cymbalta' (duloxetine hydrochloride) (Boomershine 2009) for the treatment of fibromyalgia. In contrast, to date the European Medicines Agency has not approved any pharmacotherapy for the treatment of fibromyalgia (www.fibroaction.org), suggesting that the approved FDA drugs for fibromyalgia are not readily accessible by people globally. Non-drug therapies, such as cognitive behavioural therapy (CBT) and exercise, or a combination of the two approaches, are potentially beneficial for people with fibromyalgia (Nüesch 2012).
With respect to complementary and alternative medicine (CAM), acupuncture, a physical therapy of Traditional Chinese Medicine (TCM) that has been used to treat chronic pain for over two millennia in China, is promising for alleviating the symptoms associated with fibromyalgia (Bergman 2007). Among fibromyalgia sufferers, 91% had used CAM (Pioro-Boisset 1996) and one in five sufferers had sought acupuncture for treatment within two years of diagnosis (Bombardier 1996). In 1998, the National Institutes of Health Consensus Development Conference on Acupuncture stated that acupuncture may be used as an adjunct therapy for fibromyalgia (NIH 1998).
However, no therapy alone has been demonstrated to be universally superior to the others. Consequently, it was considered appropriate that when treating fibromyalgia a multidisciplinary approach be used (Arnold 2006). In the United States of America, approximately one million consumers use acupuncture annually (Burke 2006; Ezzo 2000). Furthermore, acupuncture is a relatively safe intervention (Vincent 2001) when compared with pharmacotherapies. Adverse events associated with acupuncture tend to be mild and short-lasting, such as lethargy and pain at the needling sites (MacPherson 2004).
The plausible mechanism of acupuncture analgesia is its effect on the central nervous system and consequent regulation of neurotransmitters and hormones. Acupuncture stimulates nerve fibres (e.g. A delta afferents), which in turn activate transmission neurons in the dorsal laminae of the spinal cord and further activate three levels of the endogenous pain modulation systems at the spinal cord, midbrain, thalamus and hypothalamus. The activation results in a cascade of pain-modulating endorphins, serotonin and noradrenaline, which contributes to analgesia (Cao 2002; Han 1997; Sims 1997).
Although the pathophysiology of fibromyalgia is not well understood, data suggest that ineffective descending inhibition of the central nervous system may cause an abnormal modulation of sensory inputs (such as mechanoreceptor inputs), resulting in pain (Price 2005). Acupuncture action enhances the function of the endogenous pain inhibition systems and therefore may be beneficial to people with fibromyalgia.
The World Health Organization (WHO) defines real acupuncture, in its broadest sense, as the insertion of needles into the human body surface for therapeutic purposes (WHO 2007). Throughout its history, different treatment styles of acupuncture have been developed in relation to needle size, depth of needling and duration of needle retention as well as deqi sensation. Deqi is the feeling of soreness, numbness, distension, heaviness or the electric shock sensation that occurs around a correctly placed and manipulated acupuncture needle (WHO 2007).
A number of different styles of acupuncture are presently in use, according to acupuncture point selection and stimulation modes. In clinical practice, the selection of acupuncture points for each patient is based on either a Chinese medicine diagnosis (individualised acupuncture treatment) or symptom alleviation (formula acupuncture treatment). Sometimes trigger points are also selected for needling and this may be described as dry needling. There is also micro-system acupuncture where needles are mainly inserted into defined points on an anatomical part of the body such as the head (scalp acupuncture), the ear (auricular acupuncture) or the hand (hand acupuncture). Needles can be deeply inserted into soft tissue and manipulated to elicit deqi (also known as traditional Chinese acupuncture) or superficially inserted into the skin without eliciting deqi (which may be described as Japanese acupuncture/meridian therapy).
Apart from, and in addition to, needles, acupuncture points or other points mentioned above can be stimulated using heat (such as moxibustion), with electrical current (known as electro-acupuncture), using mechanical pressure (acupressure) or using laser (laser acupuncture). Of all the forms of stimulation of acupuncture points, needling involving skin penetration (manual acupuncture) is the most commonly used method.
In 2007, a systematic review of acupuncture for fibromyalgia concluded that "acupuncture could not be recommended for fibromyalgia" (Mayhew 2007). However, it appears that this review neither searched for nor included studies from Chinese databases. In addition, new studies have been published. Therefore, there is a need to perform a thorough review to allow an up-to-date assessment of all available studies to determine the potential role of acupuncture in the management of fibromyalgia.
The present review aims to determine whether real acupuncture is more beneficial in terms of pain reduction, function and well-being improvement than placebo and other treatments and is safe in people with fibromyalgia. We examined the following comparisons:
- Acupuncture versus no acupuncture (e.g. wait list)
- Acupuncture versus placebo or sham acupuncture
- Acupuncture versus standard/usual care (e.g. cognitive behavioural therapy (CBT) and/or exercise and/or pharmacotherapy)
- Acupuncture as an adjunct therapy to standard/usual care (evaluating additional effect)
- A particular style of acupuncture versus another (e.g. deep needling with stimulation versus deep needling without stimulation)
Criteria for considering studies for this review
Types of studies
We considered randomised and quasi-randomised controlled clinical studies of acupuncture for treating patients with fibromyalgia. Quasi-randomised studies are those that do not strictly adhere to methods of true randomisation, e.g. location by the order of admission or date of birth. Inclusion of studies was not restricted according to language, type of publication or presence of blinding. We excluded studies from which we could not extract reported clinical outcomes or data for analyses.
Types of participants
Criteria for inclusion were participants of either gender, aged 18 and over, with a diagnosis of fibromyalgia according to the American College of Rheumatology (ACR) classification criteria for fibromyalgia (Wolfe 1990).
Types of interventions
Types of intervention were restricted to acupuncture that breaks the skin for therapeutic benefit (WHO 2002). Studies comparing different styles of acupuncture were also included. In addition, studies in which acupuncture was an adjunct therapy to other therapies (e.g. herbs, cupping, physiotherapy, exercise) were included, provided the control groups also received these therapies. Studies in which acupuncture points were stimulated with methods that did not break the skin, such as transcutaneous electrical nerve stimulation (TENS), infrared light, laser or digital pressure, were excluded. Acupuncture points refer to those points as defined in the Standard Acupuncture Nomenclature by the World Health Organization (WHO) (WHO 2002). Studies that used ashi acupuncture points (i.e. tender points) or trigger points were also to be included.
The control interventions included sham/fake/placebo acupuncture, other types of placebo control, non-acupuncture treatment, different styles of acupuncture or other treatment. An example of non-acupuncture treatment is a wait list. We considered standard care to be pharmacotherapy and/or exercise and/or CBT. When selecting studies that used sham/placebo acupuncture, we chose controls that did not intend to be an effective intervention, for example, needling on irrelevant acupuncture points, superficial needling or both. Other sham controls could have a disconnected electro-acupuncture stimulator, an inactive laser, mock TENS, infrared light or digital pressure. If there were sufficient studies, we planned to examine the differences between the various types of sham acupuncture (e.g. insertion verses non-insertion, deep needle verses shallow needle, on the acupuncture point versus off the acupuncture point). We excluded studies that did not provide adequate details of the control intervention.
Types of outcome measures
Included studies must have reported one or more of the following main clinical outcome measures related to pain, function and quality of life.
- Pain (e.g. visual analogue scale (VAS), numerical pain rating scale (NRS), McGill Pain Questionnaire (SF-MPQ), Multidimensional Pain Inventory (MPI) or Regional Pain Scale score)
- Physical function (e.g. 36-Item Short-Form Health Survey (SF-36, Physical) or Health Assessment Questionnaire (HAQ))
- Global well-being as rated by participants (e.g. Fibromyalgia Impact Questionnaire (FIQ), VAS rated by participants)
- Sleep (e.g. VAS of intensity, numerical sleep scale 1 to 10)
- Fatigue (e.g. VAS, Multidimensional Fatigue Inventory (MFI))
- Morning stiffness (e.g. numerical scale)
- Adverse events: proportion of participants who experienced an adverse event and proportion who withdrew due to adverse events
Provided the studies had main outcomes, we also considered any of the following minor outcomes.
- Tenderness (e.g. number of tender points or pain threshold of tender points as measured with a dolorimeter)
- Mental well-being (e.g. SF-36 (mental), Hamilton Depression Rating Scale (HAMD))
- Analgesic use (e.g. diary)
- Changes in fibromyalgia symptoms (e.g. observer-rated change in fibromyalgia symptoms (including that rated by physicians))
- Overall well-being rated by the study care givers
In the 'Summary of findings' table, we included the main outcomes of pain, physical function, global well-being, sleep, fatigue, stiffness and total adverse events (Arnold 2011).
Search methods for identification of studies
We initially searched the following databases from their inception to April 2008 as per protocol. We updated the search in May 2010 and January 2012. Search terms used included 'fibromyalgia' and 'acupuncture' and their variations (Figure 1).
|Figure 1. A flow chart of study selection. ('English' refers to English databases and 'Chinese' refers to Chinese databases).|
- Current Contents (http://apps.webofknowledge.com.ezproxy.lib.rmit.edu.au/) (Appendix 9)
We handsearched the bibliographies of review articles, excluded studies and textbooks on acupuncture, pain and fibromyalgia for additional studies. We contacted authors of published studies in an attempt to locate any unpublished studies.
Data collection and analysis
Selection of studies
One author (JD) searched the English language databases, while another author (ZZ) searched the Chinese language databases. These two authors independently examined the abstracts of the potential studies and obtained the full-text articles for consideration based on our pre-defined inclusion and exclusion criteria. We checked all references in the retrieved full-text English, Chinese and foreign language studies. Three potential papers were translated by the Cochrane Centres in Germany (one) and Italy (two). Three native speakers of Russian, Dutch and Spanish examined one paper each in these languages for potential studies. Four authors (JD, SS, ZZ, CX) with two in each group examined the English and Chinese studies, respectively, for inclusion/exclusion. Two authors (JD and ZZ) assessed the Harris 2008, Itoh 2010 and Targino 2008 papers. There were no disagreements between each pair of review authors.
Two author pairs (English: JD, SS and Chinese: ZZ, JSS) independently extracted data for each included study using our standard data extraction sheet. JD and ZZ extracted the data from Harris 2008, Itoh 2010 and Targino 2008. Data extracted included study characteristics, items related to the 'Risk of bias' tool and adverse events for each arm of the studies. We also extracted effect measures from each trial, including mean and standard deviation for continuous outcomes at or within one month of the end of the treatment; and number of events and number of participants in each group for dichotomous outcomes at the end of the treatment.
Assessment of risk of bias in included studies
Two review author groups, with two authors in each (JD, SS and ZZ, JSS), individually assessed the methodological quality of the English and Chinese studies, respectively, and incorporated them into the 'Risk of bias' tables. Items included in the tables are adequate sequence generation, allocation concealment (selection bias), blinding of the participants (performance bias), blinding of the assessor (detection bias), incomplete outcome data and its impact on the effect of estimate (attrition bias), and selective reporting (reporting bias). Using the extracted information, two authors (JD, ZZ) assessed whether they met the guidelines by selecting one of three choices: 'Yes', 'No' or 'Unclear' and reported the details of each decision in the allocated section of the table.
Assessment of the quality of the acupuncture treatments
To assess the quality of the acupuncture treatments, the two review author groups, all experienced clinical acupuncturists (minimum 10 years of experience each), used three instruments. We used the Standards for Reporting Interventions in Controlled Trials of Acupuncture (STRICTA) (MacPherson 2002) to extract the details of acupuncture intervention (Appendix 10), including acupuncture rationale, needling details, treatment regimen, co-interventions, practitioner background and control interventions, which are not addressed by other assessment tools. The purpose of STRICTA is to improve the reporting of interventions of controlled studies in acupuncture. This allows replication of the acupuncture treatment in other studies and clinical practice. As STRICTA does not offer a rating or scale to make a critical evaluation of the reporting, we further developed two rating systems to assess the adequacy of acupuncture treatment and confidence in the acupuncture diagnosis and treatment based on STRICTA data. Similar approaches have been used in other systematic reviews (Linde 2009; Scott 2006).
Adequacy of acupuncture treatment protocol
The rationale for examining the adequacy of treatment was to ensure that the study treatment protocol was comparable to routine clinical practice and the style of treatment was consistent with the techniques applied. For instance, a study claiming to be based on Chinese medicine but not eliciting deqi or only using a single needle would be considered inappropriate. Likewise, a treatment using appropriate Chinese medicine point selection but with only a single treatment would also be viewed as inadequate.
Assessment is based the on the following parts of the STRICTA table (Appendix 10).
- Acupuncture style
- Rationale for treatment/points used
- Literature sources
- Number of needles inserted
- Depth of insertion
- Response elicited
- Type of needle stimulation (electro/manual with or without tonification/dispersion etc.)
- Needle retention time
- Number of treatment sessions
- Frequency of treatments
From the list above, the review authors were required to judge if the acupuncture treatment performed was suitable for the style of acupuncture stated in the rationale for treatment. The experienced acupuncturists (JD, ZZ, JSS, SS) on the team rated the studies as low, medium or high according to whether the acupuncture treatment protocol was adequate. If there was insufficient information, we marked the study as 'insufficient information'.
Confidence in the diagnosis and treatment delivery
The determination of confidence in the administration of the acupuncture treatments was based on whether the person making the diagnosis, delivering the treatment or both was trained to the industry standard in that style. For example, L.Ac (licensed acupuncturist) would indicate meeting the USA standard. We used information about practitioners' training and practice background from STRICTA (Appendix 10) and information about the trial procedure to assess the level of confidence. For instance, it would be inappropriate to have an acupuncturist trained in Japanese/meridian style, i.e. shallow needling on acupuncture points, to provide Chinese acupuncture. Equally, we did not consider it appropriate that acupuncturists who had no Chinese medicine differential diagnosis training to deliver Chinese medicine diagnosis and treatment, unless a well-explained protocol was in place or pre-trial training was given and competence of the trial acupuncturists was assessed prior to the commencement of the study. The review authors rated their confidence at three levels: low, medium or high.
Measures of treatment effect
We analysed the data according to the Cochrane guidelines. To examine the immediate effect, we used completed data at up to one month after the end of the treatment. This method has been used in other reviews (Vickers 2012). To examine the long-term effect, we extracted data collected up to seven months after the end of the treatment. We plotted outcomes from each study as point estimates with corresponding 95% confidence intervals (CI) expressed as mean differences (MD) for continuous outcomes using the same scale, such as a 0 to 100 VAS for the measurement of pain or standardised mean differences (SMD) for continuous outcomes that used different scales, such as VAS and NRS for pain. We reported the number of adverse events and the number of drop-outs due to adverse events using risk ratios (RR). We also analysed data at one and up to seven months after treatment.
When ranges of data were presented, we calculated the standard deviations (SD) as advised and checked by the statistician from the Cochrane Musculoskeletal Group (CMSG) according to the Cochrane guidelines. With studies using more than one control arm we combined them as per the Cochrane guidelines (Higgins 2011).
Dealing with missing data
We contacted the authors of the included/excluded articles to obtain further information. We received responses to queries from the authors of Assefi 2005 (via the last author of the article); Deluze 1992; Harris 2005; Harris 2008; Harris 2009; Itoh 2010; Martin 2006; Sprott 1998; Targino 2008.
Assessment of reporting biases and small sample biases
For studies published after 1 July 2005, we screened the Clinical Trial Register via the International Clinical Trials Registry Platform of the World Health Organization (http://www.who.int/ictrp/en/) and compared the outcome measures described in the registry with those reported in the publications to assess whether selective reporting of outcomes was present (outcome reporting bias).
As planned, we compared the fixed-effect estimate against the random-effects model to assess the possible presence of small sample bias in the published literature given that the random-effects estimate of the intervention is more beneficial than the fixed-effect estimate in the presence of small sample bias (Higgins 2011). We found no difference between the two analyses in any outcome measures except for pain under the comparison of acupuncture versus sham acupuncture. The result of the random-effects model was more conservative than the fixed-effect model. Thus, we reported only the results from random-effects model.
If there were sufficient studies (> 10 studies with the same outcome), we planned to assess for publication bias using a funnel plot (Sutton 2000). This was not conducted due to an insufficient number of trials.
As recommended by the Cochrane Musculoskeletal Group editor, we used the random-effects model as the default for data synthesis.
Subgroup analyses and assessment of heterogeneity
When there were sufficient appropriate data, we planned subgroup analyses to assess the effect of different types of acupuncture: 1) manual acupuncture versus electro-acupuncture; 2) shallow needling versus deep needling; 3) different forms of sham/placebo acupuncture.
We used the I
We also planned to conduct sensitivity analyses to examine whether aspects of methodological quality influence the effect size. For example, did inadequate or unclear concealment of allocation or failure to blind outcome assessors change the overall effect estimate of our meta-analysis for pain?
'Summary of findings' tables
We presented the main outcomes (pain, physical function, global well-being (rated by participants), sleep, fatigue, stiffness and adverse events (Arnold 2011)) in 'Summary of findings' tables. The tables include an overall grading of the evidence using the GRADE approach of high, moderate, low and very low quality:
- High quality: further research is very unlikely to change our confidence in the estimate of effect.
- Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
- Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
- Very low quality: we are very uncertain about the estimate.
The 'Summary of findings' tables also contain the available data on the main outcomes as the calculations for statistically significant outcomes, and the number needed to treat (NNT) as recommended by The Cochrane Collaboration (Higgins 2011).
For continuous outcomes, we calculated the NNT using the Wells calculator software, available from the Cochrane Musculoskeletal Group editorial office (www.cochranemsk.org), which requires a minimal clinically important difference for input into the calculator. For pain we used a 1.5-point difference out of a 0 to 10 scale or 15 out of 0 to 100 scale as a minimal clinically important change. For global well-being, we used 14 out of 100 as a minimal clinically important change as recommended by Bennett 2009 for dealing with FIQ data. For sleep, fatigue, stiffness (Martin 2006) and physical function (Harris 2005), we used 13 out of 100 or 1.3 out of 10 as a minimal clinically important change (Bennett 2009). We calculated absolute change (benefit) from the mean difference or standard mean difference and expressed this as a per cent and in the original units, and calculated relative difference in the change from baseline as the absolute benefit divided by the baseline mean of the control group.
Description of studies
Total studies located for this review
We conducted our initial search in 2008, updated it in May 2010 and then updated it again in January 2012. The search period ranged from the inception of the databases to the end of December 2011. The study selection process is illustrated in Figure 1. The search resulted in 439 studies from the English databases and 63 from Chinese databases. After removing duplicates and irrelevant papers, we identified 49 acupuncture trials for fibromyalgia, including 17 papers in English, 24 in Chinese, four in German, two in Italian, one in Spanish and one in Swedish.
Studies excluded from the review
We excluded 40 studies for the following reasons:
- One measured blood flow in the muscles upon needling, but did not assess the clinical outcomes (Sandberg 2004).
- One examined brain images using position emission tomography of participants prior to the acupuncture treatment (Harris 2007).
Studies included in the review
Nine RCTs and one quasi-RCT were included. Five studies were conducted in the United States of America (Assefi 2005 n = 96; Harris 2005 n = 56; Harris 2008 n = 10; Harris 2009 n = 20; Martin 2006 n = 49), one in Switzerland (Deluze 1992 n = 55), one in Brazil (Targino 2008 n = 58), one in Japan (Itoh 2010 n = 13) and one quasi-RCT in China (Guo 2005 n = 38). All studies were published in English except for one published in Chinese.
In total 395 participants were involved. The authors of the selected papers explained their inclusion and exclusion criteria well except for Guo 2005, which only reported inclusion without exclusion criteria. All studies used acupuncture-naive participants except for Targino 2008, while Guo 2005 did not report this. Targino 2008 admitted participants into their study if they had not received acupuncture in the last 12 months. All studies reported using ACR fibromyalgia criteria for the selection of participants. However, confirmation of the diagnosis before commencement of the studies was reported in only four studies (Assefi 2005; Itoh 2010; Martin 2006; Targino 2008). Assefi 2005 used a researcher trained in tender point examination, Itoh 2010 obtained participants direct from fibromyalgia specialists at hospitals, Martin 2006 used a rheumatologist and Targino 2008 used a physician. The other five studies did not report whether or not they performed a confirmation of diagnosis (Deluze 1992; Guo 2005; Harris 2005; Harris 2008; Harris 2009).
Main outcome measurement tools varied. Five studies (Assefi 2005; Deluze 1992; Guo 2005; Itoh 2010; Targino 2008) used a VAS for measuring pain. Other measurement tools for pain included Regional Pain Score (Deluze 1992), Numeric Rating Scale (Harris 2005), Multidisciplinary Pain Inventory (Martin 2006) and Short Form of the McGill Pain Questionnaire (Harris 2008; Harris 2009). Three studies measured quality of life (SF-36). Assefi 2005 used SF-36 including the Physical and Mental component, Harris 2005 the SF-36 Physical component, and Targino 2008 reported all eight domains of SF-36. Two studies measured function using the FIQ (Itoh 2010; Martin 2006), which is labelled as global well-being in the current review. Five studies did not include a follow-up phase after the end of the treatment (Deluze 1992; Harris 2005; Harris 2008; Harris 2009; Itoh 2010). The remaining four had follow-ups at different time points with Assefi 2005 at the 3rd and 6th months after the end of the treatment; Guo 2005 at the 6th month; Martin 2006 at the 1st and 7th months; and Targino 2008 at the 3rd, 6th, 12th and 24th months.
All studies reported withdrawal, drop-outs or both except for Guo 2005, however the reported data indicated there were no drop-outs. The most common reason for withdrawal or drop-out was time constraint, followed by worsening of fibromyalgia symptoms and scheduling conflicts (e.g. appointments). The serious events for discontinuing participation were: one experienced heart attack from the acupuncture group (Assefi 2005); three hospitalisations with one from the acupuncture group and two from the control group; one ankle oedema from the acupuncture group (Deluze 1992), which was the only case that authors reported to be directly related to the acupuncture treatment (ankle oedema). The heart attack and the hospitalisation cases were not explained and connections with interventions were not established. The low drop-out rate may suggest the treatments were well tolerated by the participants.
Assessment of the quality of the acupuncture treatments
Reporting of the acupuncture treatments was generally adequate with the exception of Guo 2005 and Itoh 2010, being the poorest. Upon our request, some authors provided missing details via e-mails, however based on the published papers as they were, it would have been impossible to reproduce any of the studies accurately (Appendix 10).
Adequacy of acupuncture treatment protocol
Reporting of the rationale of the acupuncture treatment was insufficient, making it difficult for us to give a rating. Considering all 11 criteria, we rated the studies as 'medium' for all studies except for Guo 2005 and Itoh 2010, which we noted as 'insufficient data'. Six studies (Deluze 1992; Guo 2005; Harris 2005; Harris 2009; Itoh 2010; Martin 2006) actually stated the acupuncture style, while the rest did not report the style. Based on how they reported the acupuncture treatment, we inferred that it was 'formula acupuncture' (symptom alleviation), using a set of fixed acupuncture points.
None of the studies stated whether they had made a diagnosis according to Chinese medicine, except for Deluze 1992, but there was insufficient information to confirm this. Deluze 1992 stated that they individualised treatment and provided references to support their decision. However, the authors failed to report their Chinese medicine diagnosis.
The justification of acupuncture point selection deserved some attention. Only one study provided journal references for their decision on point selection and choice of electro-acupuncture (Deluze 1992). This study reported, however, only two mandatory acupuncture points without detailing the other eight optional points. Assefi 2005 commented that no gold standard existed for acupuncture point selection in the treatment of fibromyalgia, with the authors consulting three experienced acupuncturists in treating fibromyalgia for their point selection. Guo 2005 did not state what acupuncture points were used, instead naming the channels/meridians they used. They considered 'Back Shu' points along the Bladder meridian important to strengthen Liver, Spleen and Kidney when treating Bi-Syndrome, a Chinese medicine term for a series of rheumatic conditions, including fibromyalgia. Harris 2005 chose acupuncture points based on their "ability to relieve fibromyalgia symptoms in CM", however this was referenced to a textbook that does not specify Chinese medicine treatments for fibromyalgia. Harris 2008 and Harris 2009 referenced their 2005 trial for their acupuncture point selection. Itoh 2010 did not explain. Martin 2006 stated they used "strong regulatory points that commonly recur in acupuncture literature", yet provided neither reference nor stated the acupuncture points used on the back (published a small diagram of areas used). They also stated that their acupuncture point selection might not be optimal as judged by others, but did not provide the reason. Targino 2008 used "classical acupuncture points" and they referenced an acupuncture point location book that does not include fibromyalgia. The most commonly used point in all included studies was He Gu (LI4), followed by Zu San Li (ST36).
With the reporting of unilateral/bilateral needling details, Assefi 2005, Deluze 1992, Guo 2005 and Itoh 2010 did not state which side they inserted the needles on, while Martin 2006, Harris 2005 and Harris 2009 provided diagrams. Harris 2008 based the treatment on their 2005 trial. Except for Deluze 1992, Guo 2005 and Itoh 2010 the number of needles used was clearly stated. Reporting the depth of needle insertion was clear in all studies except for Guo 2005, which we thought was subcutaneous because the needling technique was "point to point threading" along the back meridian/channels. Reporting of elicitation of deqi was clear except for Assefi 2005, Guo 2005 and Itoh 2010; two stated "stimulation" without mentioning deqi (Assefi 2005; Itoh 2010) and the other did not report this (Guo 2005). The description of the type of needle stimulation/manipulation (e.g. lifting/thrusting/even etc.) was clear in only three studies (Harris 2005; Harris 2009; Itoh 2010). With the electro-acupuncture studies (Deluze 1992; Itoh 2010; Martin 2006) both reported Hz but not where the red/black clips went or what type of stimulation setting was used, such as 'continuous'. Needle gauge/length/manufacturer or material varied greatly and were not well reported by some.
Needle retention time ranged from 20 to 30 minutes. Four studies treated the participants for 30 minutes (Assefi 2005; Guo 2005; Harris 2005; Itoh 2010), two for 25 minutes (Harris 2008; Harris 2009) and two for 20 minutes (Martin 2006; Targino 2008). One did not report the needling duration (Deluze 1992). The median duration of acupuncture treatment sessions was four weeks (range 3 to 13). Two had six sessions (Deluze 1992; Martin 2006), with the remainder, nine (Harris 2008; Harris 2009), 10 (Itoh 2010), 18 (Harris 2005), 20 (Targino 2008), 24 (Assefi 2005) and 28 sessions (Guo 2005). Itoh 2010 was a cross-over study and we used data before cross-over for analysis; that is after five sessions of treatment. Frequency of treatments was similar in most studies, with twice weekly being the commonest (Assefi 2005; Deluze 1992; Martin 2006; Targino 2008). Two trials (Harris 2008; Harris 2009) had nine sessions over four weeks, another (Harris 2005) gave 18 sessions over 13 weeks, while one (Itoh 2010) delivered weekly and the remainder (Guo 2005) daily treatment.
Confidence in the treatment delivery
We rated our confidence that acupuncture treatments were appropriately delivered by skilled practitioners as 'high' for Assefi 2005 and Harris 2005 and 'medium' for Targino 2008. The remaining studies (Deluze 1992; Guo 2005; Harris 2008; Harris 2009; Itoh 2010; Martin 2006) we noted as 'insufficient data'.
Reporting of adverse events was inconsistent. Only two studies provided details of the number of events (Assefi 2005; Targino 2008). With the remaining studies, three did not report any (Harris 2005; Harris 2008; Harris 2009), which the author confirmed as nil. Two (Deluze 1992; Itoh 2010) cited them as withdrawals, one (Martin 2006) discussed them in the results without labelling them as adverse events and one (Guo 2005) did not report any.
None of the studies reported serious adverse events. The worst events that could directly be attributed to an acupuncture treatment were oedema of the left hand and ankle, despite a lack of evidence of a causal relationship (Targino 2008). Two cases of vasovagal symptoms reported by Martin 2006 were likely due to the posture of the participants. The author did not explain which treatment group the cases were in. In that study, all participants received acupuncture treatment seated without a back support for 20 minutes, which is not a common practice (participants are mostly in a prone or supine position), however it allowed the participants to be blinded. This type of adverse event could be prevented or reduced with correct posture as suggested by the author.
We undertook subgroup analyses within the comparison of acupuncture versus placebo/sham acupuncture to compare electro- and manual acupuncture for the outcome of pain, fatigue, sleep and global well-being (rated by participants) as data were available for those outcome measures. We also undertook a subgroup analyses of studies using sham acupuncture without breaking the skin (Assefi 2005; Harris 2008; Harris 2009; Martin 2006) versus studies using breaking-skin sham interventions (Assefi 2005; Deluze 1992; Harris 2005). One arm of the control (simulating) in Assefi 2005 and the sham controls in Harris 2008 and Harris 2009 used a tooth pick in an acupuncture guide tube, which has been shown in other studies of back pain to be indistinguishable (Sherman 2002). Martin 2006 indented the skin with a blunt probe and placed over the area a small circular plaster rigged with an acupuncture needle that stuck out; they stated that preliminary trials showed volunteers could not tell the difference.
We did not perform the planned publication bias funnel plot analysis due to an insufficient number of selected studies (Sutton 2000).
We could not conduct planned sensitivity analyses due to a lack of trials with and without adequate concealment of allocation; or with and without blinded outcome assessor under one comparison. For instance, under the comparison of real and sham acupuncture, all six studies were at low risk of selection bias with adequate concealment of treatment allocation, using no treatment as the control, and at low risk of detection bias with outcome assessors being blinded (Assefi 2005; Deluze 1992; Harris 2005; Harris 2008; Harris 2009; Itoh 2010; Martin 2006). Two studies were at higher risk of selection bias with unclear or inadequate allocation concealment and had a high risk of detection bias with inadequate or unclear blinding of outcome assessor (Guo 2005; Targino 2008). They were, however, under different comparison categories and contained only one study in each. Itoh 2010 had a moderate risk of bias as a non-acupuncture treatment control was used.
Risk of bias in included studies
All studies were described as RCTs. Adequate sequence generation and allocation concealment were well described and adequate in all included studies except for Guo 2005, which used order of admission for randomisation (quasi-randomisation) and Martin 2006, which did not say how the sequence was generated. All studies used acupuncture-naive participants except for Guo 2005 (who did not report this) and Targino 2008 (patients had not had acupuncture in the last 12 months). Four studies tested for assessment of masking/blinding (Assefi 2005; Harris 2005; Harris 2009; Martin 2006) and found no difference between groups. Five studies blindfolded their participants (Assefi 2005; Harris 2005; Harris 2008; Harris 2009), while Martin 2006 blocked the vision of the participants. All used blinded assessors except for Guo 2005, which did not report this. All studies showed no missing data except for Deluze 1992 and Itoh 2010, which did not include participants who dropped out from the study in their data analysis, while Guo 2005 did not report this specifically. All reported numbers lost to follow-up except for Guo 2005, which did not report this but no participant was missing from the reported data. In terms of selective reporting, only Assefi 2005, Harris 2005 and Targino 2008 were registered with the International Clinical Trials Registry Platform of the World Health Organization and we found that the outcome measures reported were same as those included in the published protocols. Among the remainder, one did not know about it (Martin 2006), one did not need to report it as it was before the establishment of the Registry (Deluze 1992), whereas the others did not list this (Guo 2005; Harris 2008; Harris 2009; Itoh 2010) (Figure 2; Figure 3).
|Figure 2. Methodological quality summary: review authors' judgements about each methodological quality item for each included study.|
|Figure 3. Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.|
Effects of interventions
See: Summary of findings for the main comparison Acupuncture versus non-acupuncture for treating fibromyalgia; Summary of findings 2 Acupuncture versus placebo or sham acupuncture for treating fibromyalgia; Summary of findings 3 Acupuncture versus medication for treating fibromyalgia; Summary of findings 4 Acupuncture as an adjunct therapy for treating fibromyalgia; Summary of findings 5 Deep invasive acupuncture stimulation versus non-stimulated acupuncture for treating fibromyalgia
1) Real acupuncture versus non-acupuncture treatment
One study in this category (Itoh 2010) included 13 participants and compared electro-acupuncture plus trigger point acupuncture with no acupuncture treatment. This was a cross-over study, and we used data before cross-over for analysis, that is after five sessions of treatment at the end of week five of a 10-week treatment programme.
Main outcome measure 1: Pain
Pain severity was measured using a VAS (100 mm). It showed a statistically significant reduction in pain for those treated with real acupuncture compared with no acupuncture at the end of treatment (mean difference (MD) -22.40 points on a 100-point scale; 95% confidence interval (CI) -40.98 to -3.82, P = 0.02), favouring acupuncture ( Analysis 1.1).
Main outcome measure 2: Global well-being; rated by participant
Global well-being was measured using the Fibromyalgia Impact Questionnaire (FIQ) 100-point scale. It showed a statistically significant group difference at the end of treatment (MD -15.40 points on a 100-point scale; 95% CI -25.62 to -5.18, P = 0.003), favouring acupuncture ( Analysis 1.2).
Main outcome measure 3: Sleep
Sleep was measured using the subset 'rested' on the FIQ. It showed no statistically significant group difference at the end of treatment (MD -0.40 points on a 10-point scale; 95% CI -1.01 to 0.21, P = 0.20) ( Analysis 1.3).
Main outcome measure 4: Fatigue
Fatigue was measured using the subset 'fatigue' on the FIQ. It showed a statistically significant group difference at the end of treatment (MD -1.10 points on a 10-point scale; 95% CI -1.98 to -0.22, P = 0.01), favouring acupuncture ( Analysis 1.4).
Main outcome measure 5: Stiffness
Stiffness was measured using the subset 'stiffness' on the FIQ. It showed a statistically significant group difference at the end of treatment (MD -0.90 points on a 10-point scale; 95% CI -1.66 to -0.14, P = 0.02), favouring acupuncture ( Analysis 1.5).
Main outcome measure 6: Adverse events
No adverse events were reported, only withdrawals and drop-outs. One patient from the acupuncture group and two from the control group left the study as their condition was not improving.
Minor outcome measure 1: Mental well-being
Mental well-being was measured using the subset 'depression' on the FIQ. It showed no statistically significant group difference at the end of treatment (MD -0.50 points on a 10-point scale; 95% CI -1.10 to 0.10, P = 0.10) ( Analysis 1.7).
Long-term effect of acupuncture
There was no follow-up and long-term effect was not measured.
2) Real acupuncture versus placebo or sham acupuncture
Main outcome measure 1: Pain up to one month after treatment
Six studies totaling 286 participants were in this category (Assefi 2005; Deluze 1992; Harris 2005; Harris 2008; Harris 2009; Martin 2006). Measurement tools used included VAS (0 to 10 cm and 0 to 100 mm), numerical pain rating scale (NRS), Multidimensional Pain Inventory (MPI) and McGill Pain Questionnaire (SF-MPQ). Pooled analysis of the six studies showed no statistically significant difference between the groups in reducing pain (standardised mean difference (SMD) -0.14; 95% CI -0.53 to 0.25, P = 0.48; corresponding to a reduction of 2.8 points on a 100-point scale) ( Analysis 2.1). Moderate heterogeneity was found (I
Pain subgroup analysis (electro- versus manual acupuncture)
Pooled subgroup analysis of two electro-acupuncture studies, including 104 participants (Deluze 1992; Martin 2006), indicated that real electro-acupuncture was statistically significantly better than sham electro-acupuncture in reducing pain (SMD -0.63; 95% CI -1.02 to -0.23, P = 0.002, about 13 points on a 100-point scale) ( Analysis 2.1) up to one month after treatment, with low heterogeneity (I
Main outcome measure 2: Physical function (SF-36) up to one month after treatment
One study totaling 56 participants was in this category (Harris 2005). Physical function was measured with the SF-36. Analysis of the study indicated that sham manual acupuncture was superior to manual acupuncture in improving SF-36 physical function (MD -5.80 points on a 100-point scale; 95% CI -10.91 to -0.69, P = 0.03, Analysis 2.4).
Main outcome measure 3: Global well-being: rated by participants up to one month after treatment
Three studies totaling 200 participants were in this category (Assefi 2005; Deluze 1992; Martin 2006). Measurement tools included VAS and FIQ. Pooled analysis of the three studies showed no statistically significant difference between real and sham acupuncture (SMD 0.29; 95% CI -0.44 to 1.01, P = 0.44, 5.8 points on a 100-point scale) ( Analysis 2.5), with high heterogeneity (I
Global well-being subgroup analysis (electro-acupuncture versus manual acupuncture)
Pooled subgroup analysis of two electro-acupuncture studies with 104 participants (Deluze 1992; Martin 2006) indicated that real electro-acupuncture was statistically significantly better than sham electro-acupuncture in reducing global well-being as rated by participants (SMD 0.65; 95% CI 0.26 to 1.05, P = 0.001, about 11 points on a 100-point scale) ( Analysis 2.5), up to one month after treatment, with low heterogeneity (I
Main outcome measure 4: Sleep up to one month after treatment
Three studies totaling 200 participants were in this category (Assefi 2005; Deluze 1992; Martin 2006). Sleep quality was measured with a VAS sleep scale and the subset 'rested' on the FIQ. Pooled analysis showed no statistically significant difference with real acupuncture when compared with sham interventions (SMD 0.16; 95% CI -0.29 to 0.61, P = 0.49, about 3.2 points on a 100-point scale) ( Analysis 2.7), with moderate heterogeneity (I
Sleep subgroup analysis (electro-acupuncture versus manual acupuncture)
Pooled subgroup analysis of two electro-acupuncture studies with 104 participants (Deluze 1992; Martin 2006) indicated that real electro-acupuncture was statistically significantly better than sham electro-acupuncture in improving sleep quality (SMD 0.40; 95% CI 0.01 to 0.79, P = 0.05, about eight points on a 100-point scale) ( Analysis 2.7) up to one month after treatment, with low heterogeneity (I
Main outcome measure 5: Fatigue up to one month after treatment
Three studies totaling 201 participants were in this category (Assefi 2005; Harris 2005; Martin 2006). Fatigue was measured with a VAS, the Multidimensional Fatigue Inventory (MFI) and the subset 'fatigue' on the FIQ. Pooled analysis showed no statistically significant difference between real and sham acupuncture in reducing fatigue (SMD -0.10; 95% CI -0.81 to 0.61, P = 0.78, about 1.7 points on a 100-point scale, Analysis 2.9), with high heterogeneity (I
Fatigue subgroup analysis (electro-acupuncture versus manual acupuncture)
Subgroup analysis of one electro-acupuncture study (Martin 2006) of 49 participants indicated that real electro-acupuncture was statistically significantly better than sham electro-acupuncture in reducing fatigue (SMD -0.85; 95% CI -1.44 to -0.27, P = 0.004, about 15.3 points on a 100-point scale) ( Analysis 2.9) up to one month after treatment. Pooled subgroup analysis of two manual acupuncture studies (Assefi 2005; Harris 2005) with 152 participants showed no group difference between real and sham manual acupuncture in reducing fatigue (SMD 0.26; 95% CI -0.08 to 0.61, P = 0.13, about 4.3 points worse on a 100-point scale) ( Analysis 2.9), with low heterogeneity (I
Main outcome measure 6: Stiffness up to one month after treatment
Two studies totaling 104 participants using electro-acupuncture treatment (Deluze 1992; Martin 2006) were in this category. Stiffness was measured as minutes (Deluze 1992) and the subset 'stiffness' on the FIQ (Martin 2006). Pooled analysis showed that real electro-acupuncture was statistically significantly better than sham electro-acupuncture in improving stiffness (SMD -0.45; 95% CI -0.84 to -0.06, P = 0.02, nine points on a 100-point scale) ( Analysis 2.11) up to one month after treatment, with low heterogeneity (I
Main outcome measure 7: Adverse events
There were no serious adverse events reported. Minor adverse events were reported in 11 out of 113 participants in the acupuncture groups and 58 out of 156 in the control groups (risk ratio (RR) 0.44; 95% CI 0.12 to 1.63, P = 0.22, Analysis 2.13), with moderate heterogeneity (I
Minor outcome 1: Mental well-being up to one month after treatment
One study totaling 49 participants using electro-acupuncture was in this category (Martin 2006). Mental well-being was measured with the subset 'depression' of the FIQ. Analysis showed a statistically significantly better result with real electro-acupuncture when compared to sham electro-acupuncture in improving mental well-being (MD -1.70 points on a 10-point scale; 95% CI -3.13 to -0.27, P = 0.02) ( Analysis 2.14), up to one month after treatment.
Minor outcome measure 2: Analgesic use (number of tablets) up to one month after treatment
One study with 55 participants using electro-acupuncture treatment (Deluze 1992) measured analgesic use by the number of tablets per week. There was no difference between real and sham electro-acupuncture (MD -3.20 tablets less per week; 95% CI -10.20 to 3.80, P = 0.37) ( Analysis 2.16).
Minor outcome measure 3: Analgesic use (number of participants) up to one month after treatment
One study with 80 participants using manual acupuncture treatment (Assefi 2005) measured analgesic use by number of participants taking analgesics. There was no difference between real and sham manual acupuncture (RR 0.94; 95% CI 0.66 to 1.32, P = 0.71) ( Analysis 2.17).
Minor outcome 4: Tenderness up to one month after treatment
One study with 55 participants using electro-acupuncture treatment (Deluze 1992) measured pressure pain threshold (kg/cm
Minor outcome measure 5: Overall well-being: rated by care giver
One study with 55 participants using electro-acupuncture treatment (Deluze 1992) reported overall well-being rated by care givers. Analysis showed a statistically significantly better result with real electro-acupuncture when compared with sham electro-acupuncture (MD 2.00 points on a 10-point scale; 95% CI 0.81 to 3.19, P = 0.001) ( Analysis 2.19), up to one month after treatment.
Long-term effects of acupuncture
Two studies (Assefi 2005; Martin 2006) measured long-term effects of acupuncture for up to seven months after the end of the treatment. There was no difference between real and sham acupuncture on any outcome measures, including pain ( Analysis 2.2), global well-being ( Analysis 2.6), sleep ( Analysis 2.8), fatigue ( Analysis 2.10), stiffness ( Analysis 2.12) and mental well-being ( Analysis 2.15). Subgroup comparison indicated that electro-acupuncture was not statistically significantly better than manual acupuncture at improving any of the outcomes at seven months after treatment.
Sham non-invasive (not breaking skin) acupuncture versus sham invasive (breaking skin) acupuncture
Four studies (Assefi 2005; Harris 2008; Harris 2009; Martin 2006) with 116 participants using non-invasive sham acupuncture were compared with three studies (Assefi 2005; Harris 2005; Deluze 1992) with 170 participants using invasive sham interventions. There was no statistically significant difference between the two subgroups on pain rating (Chi
3) Real acupuncture versus standard or usual care (medication)
One study in this category (Guo 2005) included 38 participants and compared manual acupuncture with Western medicine (amitriptyline).
Main outcome measure 1: Pain at up to one month after treatment
Pain severity was measured using a VAS. It showed a statistically significant group difference favouring acupuncture (MD -17.30 points on a 100-point scale; 95% CI -24.13 to -10.47, P < 0.00001) ( Analysis 3.1).
Main outcome measure 2: Adverse events
No adverse events were reported, however all participants were included in the final analyses. No withdrawals or drop-outs were reported either.
Minor outcome measure 1: Number of tender points at up to one month after treatment
A statistically significant group difference was shown for number of tender points, favouring acupuncture (MD -4.00 number of tender points; 95% CI -6.73 to -1.27, P = 0.004) ( Analysis 3.3).
Long-term effect of acupuncture at the sixth month after treatment
The authors stated there was follow-up at six months but no data were provided.
The poor reporting of the trial raises questions about its quality. For example, the authors claimed 'cure' of 12 fibromyalgia participants, with nine in the acupuncture group and three in the control, without a definition of what 'cure' was, except for saying "signs and symptoms free with no tender point" without further explanation as to which time point these were measured at.
4) Real acupuncture as an adjunct therapy
One study in this category (Targino 2008) with 58 participants compared manual acupuncture plus standard therapy, which included tricyclic antidepressants and exercise, with standard therapy alone.
Main outcome measure 1: Pain at up to one month after treatment
Pain severity was measured using a VAS. It showed a statistically significant group difference favouring acupuncture (MD -3.00 points on a 10-point scale; 95% CI -3.90 to -2.10, P < 0.00001) ( Analysis 4.1).
Main outcome measure 2: Adverse events
There were no serious adverse events reported. No group difference in the number of minor adverse events was found (RR 3.57; 95% CI 0.18 to 71.21, P = 0.40) ( Analysis 4.3).
Minor outcome measure 1: Tenderness - number of tender points below kg/cm
2 up to one month after treatment
Tenderness was measured the number of tender points below the threshold. The results showed a statistically significant group difference favouring acupuncture (MD -4.50 number of tender points; 95% CI -6.20 to -2.80, P < 0.00001) ( Analysis 4.4).
Minor outcome measure 2: Tenderness - mean pressure threshold by pressure algometry at end of treatment
Tenderness was measured with the mean pressure threshold (kg/cm
Long-term effect of acupuncture (follow-up at six months)
At the six-month follow-up, the acupuncture as an adjunct therapy group continued to be better than the standard therapy alone group for tender points (MD -2.00 number of tender points; 95% CI -3.51 to -0.49, P = 0.009) ( Analysis 4.5) and mean pressure pain threshold (MD 0.60 kg/cm
5) A particular style of acupuncture versus another (deep invasive needling with stimulation (deqi) (T/S) versus deep invasive needling without stimulation (T/O))
Two different styles of acupuncture, deep needling using manual acupuncture on the point with stimulation to achieve deqi as in traditional Chinese acupuncture (T/S) versus deep needling on the point without stimulation (T/O), were compared in one study of 41 participants (Harris 2005).
Main outcome measure 1: Pain at the end of the treatment
Pain was measured using the NRS. It showed no statistically significant difference between the two interventions (MD 0.30 on a 10-point scale; 95% CI -18.34 to 18.94, P = 0.97) ( Analysis 5.1).
Main outcome measure 2: Physical function (SF-36) at the end of the treatment
Physical function was measured using the SF-36 (physical). There was no group difference between the two interventions (MD -5.50 points on a 100-point scale; 95% CI -11.43 to 0.43, P = 0.07) ( Analysis 5.2).
Main outcome measure 3: Fatigue at the end of the treatment
Fatigue was measured using the MFI. There was no group difference between the two interventions (MD 1.10 points on 20-point scale; 95% CI -1.41 to 3.61, P = 0.39) ( Analysis 5.3).
Summary of main findings
Out of 124 studies screened, we identified nine randomised controlled trials (RCTs) involving 395 participants. Most of the studies were excluded because of insufficient data. All selected studies used a fixed set of acupuncture points (formula acupuncture) with six using manual acupuncture and three electro-acupuncture.
When compared with the group not receiving acupuncture, the acupuncture treatment group improved in terms of pain, global well-being, fatigue and stiffness, but not sleep. We found no difference between real and sham acupuncture on any outcome measures except for stiffness and physical functioning. Stiffness was measured in two electro-acupuncture studies, which showed a moderate effect of electro-acupuncture over sham interventions. On the contrary, sham intervention produced better improvement in physical functioning. Subgroup analyses demonstrated that electro-acupuncture was consistently better than manual acupuncture in eliciting moderate benefits on pain, fatigue, sleep and global well-being as rated by participants.
Comparing acupuncture with standard pharmacotherapy (amitriptyline), the result of a single trial favoured acupuncture for pain and muscle tenderness. The quality of that paper was poor, affecting the validity of the result. One study examined acupuncture as an adjunct therapy to standard care comprising of a tricyclic antidepressant and exercise and found an additive effect of acupuncture for pain relief and reduction of muscle tenderness.
Measurement of treatment effects was within one month of the end of treatment. Many effects of acupuncture were short-lasting and not maintained at six to seven-month follow-ups. Adverse events reported were mild and no difference between real and sham acupuncture, or other control interventions, was found.
Overall, there is a low to moderate level of evidence indicating that formula acupuncture could be a safe option for fibromyalgia. There is a low to moderate level of evidence that acupuncture is better than non-acupuncture, Western medication and standard therapy in improving pain and stiffness for people with fibromyalgia. There is a moderate level of evidence that the effect of acupuncture does not differ from sham acupuncture in terms of reduction of pain, fatigue, improvement of sleep or global well-being. Subgroup analyses indicate that electro-acupuncture was consistently better than sham interventions. When considering acupuncture, electro-acupuncture could be an effective modality for short-term pain relief. We reached these gradings because of the small sample sizes in all included studies, although the risks of biases were low. None of the studies had more than 50 participants in any of the trial arms and there is a possibility of random errors due to small sample size. As a result, our findings warrant further research with an adequate sample size and long-term follow-up.
Quality of the evidence
With additional information provided by the authors, we were able to ascertain that the risk of bias of the included studies was acceptable in all studies except for three pragmatic trials (Guo 2005; Itoh 2010; Targino 2002). The other six studies (Assefi 2005; Deluze 1992; Harris 2005; Harris 2008; Harris 2009; Martin 2006) that compared acupuncture with sham controls included acupuncture-naive participants, adopted adequate randomisation procedures, blinded assessors, properly recorded drop-outs and five out of six studies used intention-to-treat analysis. Five studies also blinded participants (Assefi 2005; Harris 2005; Harris 2008; Harris 2009; Martin 2006), tested the blinding of participants (except for Harris 2008) and reported that the participants could not tell to which group they were allocated. We consider that the risk of bias is low in these six out of the nine included studies.
Acupuncture versus non-acupuncture
There is low-quality evidence based on one trial (electro-acupuncture, 13 participants) that acupuncture significantly reduced pain and stiffness and improved global well-being and fatigue when compared with the non-acupuncture group. We downgraded the quality of evidence because participant blinding was impossible, intention-to-treat analysis was not used and due to the small sample size. Adverse events were not reported. Three participants withdrew due to ineffective treatment ( Summary of findings for the main comparison).
Acupuncture versus sham acupuncture
There is moderate-quality evidence based on data from six trials (289 participants) (combined manual and electro-acupuncture) that acupuncture did not significantly reduce pain compared with sham acupuncture, but subgroup analysis indicates that electro-acupuncture was significantly better than sham electro-acupuncture; whereas manual acupuncture showed no significant difference from sham manual acupuncture. Based on one manual acupuncture trial (56 participants), there is moderate-quality evidence that sham acupuncture improved physical function better than acupuncture, with the quality being downgraded due to inconsistency with other outcome measures. Global well-being has moderate-quality evidence based on three trials (203 participants, combined manual and electro-acupuncture) that acupuncture was not better than sham acupuncture, with subgroup analysis indicating that electro-acupuncture showed greater improvement than manual acupuncture did. Sleep had moderate-quality evidence with data from three trials (203 participants) that acupuncture (combined electro- and manual) did not significantly improve sleep time over sham acupuncture. However, subgroup analysis indicates that electro-acupuncture improved sleep quality significantly. We downgraded the quality of evidence for 'pain', 'global well-being' and 'sleep' due to one study (Deluze 1992) not using intention-to-treat analysis. There is high-quality evidence based on three trials (204 participants, combined manual and electro-acupuncture) that acupuncture did not significantly reduce fatigue, however subgroup analysis indicates that electro-acupuncture reduced fatigue significantly. Stiffness has moderate-quality evidence based on two trials (104 participants, electro-acupuncture only) that acupuncture significantly reduced stiffness compared with sham acupuncture and was downgraded due to one study (Deluze 1992) not using intention-to-treat analysis. Moderate-quality evidence from six trials (289 participants) showed no statistically significant difference between real and sham acupuncture in the number of adverse events associated with acupuncture. We downgraded the quality due to the small sample size within the studies. One in six people who had acupuncture reported adverse events, in contrast to one in three in the sham treatment groups. Such events were minor and lasted less than one day ( Summary of findings 2).
Acupuncture versus medication
There is low-quality evidence based on one trial (38 participants, manual acupuncture only) that acupuncture significantly reduced pain when compared with medication. We downgraded the quality of evidence due to the poor reporting of the paper. No details about adverse events were reported. From the data it would appear that there were no drop-outs or withdrawals ( Summary of findings 3).
Acupuncture as an adjunct therapy
There is moderate-quality evidence based on one trial (manual acupuncture, 58 participants) that acupuncture significantly reduced pain as an adjunct therapy to medication and exercise. We downgraded the quality of evidence due to the small sample size. There were two adverse events in the acupuncture group, which was not significantly different from the control group. We downgraded the evidence due to small sample size ( Summary of findings 4).
Deep needling with stimulation versus deep needling without stimulation
There is moderate-quality evidence based on one trial (manual acupuncture, 41 participants) which showed that there was no significant difference between the two needling styles in the reduction of pain or improvement of their physical function. We downgraded the evidence due to small sample size ( Summary of findings 5).
Comparison with other systematic reviews
Three meta-analyses of RCTs of acupuncture for the treatment of fibromyalgia have recently been published (Cao 2010; Langhorst 2010; Martin-Sanchez 2009) with conflicting conclusions. Cao 2010 considered that acupuncture could be a safe and effective therapy for treating fibromyalgia, while Martin-Sanchez 2009 and Langhorst 2010 concluded that acupuncture was neither effective nor could the effect be distinguished from bias.
In comparison, the present review has the following strengths: our search was comprehensive, including both English and Chinese databases; we adopted strict trial selection criteria based on the American College of Rheumatology (ACR) requirements; the adequacy of acupuncture protocol and treatment delivery were assessed by experts in the field; we contacted all authors to obtain additional data; we extracted data for multiple outcome measures and we limited acupuncture intervention to needling only. Laser acupuncture differs from manual or electro-acupuncture due to its mechanism and depth of stimulation. We selected studies using invasive needling acupuncture as the main or adjunct therapy. Consequently, all studies identified for inclusion in those three reviews have been either included in or excluded from our review. Martin-Sanchez 2009 only examined pain and did not include other outcome measures that are associated with fibromyalgia. Langhorst 2010 included most of the studies selected for this review. Langhorst 2010 found the reduction of pain to be significantly better in the real acupuncture group post-treatment (standardised mean difference (SMD) -0.25; 95% confidence interval (CI) -0.49 to -0.02, P = 0.04) and their effect size was smaller when compared with our data (SMD -0.42) due to inclusion of two studies that we excluded (Lautenschlager 1989; Sprott 1998) for not using ACR criteria or not reporting confirmable data, respectively. They also excluded one study which was included in our review (Harris 2008). Langhorst 2010 went on to conclude that "significant reduction of pain was only present in studies with risk of bias" because they considered three positive studies (Assefi 2005; Deluze 1992; Martin 2006) as having a high risk of bias. As indicated in Figure 3 and Figure 2, our data do not support this claim as explained above in the 'Quality of the evidence' section. We also included three other studies comparing acupuncture with non-acupuncture, medication and standard therapy.
Outcome measures for fibromyalgia
The top three core domains for outcome measures in any pain studies as recommended in IMMPACT are pain, function and emotion (Dworkin 2010). Furthermore, the 2010 ACR preliminary diagnostic criteria identified pain as well as a range of non-pain symptoms, for instance cognitive symptoms, headache and irritable bowel syndrome (Wolfe 2010). In all studies included in this review, the measurement tools for pain were adequate and validated. However, only four studies measured function or quality of life, two studies measured mental well-being, with one using the depression and anxiety sub-scales of the Fibromyalgia Impact Questionnaire (FIQ) and the other using the mental component of the SF-36. None measured cognition or somatic symptoms apart from sleep and fatigue. The FIQ, a condition-specific, validated function measure, was used only in two studies (Itoh 2010; Martin 2006). The FIQ has been in existence for 18 years and translated into eight languages (Bennett 2005). Targino 2008 explained they could not use the FIQ because the Brazilian version had not been validated at the time of the trial. The other two (Assefi 2005; Harris 2005) used the SF-36, which assesses quality of life but not function in fibromyalgia. The data from Assefi 2005 could not be used for the current review due to incorrect labelling of data in the published papers. To our knowledge, there is only one trial comparing the FIQ and the SF-36 in fibromyalgia and rheumatoid arthritis participants (Birtane 2006). The total score on the FIQ was moderately correlated with physical function, physical role and bodily pain on the SF-36, but not with other domains. Sub-scales of the FIQ were not correlated with relevant domains on the SF-36. For instance, bodily pain on the SF-36 was correlated with stiffness but not pain on the FIQ, and mental health on the SF-36 was correlated with anxiety but not depression on the FIQ. For this reason, we analysed data from the SF-36 and the FIQ separately, with the SF-36 measuring physical and mental function and the FIQ measuring overall well-being. We question the suitability of the SF-36 for measuring function in fibromyalgia participants. Physical function measured with the SF-36 physical domain was poorer in the acupuncture group than in the sham intervention group. The change was statistically, but not clinically, significant. We could not explain this finding. Given that the finding was from one trial, future studies with large sample sizes might impact on the direction of changes.
Fibromyalgia is characterised by widespread chronic pain as well as a range of non-pain symptoms and co-morbidities. A recent review indicates that when rating global improvement, fibromyalgia participants consider not only pain reduction, but also improvement in fatigue, functioning, mood and daily living (Hudson 2009). It is therefore important to assess a wide range of measures when examining the effect of any interventions for fibromyalgia. The FIQ consists of measures of pain, fatigue, sleep and physical and emotional functioning and is an ideal outcome measurement tool. OMERACT (Outcome Measures in Rheumatology) participants have agreed that pain, tenderness, fatigue, participant global rating or well-being, function and sleep are the core outcomes to be measured (Mease 2009). The FIQ measures most of these domains. In future studies, researchers should consider using the FIQ or include the assessment of the key co-morbidities and emotional and cognitive aspects of fibromyalgia. Such a design would help identify the specific effects of acupuncture on fibromyalgia.
Quality of acupuncture treatment
Overall, the treatment was adequate in terms of frequency (two to three sessions per week), number of treatments (six to 28 sessions) and length of each session of treatment (20 to 30 minutes). However, reporting of some details of the treatment, such as needling depth and unilateral or bilateral needling, were unavailable. It is important that both authors and journals adhere to the STRICTA guidelines for adequate reporting of acupuncture treatments.
The major weakness of reporting of acupuncture treatments was a lack of rationale for the acupuncture treatment in all but one included trial (Deluze 1992), and justification of point selection was rarely provided, which could be due to a lack of standard Chinese medicine syndrome differential criteria for fibromyalgia. Acupuncture therapy in a clinical setting relies on the syndrome pattern differentiation for accurate point selection. None of the studies included offered a diagnosis or attempted a syndrome pattern differentiation for fibromyalgia according to Chinese medicine.
The current Western medicine diagnosis of fibromyalgia does not result in a single entity or homogenous group. Reliance on the two main criteria of chronic widespread pain and 11 out of 18 tender points according to the ACR diagnostic criteria has been criticised for not considering other important symptoms and co-morbidities (Mease 2005; Wilke 2009; Wolfe 2003). It was never intended for the ACR criteria to be used for clinical diagnosis but rather for research as a standardised definition of fibromyalgia, and there is no gold standard for fibromyalgia diagnosis (Katz 2005). To address this, Western medical research is being undertaken to examine the differentiation of fibromyalgia into subgroup/symptom clusters (Muller 2007; Schneider 2005; Wilson 2009). The 2010 ARC preliminary criteria (Wolfe 2010) are a positive step towards clinically orientated approaches.
Although fibromyalgia is not a diagnosis of Chinese medicine, the types of pain and co-morbidities associated with it may fit into the Chinese medicine diagnostic concept of Bi-Syndrome, documented 2500 years ago (Ni 1995). Dividing Bi-Syndrome into a number of patterns depends on the characteristics of pain, as well as the accompanying signs and symptoms, which allows syndrome pattern differentiation, leading to an individualised approach to treatment that is part of the clinical decision-making process within traditional/clinical acupuncture practice. However, the Chinese medicine diagnostic criteria for fibromyalgia are yet to be developed. This might explain why only formula acupuncture treatments were used in all included studies. It is common in modern Chinese medicine for a disease or condition in Western medicine to be given a set of differentiation diagnoses so that understanding of subgroups can be standardised to provide guidance for treatment. There is an urgent need for research into developing Chinese medicine syndrome differentiation diagnostic criteria.
Modes of acupuncture
The current data do not allow us to conclude the best acupuncture stimulation mode for the treatment of fibromyalgia. However, only one trial examined the two types of stimulation, and found deep needling with stimulation did not differ from deep needling without stimulation. That is to say deqi, one of the essences of acupuncture stimulation, might not play the expected role in the treatment of fibromyalgia. Subgroup analyses indicate that electro-acupuncture was consistently superior to manual acupuncture for a number of major outcome measures. However, no trial directly compared electro- with manual acupuncture.
Sensitivity of the nervous system of fibromyalgia participants may influence the treatment outcome of different types of stimulation, however dose of treatment could be another explanation. A recent trial in healthy humans compared the effect of sham acupuncture with manual and electro-acupuncture on electrical pain thresholds (Zheng 2010). The researchers found that electro-acupuncture induced the best analgesia, followed by manual acupuncture, then the sham intervention. In electro-acupuncture, the stimulation was delivered constantly for 25 minutes, whereas in manual acupuncture the stimulation was about one minute and in sham acupuncture was close to zero. The treatments were similar to those in the selected studies for this review. That is, any difference between electro- and manual or manual with sham manual acupuncture could be due to the duration and strength of stimulation, or dose. This hypothesis will need to be tested in a trial examining all three modes of stimulation in the fibromyalgia population.
Challenges of sham acupuncture design in fibromyalgia studies
Sham acupuncture controls varied amongst the studies. We conducted a subgroup analysis comparing studies using invasive sham acupuncture with studies using a non-invasive method and found no subgroup difference. This comparison is, however, influenced by the small number of studies and mixed studies using electro- and manual acupuncture. Consequently, we could not draw a strong inference as to what the ideal sham control is. Sham controls were non-invasive (Assefi 2005 (one arm) and Harris 2008; Harris 2009; Martin 2006), invasive, off the point/channel (Assefi 2005 (one arm); Deluze 1992; Harris 2005 (two arms)) plus invasive on irrelevant point (Assefi 2005 (one arm)). The inert nature of these sham methods is debatable and no agreed standard for sham controls exists for acupuncture (Birch 2006). Penetrating the skin anywhere would appear to activate one of the commonly proposed mechanisms of acupuncture analgesia, i.e. diffuse noxious inhibitory control (Lewith 1983; Pomeranz 1988). Four of the included studies reported that their choice of sham might in fact be active (Assefi 2005; Harris 2005; Martin 2006; Targino 2008). Indeed, in a trial comparing muscle blood flow in fibromyalgia participants with that of healthy controls, the researchers found that in healthy controls only deep needle insertion into an acupuncture point increased blood flow, but in fibromyalgia participants both shallow and deep insertion were equally effective (Sandberg 2004).
In a review, Lundeberg 2007 questioned whether sham acupuncture was a valid procedure for fibromyalgia participants due to their dysfunctional central nervous system (central sensitisation). As such, the nervous system may be responsive to the sub-pain threshold stimulus involved in any invasive sham acupuncture, subsequently activating the endogenous pain inhibition pathways (Mense 2003) that are usually activated by painful stimulation. Further clouding the issue is the result of a positron emission tomography (PET) trial of participants with fibromyalgia (Harris 2009). They found no difference between real manual acupuncture and non-invasive sham manual acupuncture in pain reduction. However, they identified significant group difference in brain activities. Morphine binding potential was increased in the real manual acupuncture group in the brain centres that modulated pain, whereas it was reduced or there was no change in the non-invasive sham manual acupuncture group. The results indicate that a non-invasive sham acupuncture technique may become active treatment in this population group, and its mechanism is likely due to non-opioid mediated pain modulation. This might also explain why there was no difference between deep needling with and without stimulation.
The placebo effect, including a range of components such as patient expectation, patient/therapist relationship and conditioning, has also been considered as one of the mechanisms explaining acupuncture analgesia (Finniss 2010). An analysis of data from four acupuncture trials totaling 864 participants concluded that there is a strong association between expectation and pain relief (Linde 2007). In a qualitative study, Kerr and colleagues (Kerr 2011) found that trial participants interpreted the sensation elicited by non-invasive placebo acupuncture needles as being meaningful and therapeutic. Those studies indicate that acupuncture is a complex intervention with multiple components. Indeed, some researchers challenge the usefulness of sham acupuncture controlled trials (Langevin 2011).
However, having some form of placebo is important in establishing the efficacy of a therapy. Future studies need to identify an adequate sham acupuncture intervention for fibromyalgia participants before studies are commenced. It is also important to conduct high-quality pragmatic trials to compare acupuncture with other proven therapies. In the current review, we found that acupuncture was superior to antidepressants and a combination of antidepressants and exercise for fibromyalgia, but the findings were from two studies with a small sample size.
Reporting of adverse events
We identified inconsistent reporting of adverse events in the included studies, with some studies reporting no adverse events (Harris 2005; Harris 2008) and others reporting 53% of participants experiencing them (Assefi 2005). So far, there is no uniform understanding of what constitutes an adverse event in acupuncture treatment or what should be recorded. For example, should 'pain at site of needling' be an adverse event, when for some techniques this is normal? Is being 'relaxed/tired' an adverse event or a typical indication of the therapeutic effect of acupuncture?
The discrepancy of reporting adverse events is also reflected by other published studies. For example, one clinic audit found that bleeding occurring in 53% cases and pain in 24% (White 2001), whereas another reported bleeding in 0.4% and pain in 1.2% of cases (MacPerson 2001). Although the former study audited treatments performed by medical or physiotherapy acupuncturists and the latter by traditional Chinese acupuncturists, the significant differences are likely due to reporting discrepancy and the definition of adverse events. Generally, acupuncture is considered safe. A consensus on how to report adverse events in acupuncture treatment is needed.
Implications for practice
Due to the weaknesses of the included studies, the implications for practice are limited. Overall, there is a low to moderate-quality level of evidence that formula acupuncture for the treatment of fibromyalgia is safe. There is a moderate level of evidence that acupuncture is not better than sham controls. Electro-acupuncture is found to be consistently better than sham interventions in improving pain, global well-being, sleep, stiffness and fatigue. The effect of acupuncture was not maintained at six to seven months after treatment. The same level of evidence supports acupuncture as an adjunct therapy to medication and exercise or acupuncture when compared with a medication and exercise control. When comparing acupuncture with medication or a wait list, there is low quality evidence in favour of acupuncture but this needs more rigorous and methodologically sound studies.
Evidence suggests that treatment sessions should be twice per week, over four weeks, with each session lasting for 25 minutes. Electro-acupuncture seems to provide a number of benefits for fibromyalgia participants. Practitioners should consider electro-acupuncture with 2 to 5 Hz electrical stimulation and acupuncture points could include ST36 and LI4. Optimal needling depth, point selection and needle stimulation are yet to be identified.
Like any treatment for chronic pain, maintenance acupuncture treatment is likely to be required for long-term benefit for fibromyalgia. How frequent the treatment should be is unknown.
Implications for research
We recommend a number of ways in which to address the weaknesses identified in the included studies. To further test the usefulness of acupuncture in treating fibromyalgia, researchers need to develop Chinese medicine diagnostic and subgroup differentiation criteria. The suitability of any sham acupuncture needs to be tested in this population prior to any further studies. In regards to the safety profile, a clear definition of what adverse events are associated with acupuncture is needed. Future studies testing the efficacy of acupuncture should use an adequate sample size, apply electro-acupuncture and assess the long-term results. Use of a disease-specific tool, such as the Fibromyalgia Impact Questionnaire, and accurate reporting of treatment using the Standards for Reporting Interventions in Controlled studies of Acupuncture (STRICTA) guidelines would be desirable. Future studies also need to assess how often acupuncture should be delivered to maintain its long-term benefit and the cost-effectiveness of such a treatment plan.
We would like to thank Robin Christensen at CMSG for statistical support and advice; Anette Bluemie at the German Cochrane Centre, Lorenzo Moja at the Italian Cochrane Centre and Juan Molero at the School of Health Sciences, RMIT University, Australia, for translating potential papers; and the authors of Targino 2008, Harris 2007 (excluded), Harris 2008 and Itoh 2010 for providing papers and raw data directly to us.
Data and analyses
- Top of page
- Summary of findings [Explanations]
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Differences between protocol and review
- Index terms
Appendix 1. Cochrane search strategy and results
(Updated search on 21 January 2012)
Appendix 2. MEDLINE search strategy and results
(Updated searched on 21 January 2012)
Appendix 3. EMBASE search strategy and results
(Updated search on 17 January 2012)
Appendix 4. CINAHL search strategy and results
(Via EBSCOhost updated search on 17 January 2012 then 28 March 2012)
Appendix 5. Chongqing Weipu (VIP) search strategy and results
Appendix 6. Wanfang search strategy and results
Appendix 7. National Research Register search strategy and results
(Updated search on 21 January 2012)
You searched for fibromyalgia AND acupuncture
"There are no results."
Appendix 8. HSRProj search strategy and results
(Updated search on 21 January 2012)
((fibromyalgia AND acupuncture) AND (randomised controlled trial))
0 result found
Appendix 9. Current Contents search strategy and results
(via Web of Science updated search on 17 January 2012 then 28 March 2012)
Appendix 10. STRICTA
Detailed information of acupuncture treatment (modified STRICTA items)
Abbreviations used: CM: Chinese medicine; EA: electro-acupuncture; mm: millimetre; NR: not reported; tx: treatments; wk = weeks: wkly = weekly; yrs: years
Last assessed as up-to-date: 15 January 2012.
Contributions of authors
John C Deare (JD)
- Topic conception, protocol development and revision
- Systematic review study selection, methodology, adequacy of treatment, data extraction, data analysis and interpretation of findings
- Co-author of the review
Zhen Zheng (ZZ)
- Topic conception, protocol development review and revision
- Systematic review study selection, methodology, adequacy of treatment, data extraction, data analysis and interpretation of findings
- Co-author of the review
Charlie C Xue (CX)
- Topic conception and methodological aspects
- Protocol review
- Systematic review dispute resolution for study selection, adequacy of treatment, and revision and review of final interpretations of findings
Jian Ping Liu (JPL)
- Topic conception, methodological perspectives, data analysis
- Protocol revision and review
- Systematic review dispute resolution for methodological quality and final proof
Jingsheng Shang (JSS)
- Topic conception
- Protocol review proof
- Systematic review final proof
Sean W Scott (SS)
- Protocol review proof
- Systematic review data extraction
- Systematic review final proof
Geoff Littlejohn (GL)
- Protocol review proof
- Systematic review final proof
Declarations of interest
The authors plan to conduct an acupuncture clinical trial on fibromyalgia in the future. No other potential conflicts of interest have been noted.
Sources of support
- Traditional & Complementary Medicine Program, Health Innovations Research Institute, Discipline of Chinese Medicine, School of Health Sciences, RMIT University, Australia.
- Compmed Health Institute, Southport, Australia.
- Centre for Clinical Research and Evaluation, Beijing University of Chinese Medicine, Beijing, China.
- National Center for Complementary and Alternative Medicine (NCCAM, Number R24 AT001293), National Institutes of Health (NIH), USA.JD was supported by NCCAM grant. The contents of this systematic review are solely the responsibility of the authors and do not necessarily represent the official views of the NCCAM or the NIH.
- National Health and Medical Research Council (NHMRC Project Grants 555411 and 555414), Australia.ZZ and CCLX were partially supported by the NHMRC grants. The contents of this systematic review are solely the responsibility of the authors and do not necessarily represent the official views of the NHMRC.
Differences between protocol and review
We updated the background and changed the order of importance of the main and minor outcomes, based on findings from reviews. In accordance with new Cochrane Collaboration guidelines, we have included in the review 'Risk of bias' and replaced the planned 'Clinical relevance tables' with the 'Summary of findings' tables. We did not conduct the searches of ACULARS, AcuBriefs, SIGLE or AMED as they could either not be accessed via RMIT University, it was a pay for service or the content of the databases was covered by our other searches.
Medical Subject Headings (MeSH)
MeSH check words
* Indicates the major publication for the study