Sham Electroacupuncture Methods in Randomized Controlled Trials.

Sham electroacupuncture (EA) control is commonly used to evaluate the specific effects of EA in randomized-controlled trials (RCTs). However, establishing an inert and concealable sham EA control remains methodologically challenging. Here, we aimed to systematically investigate the sham EA methods. Eight electronic databases were searched from their inception to April 2015. Ten out of the 17 sham EA methods were identified from 94 RCTs involving 6134 participants according to three aspects: needle location, depth of needle insertion and electrical stimulation. The top three most frequently used types were sham EA type A, type L and type O ordinally. Only 24 out of the 94 trials reported credibility tests in six types of sham EA methods and the results were mainly as follows: sham EA type A (10/24), type B (5/24) and type Q (5/24). Compared with sham EA controls, EA therapy in 56.2% trials reported the specific effects, of which the highest positive rate was observed in type N (3/4), type F (5/7), type D (4/6) and type M (2/3). In conclusion, several sham EA types were identified as a promising candidate for further application in RCTs. Nonetheless, more evidence for inert and concealable sham EA control methods is needed.

A randomized controlled trial (RCT) has been the cornerstone of medical clinical research since the first RCT paper entitled “Streptomycin treatment of pulmonary tuberculosis: a Medical Research Council investigation” was published in 19481. By the late 20th century, RCT was recognized as the gold standard for a clinical trial2. To improve the quality of clinical research, the methodology has been refined to avoid any bias over the past several decades. The most important design techniques for avoiding bias in clinical trials are randomization and blinding. Blinding is intended to limit the occurrence of conscious and unconscious bias in clinical trials (performance bias) conduction and interpretation of outcomes (ascertainment bias)3. Blinding is crucial for treatment evaluation because lack of blinding can bias the reliable assessment of treatment effects. For RCTs, placebo is a standard control method to blind the participants and health care providers. The purpose of placebo group is to account for the placebo effect, i.e., effects from treatment that do not depend on the treatment itself. However, blinding is difficult to ensure in non-pharmacological treatment trials because fabrication of placebo such as placebo/sham acupuncture controls requires the placebo to be both inert and indistinguishable, which is relatively difficult4.

RCTs for acupuncture appeared in 1970s5. Since then, a number of RCTs on acupuncture have been published6. The “sham” acupuncture is identified as the procedure controlling for the acupuncture treatment components with the aim to blind the participants and control for non-specific placebo effects7. Since participants are to a large extent ignorant of the components of acupuncture such as needle location, depth of needle insertion, needle stimulation and patient/practitioner interactions, sham acupuncture can be considered to be therapeutically inactive. However, it is difficult to design a standard method for sham acupuncture avoiding all therapeutically active components. Thus, the methodological difficulties in designing appropriate sham acupuncture controls for RCTs remained challenging8910.

Electroacupuncture (EA) is an extension technique based on traditional acupuncture combined with modern electrotherapy1112. Owing to its accurate, reproducible and standardized intensity and duration of stimulation with simple, verifiable electrical parameters, EA has been widely used in clinical studies and basic research into underlying mechanisms of acupuncture treatment1314. Currently, EA is being used extensively in China and elsewhere around the world. However, no systematic analyses have yet been published to describe the sham EA procedures. Thus, the objective of this study is to investigate the sham EA methods utilized in EA RCTs.

Methods

Search strategy

Eight electronic databases, including Cochrane Controlled Trials Register, PubMed, EMBASE, AMED, China National Knowledge Infrastructure (CNKI), VIP Journals Database, Wanfang database, and Chinese Biomedical Database (CBM) were searched from their inceptions to April 2015. The search terms were confined to “Electroacupuncture” AND “sham acupuncture OR placebo acupuncture” AND “randomized controlled trial (RCT)”. All searches were limited to studies on human.

Eligibility Criteria

RCTs concerning the effects of EA on any kind of diseases with at least one control group receiving sham EA were included, regardless of publication status and languages. Quasi-RCTs and non-RCTs were excluded.

The studies were eligible if EA therapy alone or adjunct therapy were given in treatment group and secondly, if sham EA or any type of faked manipulation mimicking real EA in aspects of acupoint, penetration and electro-stimulation were given in control group. There were no restrictions on needle parameters or intensity, frequency and mode of stimulation. Studies that compared EA with transcutaneous nerve electrical stimulation (TNES), another acupuncture plus sham EA or placebo medications were excluded. If three or more groups were designed in one study, only real EA versus sham EA groups were included.

Study selection and data extraction

Two authors (ZXC, YL) reviewed the titles and abstracts of the potential references independently. All the potentially relevant studies were marked and their full articles were retrieved. Further examinations were carried out to make a final selection decision. The same two authors performed the data extraction independently for the predefined items: author, year, country, EA indications, sample size, the characteristics of interventions, outcome measures, results and dropouts. The disagreements were resolved through consulting a third part (GQZ).

Risk of bias assessment

Two authors (ZXC and YL) performed the methodological quality assessment of each included trial independently based on the Cochrane Collaboration’s tool for assessing risk of bias14. The criteria consisted of the following: adequate sequence generation, allocation concealment, blinding of participants, blinding of personnel, blinding of outcome assessor, free of incomplete outcome data, free of selective reporting and free of other bias.

Description of sham EA methods

The sham EA methods used in each control group were examined and the details were extracted according to three respects: needle location, depth of needle insertion and electrical stimulation. Partially based on the previous sham acupuncture type I~V classification published by Dincer et al.8 we summarized seventeen kinds of sham EA methods: (1) Sham EA on therapeutic acupoints plus no skin penetration plus no electrical stimulation (Sham EA type A); (2) Sham EA on therapeutic acupoints plus no skin penetration plus electrical stimulation (Sham EA type B); (3) Sham EA on therapeutic acupoints plus the same depth plus no electrical stimulation (Sham EA type C); (4) Sham EA on therapeutic acupoints plus superficial insertion plus no electrical stimulation (Sham EA type D); (5) Sham EA on therapeutic acupoints plus superficial insertion plus electrical stimulation (Sham EA type E); (6) Sham EA on nonspecific acupuncture points plus the same depth plus electrical stimulation (Sham EA type F); (7) Sham EA on nonspecific acupuncture points plus the same depth plus no electrical stimulation (Sham EA type G); (8) Sham EA on nonspecific acupuncture points plus superficial insertion plus electrical stimulation (Sham EA type H); (9) Sham EA on nonspecific acupuncture points plus superficial insertion plus no electrical stimulation (Sham EA type I); (10) Sham EA on nonspecific acupuncture points plus no skin penetration plus electrical stimulation (Sham EA type J); (11) Sham EA on nonspecific acupuncture points plus no skin penetration plus no electrical stimulation (Sham EA type K); (12) Sham EA on non-acupuncture points plus the same depth plus electrical stimulation (Sham EA type L); (13) Sham EA on non-acupuncture points plus the same depth plus no electrical stimulation (Sham EA type M); (14) Sham EA on non-acupuncture points plus superficial insertion plus electrical stimulation (Sham EA type N); (15) Sham EA on non-acupuncture points plus superficial insertion plus no electrical stimulation (Sham EA type O); (16) Sham EA on non-acupuncture points plus no skin penetration plus electrical stimulation (Sham EA type P); (17) Sham EA on non-acupuncture points plus no skin penetration plus no electrical stimulation (Sham EA type Q).

Assessment of the effectiveness

Considering wildly varying outcome measures across different disease conditions, treatment efficacy was evaluated for each study according to the modified method based on a previous publication10. The results of each trial were presented by using the following primary outcome measures: “T > C” meaning that real EA treatment group was significantly superior to sham EA control group; “ND” meaning no difference between EA and sham EA groups; “T < C” meaning that real EA group was significantly inferior to sham EA group. If the efficacy of a trial was reported as “T > C” or “T < C” without between-groups comparison having been conducted, we collected the original data by reviewing the articles or contacting the corresponding author. If the original data were available, an effect-size analysis was conducted to reconfirm the between-groups difference. If the original data were not available, the efficacy results were presented as “T > C?” or “T < C?”.

The credibility of blinding

The credibility test was formally performed in validation studies to assess the blinding effect of sham acupuncture based on credibility questionnaire and statistical analysis151617. The information on the credibility test was extracted to explore the relationship between the credibility of blinding and the type of sham EA method.

Results

Study selection

A total of 679 potentially relevant articles were identified. By reviewing titles and abstracts, 374 papers were excluded for at least one of following reasons: (1) not clinical trials; (2) case report, comment, review, letter, news or editorial; (3) not in contrast with sham EA; (4) lack of EA intervention. After examining the full content of the remaining 305 articles, we removed 211 records, of which 127 articles were due to lack of sham EA controls, including electro-acupuncture (14 studies), manual acupuncture (13 studies), TNES (7 studies), other treatment (69 studies) or no treatment (24 studies); 49 articles removed for lack of real EA groups, with their target intervention designed as manual acupuncture (10 studies), TNES (38 studies) or periosteal stimulation therapy (PST) (1 study); 10 articles were not RCTs; 13 articles were double publications; 12 articles were cross-over design. Ultimately, 94 studies18192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111 involving 6134 participants were selected (Fig. 1).

Figure 1

Flow diagram.

Characteristics of included studies

The 94 included articles were published from 1992 to 2015. Among them, 5 studies29549095107 were published between 1992 and 1999; 33 studies22242527303138454752535657586768717273747778818384929397104105106110111 were published between 2000 and 2010; the remaining 56 studies1819202123262832333435363739404142434446484950515559606162636465666970757679808285868788899194969899100101102103108109 were reported from 2010 to 2015 (Fig. 2). Indications for EA included pain (32 studies)1819212426273133363740424547515256606667727682848890929697103110111, anesthesia (8 studies)46507374868994102, stroke (7 studies)2528293462105106, depression (6 studies)235359656880, obesity (4 studies)32495470, primary dysmenorrheal/menstrual pain (4 studies)619899101, substance abuse (heroin or smoking) (3 studies)6495107, osteoarthritis (2 studies)22104, migraine (2 studies)3978, nausea and vomiting (2 studies)3857, postoperative ileus (2 studies)3591, insomnia (2 studies)6381, benign prostate hyperplasia (2 studies)7987, diabetic mellitus related diseases (2 studies)83109, carpal tunnel syndrome (1 study)100, rheumatoid arthritis (1 study)93, whiplash-associated disorders (1 study)69, constipation (1 study)48, multiple sclerosis (1 study)41, tinnitus (1 study)20, auditory hallucination (1 study)30, attention deficit hyperactivity disorder (1 study)44, polycystic ovary syndrome (1 study)55, hot flushes (1 study)58, postpartum insufficient lactation (1 study)71, cardiac ischemia-reperfusion injury (1study)75, stress-related symptoms (1 study)108. The rest three studies437785 reported the effects of EA on healthy subjects.

Figure 2

The time distribution of the included articles.

EA treatment alone was adopted in 55 trials1920212223242526282930313435363738394243535558606166686971727577787980838485868788899091959899100101103104107108110111, while the interventions of the remaining 39 trials were a combination of EA and western conventional medicine (WCM). Four trials244754100 were designed as two groups of EA, and seven trials384361639899101 were conducted with two groups of sham EA. Compared with sham EA group, real EA group in 83 studies selected the same number of acupoints; nine studies2546668990105106108111 used more number of acupoints; one study20 used less number of acupoints; one study52 did not report the number of acupoints. Eight studies26495476959697110 identified acupoints by using a point detector. The “deqi” sensation was required in 65 real EA groups181920212223242526293031333435363940414344474850515657585960616263646667697175777879808182838586879091939899100101102104105106109111 and 3 sham EA groups selected nonspecific acupoints6178101. Eight studies utilized pricking sensation to mimic needle sensation and blind participants in control group2021235260636781. Forty-two studies1920212226272831323334353739404243444647505158606162646566707177787983858992100104105106 applied EA at high intensity with maximum tolerance, and seventeen studies24295456598082868790959899102107108111 applied EA with low intensity below pain threshold or at a comfortable level with presence or absence of muscle contractions. The other trials were lacking in details on the intensity of stimulation. The duration of each session ranged from 5 minutes100 to 72 hours96; the total number of treatment sessions varied from 1333738394647505152727377828486888994959697100101102109 to 7244; the total duration of treatment ranged from 5 minutes100 to 6 months111. Ten studies274672767779899097102 did not report the duration of each session. In one study76, the total number of treatment session was not mentioned. Eight studies27467276778997102 did not provide any information on the total duration of treatment. The characteristics of the studies were listed in Table 1.

Table 1

Characteristics of 94 included studies.

Reference (author, year and country)	Diseases	No. of acupoint (T/C)	Sample size (T/C)	dropout (T/C)	Primary outcome measures	Difference between groups	Characteristic of sham electro-acupuncture (EA) methods
							needle location	degree of needle insertion	electrical stimulation	The type of sham EA method
Ntritsou et al. 2014 USA18	Pain	4/4	37/38	5(2/3)	NRS and SF_MPQ scores	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Chu et al. 2012 Hong Kong19	Pain perception in Irritable Bowel Syndrome	6/6	15/15	0(0/0)	FMRI	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Wang et al. 2010 Denmark20	Tinnitus	4/14	20/20	9(4/5)	The frequency of tinnitus occurrence and the tinnitus loudness	ND	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Zyloney et al. 2010 USA21	Pain	2/2	N/N	29(N/N)	FMRI and MASS ratings	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Jubb et al. 2008 UK22	Osteoarthritis	6/6	34/34	4(2/2)	WOMAC pain score	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Chung et al. 2012 Hong Kong23	Postpartum Depression	18/18	10/10	6(5/1)	EPDS, HADS, HDRS17 and CGI scores	ND	therapeutic acupoints	no penetration	electrical stimulation	Sham EA type B
Barlas et al. 2006 UK24 1	Pain	4/4	12/12	N(N/N)	Pressure pain threshold	T > C	therapeutic acupoints	no penetration	electrical stimulation	Sham EA type B
Barlas et al. 2006 UK24 2	Pain	4/4	12/12	N(N/N)	Pressure pain threshold	ND	therapeutic acupoints	no penetration	electrical stimulation	Sham EA type B
Wayne et al. 2005 US25	Stroke Rehabilitation	14–22/2–3	16/17	9(3/6)	FMA score	ND	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Sahin et al. 2010 Turkey26	Pain	14/14	15/16	2(2/0)	VAS scores	ND	non-acupuncture points	the same depth	electrical stimulation	Sham EA type L
Fanti et al. 2003 Italy27	Pain	10/10	10/10	0(0/0)	VAS scores	ND	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Hsing et al. 2012 Brazil28	Stroke recovery	11–14/11–14	31/31	0(0/0)	NIHSS, Barthel Index and modified Rankin scales scores	T > C	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Gosman-Hedström et al. 1998 Sweden29	Stroke	4/4	37/34	3(2/1)	The neurological score and the Barthel and Sunnaas index scores	ND	therapeutic acupoints	superficial penetration	no electrical stimulation	Sham EA type D
Jing et al. 2009 China30	Auditory hallucination	6/6	30/30	7(4/3)	The Psychotic Symptom Rating Scales and Auditory Hallucination Subscale total score	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Kong et al. 2009 USA31	Pain	2/2	N/N	29(N/N)	Gracely Sensory and Affective scales scores and fMRI	ND	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Darbandi et al. 2013 Iran32	Obesity	4/4	47/47	8(5/3)	BW, BMI and BFM	ND	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Yang et al. 2012 China33	Pain	12/12	40/40	0(0/0)	the consumption of sevoflurane and the recovery profile	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Liu et al. 2013 China34	Post-stroke detrusor overactivity	10/10	35/36	5(2/3)	maximum cystometric capacity and bladder compliance	T > C	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Zhang et al. 2014 China35	Postoperative ileus	2/2	20/20	1(1/0)	Time of the first bowel sounds and passage of flatus	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Mao et al. 2014 USA36	Aromatase Inhibitor-Related Arthralgia	4/4	22/22	6(3/3)	BPI	ND	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Leung et al. 2011 Hong Kong37	Pain	3/3	20/20	0(0/0)	the maximal tolerable pressure, VAS score and beta-endorphin level	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Rusy et al. 2002 USA38 a	Postoperative Nausea and Vomiting	1/1	40/40	0(0/0)	occurrence of nausea and vomiting, and use of antiemetic rescue medication	T > C	nonspecific acupuncture points	the same depth	electrical stimulation	Sham EA type F
Rusy et al. 2002 USA38 b	Postoperative Nausea and Vomiting	1/1	40/40	0(0/0)	occurrence of nausea and vomiting, and use of antiemetic rescue medication	T > C	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Yang et al. 2014 China39	Migraine	3/3	10/10	0(0/0)	VA S scores	ND	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Chen et al. 2013 China40	Pancreatic cancer pain	10/10	30/30	1(0/1)	NRS	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Quispe-Cabanillas et al. 2012 Brazi41	Multiple sclerosis	9/9	16/15	0(0/0)	EDSS, pain VASscore and quality of life FAMS	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Aranha et al. 2015 Brazil42	Pain	7–8/7–8	24/23	17(7/10)	VAS scores and cervical movements	T > C	non-acupuncture points	the same depth	no electrical stimulation	Sham EA type M
Yu et al. 2013 Hong Kong43 a		2/2	12/12	0(0/0)	C-MMASS and HR, MAP	ND	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Yu et al. 2013 Hong Kong43 b		2/2	12/12	0(0/0)	C-MMASS and HR, MAP	T > C	therapeutic acupoints	superficial penetration	no electrical stimulation	Sham EA type D
Li et al. 2010 Chin44	Attention deficit hyperactivity disorder	15–16/15–16	92/88	10(6/4)	relapse rate	T > C	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Zheng et al. 2007 Australia45	Chronic pain	4/4	17/18	12(8/4)	the dosage reduction of OLM, the incidence of side effect, and VAS score	ND	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Li et al. 2013 China46	General anesthesia	10/6	9/10	0(0/0)	the levels of TNF-α, IL-8 and IL-10	ND	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Lin et al. 2002 Taiwan47 1	Postoperative pain	2/2	25/25	0(0/0)	VAS score	ND	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Lin et al. 2002 Taiwan47 2	Postoperative pain	2/2	25/25	0(0/0)	VAS score	ND	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Chen et al. 2013 Taiwan48	Constipation	6/6	30/30	30 (16/14)	the defecation rate	T > C	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Schukro et al. 2014 Austria49	Obesity	3/3	28/28	14(7/7)	the relative reduction of body weight	T > C	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Yu et al. 2014 China50	General anesthesia	2/2	20/20	0(0/0)	The serum cortisol and ACTH	T > C ?	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Xie et al. 2014 China51	Postoperative pain	2/2	20/20	0(0/0)	VAS score, Total Doses of Sufentanil and Dezocine	T > C	therapeutic acupoints	the same depth	no electrical stimulation	Sham EA type C
Sim et al. 2002 Singapore52	Intraoperative pain	N/N	30/30	0(0/0)	The total intraoperative usage of alfentanil, The total morphine consumption and VAS score	ND	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Song et al. 2009 China53	Depression	2/2	31/32	10(3/7)	HDRS and CGI	T > C	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Shafshak 1995 Egypt54 1	Obesity	2/2	10/10	N(N/N)	the success rate of going on the diet trigmen	T > C	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Shafshak 1995 Egypt54 2	Obesity	2/2	10/10	N(N/N)	the success rate of going on the diet trigmen	T > C	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Franasiak et al. 2012 USA55	Polycystic Ovary Syndrome	8/8	46/50	16(9/7)	Serum LH and FSH.The monthly rates of ovulation	ND	non-acupuncture points	no penetration	no electrical stimulation	Sham EA type Q
Naslund et al. 2002 Sweden56	Idiopathic anterior knee pain	6/6	30/28	1(0/1)	one leg vertical jump, functional score, daily VAS recording and skin temperature	ND	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Shen et al. 2000 US57	Chemotherapy–Induced Emesis	4/4	37/33	1(1/1)	Total number of emesis episodes occurring	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Wyon et al. 2004 Sweden58	Hot flushes in postmenopausal women.	6/6	15/15	7(4/3)	The number of flushes/24 h	ND	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Zhang et al. 2012 Hong Kong59	Depression	12/12	38/35	10(7/3)	score of HAMD-17 and SDS	T > C	therapeutic acupoints	no penetration	electrical stimulation	Sham EA type B
Zheng et al. 2010 Australia60	Pain	2/2	12/12	0(0/0)	SPT and TST	T > C	therapeutic acupoints	no penetration	no electrical stimulation	Sham EA type A
Ma et al. 2010 China61 a	Menstrual Pain	2/2	13/14	1(1/0)	VAS scores	T > C	nonspecific acupuncture points	the same depth	electrical stimulation	Sham EA type F
Ma et al. 2010 China61 b	Menstrual Pain	2/2	13/12	1(1/0)	VA S scores	T > C	non-acupuncture points	the same depth	Electrical stimulation	Sham EA type L
Wang et al. 2014 Taiwan62	Chronic stroke	4/4	10/10	5(1/4)	R1, R2 and R2–R1	T > C	therapeutic acupoint	superficial penetration	No electrical stimulation	Sham EA type D
Yeung et al. 2011 Hong Kong63 a	Insomnia	8/8	26/26	7(4/3)	ISI and PSQI	ND	non-acupuncture points	superficial penetration	electrical stimulation	Sham EA type N
Yeung et al. 2011 Hong Kong63 b	Insomnia	8/8	26/26	7(4/3)	ISI and PSQI	T > C	non-acupuncture points	No penetration	No electrical stimulation	Sham EA type Q
Chan et al. 2014 Taiwan64	Heroin Addicts	4/4	30/30	2(1/1)	the daily consumption of methadone	T > C	therapeutic acupoint	superficial penetration	no electrical stimulation	Sham EA type D
Man et al. 2014 Hong Kong65	Post-stroke depression	20/20	23/20	10(4/6)	HAMD-17 and CGI-S	T > C	therapeutic acupoint	No penetration	electrical stimulation	Sham EA type B
Oh et al. 2013 Australia66	Pain	16/12	16/16	3(2/1)	WOMAC, BPI-SF and FACT-G instrument	ND	therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Wong et al. 2006 Hong Kong67	Pain	4/4	13/14	2(0/2)	VAS pain score	ND	therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Song et al. 2007 USA68	Depression	2/2	N/N	N(N/N)	24-item HAMD and the level of G protein α subtypes in the platelet membrane	ND	non-acupuncture points	the same depth	no electrical stimulation	Sham EA type M
Cameron et al. 2011 Australia69	Whiplash-associated Disorders	8/8	64/60	8(0/8)	VAS	T > C	non-acupuncture points	the same depth	no electrical stimulation	Sham EA type M
Darbandi et al. 2014 Iran70	Obesity	4/4	20/20	0(0/0)	BMI, TFM, WC and HC	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Wei et al. 2008 China71	Postpartum Insufficient Lactation	2/2	46/46	0(0/0)	total therapeutic effect, 24-hour milk secretion quantity, prolactin level	T > C	nonspecific acupuncture points	the same depth	electrical stimulation	Sham EA type F
Wang et al. 2007 USA72	Pain	4/4	29/27	0(0/0)	intraprocedural alfentanil consumption, VAS score	T > C	therapeutic acupoint	superficial penetration	no electrical stimulation	Sham EA type D
Kvorning et al. 2003 Sweden73	Anaesthesia	12/12	23/23	1(1/0)	Physiological reactions to skin incision	T > C	therapeutic acupoint	No penetration	no electrical stimulation	Sham EA type A
Kvorning et al. 2003 Sweden74	Anaesthesia	6/6	23/23	0(0/0)	The minimal alveolar concentration of sevoflurane	T < C	therapeutic acupoint	No penetration	no electrical stimulation	Sham EA type A
Yang et al. 2010 China75	Cardiac ischemia-reperfusion injury	6/6	30/30	0(0/0)	levels of serum cardiac troponin I	T > C	therapeutic acupoint	the same depth	no electrical stimulation	Sham EA type C
Sahmeddini et al. 2010 Iran76	Perioperative Pain	4/4	45/45	0(0/0)	score on VAS-100	ND	therapeutic acupoint	no penetration	no electrical stimulation	Sham EA type A
Wang et al. 2007 China77		1/1	9/5	3(3/0)	BOLD fMRI	T > C	non-acupuncture points	the same depth	electrical stimulation	Sham EA type L
Jia et al. 2009 China78	Migraine	2/2	138/138	1(0/1)	VAS score and the plasma 5-HT level	T > C	nonspecific acupuncture points	the same depth	electrical stimulation	Sham EA type F
Wang et al. 2013 China79	Benign prostate hyperplasia	2/2	50/50	23(9/14)	IPSS	T > C	non-acupuncture points	the same depth	electrical stimulation	Sham EA type L
Andreescu et al. 2011 Canada80	Depression	2/2	28/29	11(4/7)	HDRS score	ND	non-acupuncture points	superficial penetration	No electrical stimulation	Sham EA type O
Yeung et al. 2009 Hong Kong81	Insomnia	8/8	30/30	3(1/2)	ISI	ND	therapeutic acupoint	no penetration	no electrical stimulation	Sham EA type A
Chen et al. 2014 Taiwan82	Pain	1/1	25/24	0(0/0)	VAS scores and the dosage of opium derivative analgesic	ND	therapeutic acupoint	the same depth	no electrical stimulation	Sham EA type C
Wang et al. 2008 China83	Diabetic Gastroparesis	4/4	11/12	4(2/2)	GCSI score	T > C	non-acupuncture points	superficial penetration	electrical stimulation	Sham EA type N
Meissner et al. 2004 Germany84	Pain	6/6	8/8	0(0/0)	SEPs	T > C	therapeutic acupoint	no penetration	electrical stimulation	Sham EA type B
Zhou et al. 2012 China85		2/2	11/11	0(0/0)	MVC strength	ND	non-acupuncture points	the same depth	electrical stimulation	Sham EA type L
Yeh et al. 2012 Taiwan86	Shivering during regional anesthesia	4/4	40/40	0(0/0)	Shivering score and tympanic temperature	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Yu et al. 2011 Taiwan87	Benign Prostate Hyperplasia	6/6	21/21	5(3/2)	The change of the maximum flow rate, average flow rate, void volume	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Ma et al. 2011 China88	Pain	2/2	116/117	0(0/0)	VAS score	T > C	therapeutic acupoint	no penetration	no electrical stimulation	Sham EA type B
Li et al. 2013 China89	Intraoperative immunosuppression	10/6	19/19	0(0/0)	the levels of TNFα, IL-8, IL-10, IgM, IgA, IgG and full blood count	ND	non-acupuncture points	the same depth	electrical stimulation	Sham EA type L
Deluze et al. 1992 Switzerland90	Fibromyalgia	4-10/4	36/34	15(8/7)	Pain threshold, number of analgesic tablets used, VAS score	T > C	non-acupuncture points	superficial penetration	electrical stimulation	Sham EA type N
Ng et al. 2013 Hong Kong91	Postoperative ileus	4/4	55/55	0(0/0)	Time to defecation	T > C	non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Lee and Lee 2009 Republic of Korea92	Chronic Prostatitis / Chronic Pelvic Pain Syndrome	6/6	13/13	5(2/3)	NIH-CPSI	T > C	Non-acupuncture points	superficial penetration	no electrical stimulation	Sham EA type O
Tam et al. 2007 Hong Kong93	Rheumatoid arthritis	6/6	12/12	5(0/5)	VAS	ND	Therapeutic acupoint	Superficial penetration	No electrical stimulation	Sham EA type D
Kvorning and Akeson 2010 Sweden94	Anaesthesia	12/12	22/23	0(0/0)	Plasma levels of adrenaline	T > C	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Waite and Clough 1998 UK95	Smoking cessation	2/2	40/38	0(0/0)	Biochemically validated total cessation of smoking	T > C	Non-acupuncture points	superficial penetration	Electrical stimulation	Sham EA type N
Holzer et al. 2011 Austria96	Postoperative pain	3/3	20/20	0(0/0)	VAS scores and the consumption of piritramide	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Sator-Katzenschlager et al. 2006 Austria97	Perioperative pain	3/3	32/30	1(0/1)	VAS scores, adverse event and analgesic drug consumption	T > C	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Liu et al. 2011 China98 a	Primary dysmenorrhea	1/1	50/50	5(3/2)	VAS scores	ND	Nonspecific acupuncture points	The same depth	Electrical stimulation	Sham EA type F
Liu et al. 2011 China98 b	Primary dysmenorrhea	1/1	50/46	6(3/3)	VAS scores	ND	Non-acupuncture points	The same depth	Electrical stimulation	Sham EA type L
Liu et al. 2014 China99 a	Primary dysmenorrhea	2/2	167/167	6(4/2)	VAS scores	T > C	Nonspecific acupuncture points	The same depth	Electrical stimulation	Sham EA type F
Liu et al. 2014 China99 b	Primary dysmenorrhea	2/2	167/167	6(4/2)	VAS scores	T > C	Non-acupuncture points	The same depth	Electrical stimulation	Sham EA type L
Maeda et al. 2013 USA100 1	Carpal tunnel syndrome	2/2	22/19	0(0/0)	functional MRI, VAS scores	T > C	Non-acupuncture points	No penetration	No electrical stimulation	Sham EA type F
Maeda et al. 2013 USA100 2	Carpal tunnel syndrome	2/2	18/19	0(0/0)	functional MRI, VAS scores	T > C	Non-acupuncture points	No penetration	No electrical stimulation	Sham EA type F
Shi et al. 2011 China101 a	Primary dysmenorrhea	1/1	10/10	0(0/0)	VAS scores, The plasma PGE2, PGF2a, TXB2, and 6-keto PGF1a levels	ND	Nonspecific acupuncture points	The same depth	Electrical stimulation	Sham EA type F
Shi et al. 2011 China101 b	Primary dysmenorrhea	1/1	10/10	0(0/0)	VAS scores, The plasma PGE2, PGF2a, TXB2, and 6-keto PGF1a levels	ND	Non-acupuncture points	The same depth	Electrical stimulation	Sham EA type L
Dias et al. 2010 Brazil102	Local anaesthesia	8/8	16/17	0(0/0)	VAS scores	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Zhang et al. 2013 Hong Kong103	Chronic neck Pain	5/5	103/103	46(19/27)	NPQ scores	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Sangdee et al. 2002 Thailand104	Osteoarthritis of the knee	4/4	48/47	4(2/2)	VAS score, and Lequesne’s functional index	T > C	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Johansson et al. 2001 Sweden105	Stroke rehabilitation	9–10/4	48/51	20(11/9)	The scores of Barthel Index, the Rivermead Mobility Index and NHP, the time needed to walk 10 meters	ND	Therapeutic acupoint	No penetration	Electrical stimulation	Sham EA type B
Hopwood et al. 2008 UK106	Stroke recovery	8–10/6	57/48	13(10/3)	The scores of Barthel Index	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
White et al. 1998 England107	Tobacco addiction	2/2	38/38	24 (11/13)	VAS score	ND	Non-acupuncture points	No penetration	No electrical stimulation	Sham EA type Q
Dias et al. 2014 Brazil108	Stress-related symptoms	13/8	33/20	5(3/2)	The scores of MSQ, PSQI and MBI-SS	T > C	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Lin et al. 2013Taiwan109	Insulin Resistance	2/2	16/15	1(1/0)	Plasma glucose	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Michalek-Sauberer et al. 2007 Austria110	Perioperative pain	3/3	76/36	24(16/8)	5-point verbal rating scale, Time and amount of analgesic intake	ND	Therapeutic acupoint	No penetration	No electrical stimulation	Sham EA type A
Carlsson and Sjolund 2001 Sweden111	Chronic Low Back Pain	4/2	16/16	0(0/0)	VAS scores, Intake of analgesics, Sleep quality and level of activity	T > C?	Non-acupuncture points	No penetration	No electrical stimulation	Sham EA type Q

T, treatment group/real EA group; C, control group/sham EA group; NS, not stated; T > C, EA treatment group was significantly superior to sham EA control group; ND, no difference between EA and sham EA group; T < C, real EA group was significantly inferior to sham EA group; T > C?, the efficacy result of trial was reported as “T > C” without conducting the between-group analysis and with the original data not available; NRS, Numerical Rating Scale; SF_MPQ, Short-Form McGill Scale; FMRI, Functional magnetic resonance imaging; MASS, the Massachusetts General Hospital Acupuncture Sensation Scale; WOMAC, The Western Ontario and McMaster University Osteoarthritis Index; EPDS, Edinburgh Postnatal Depression Scale; HADS, Hospital Anxiety and Depression Scale; HDRS17, 17-item Hamilton Rating Scale for Depression; CGI, Clinical Global Impression; FMA, Fugl-Meyer Assessment; VAS, Visual Analog Scale; NIHSS, the National Institutes of Health Stroke Scale. BW, body weight; BMI, body mass index; BFM, body fat mass; BPI, Brief Pain Inventory; EDSS, Expanded Disability Status Scale; FAMS, Functional Assessment of multiple Sclerosis; C-MMASS, Modified Massachusetts General Hospital Acupuncture Sensation Scale – Chinese version; HR, Heart rate; MAP, mean arterial blood pressure; OLM, opioid-like medication; TNF-α, tumor necrosis factor-α; IL-8, interleukin-8; IL-10, interleukin-10; HDRS, Hamilton Depression Rating Scale; HAMD-17, the 17-item Hamilton Rating Scale for Depression; SDS, the Chinese-version Self-rating Depression Scale; SPT, single pain threshold; TST, temporal summation thresholds; R1, angle of muscle reaction; R2, passive range of motion; R2–R1, dynamic component; ISI, The Insomnia Severity Index; PSQI, the Pittsburgh Sleep Quality Index; CGI-S, the Clinical Global Impression - Severity scale; BPI-SF, Brief Pain Inventory Short Form; FACT-G instrument, the Functional Assessment of Cancer Therapy-General instrument; 24-item HAMD, the 24-item Hamilton Depression Rating Scale; TFM, Trunk Fat Mass; WC, Waist Circumference; HC, Hip Circumference; VAS-100, a 100-mm visual analogue scale; BOLD fMRI, blood oxygen level dependent functional magnetic resonance imaging; IPSS, the International Prostate Symptom Score; GCSI, the Gastroparesis Cardinal Symptom Index; SEPs, somatosensory evoked potentials; MVC, maximal voluntary contraction; NIH-CPSI, NIH-Chronic Prostatitis Symptom Index; NPQ, the Northwick Park Neck Pain Questionnaire; NHP, the Nottingham Health Profile; MSQ, Mini-Sleep Questionnaire; MBI-SS, the Maslach Burnout Inventory—Student Survey.

Note: Sham EA type A: sham EA on therapeutic acupoints plus no penetration plus no electrical stimulation; Sham EA type B: sham EA on therapeutic acupoints plus no penetration plus electrical stimulation; Sham EA type C: sham EA on therapeutic acupoints plus the same depth plus no electrical stimulation; Sham EA type D: sham EA on therapeutic acupoints plus superficial penetration plus no electrical stimulation; Sham EA type F: sham EA on nonspecific acupuncture points plus the same depth plus electrical stimulation; Sham EA type L: sham EA on non-acupuncture points plus the same depth plus electrical stimulation; Sham EA type M: sham EA on non-acupuncture points plus the same depth plus no electrical stimulation; Sham EA type N: sham EA on non-acupuncture points plus superficial penetration plus electrical stimulation; Sham EA type O: sham EA on non-acupuncture points plus superficial penetration plus no electrical stimulation; Sham EA type Q: sham EA on non-acupuncture points plus no penetration plus no electrical stimulation.

Characteristics of sham EA

Ten different types of sham EA methods used in the trials were identified as follows: (1) sham EA type A were used in twenty-six control groups18192021223133374060666773747681949697102103104106108109110; (2) sham EA type B were used in seven control groups232459658488105; (3) sham EA type C were used in seven control groups28444749517582; (4) sham EA type D were used in six control groups294362647293; (5) sham EA type F were used in seven control groups386171789899101; (6) sham EA type L were used in seventeen control groups26273943464850535461777985899899101; (7) sham EA type M were used in three control groups426869; (8) sham EA type N were used in four control groups63839095; (9) sham EA type O were used in fourteen control groups3032354145565758708086879192; and (10) sham EA type Q were used in ten control groups25343638525563100107111.

For the needle location, 48 sham EA groups25262730323435363839414243454648505253545556575861636869707779808385868789909192959899100101107111 chose non-acupoints that were either located away from the therapeutic acupoints with a distance ranging from 1 cm to 40 cm or devised in advance to avoid any known meridian or extra-point. Seven sham EA groups386171789899101 received nonspecific acupuncture points which were thought to be ineffective for treating the diseases. For the depth of needle insertion, 34 sham EA groups2627283839424344464748495051535461686971757778798285899899101 conducted the needle insertion to the same depth as corresponding real EA groups. Twenty-four sham EA groups293032354143455657586263647072808386879091929395 performed either superficial or subcutaneous needle insertion with depth varying from 0.5 mm to 2 cm, whereas one study93 retracted the needle after superficial penetration. The remaining 43 sham EA groups had sham EA without skin penetration. Forty-one out of 43 trials did not apply any needle insertion by using Streitberger needles2123253136555963656681, contractible placebo needles with dull tips and tubes243752, placebo needles with blunted tips1920406067100, verum needles fixed by tapes or rings without piercing182234, leading wires alone without needles33387697 and mock laser pen103 or electrodes73749496102104105106107108109110111. Two sham EA groups8488 did not describe any details on the sham needles. For electrical stimulation, 35 sham EA groups2324262738394346485053545961636571777879838485888990959899101105 used electrical stimulation, whereas two sham EA groups stimulated with current just at the beginning of sham procedure. The other sham EA groups1819202122252829303132333435363738404142434445474951525556575860626364666768697072737475768081828687919293949697100102103104106107108109110111 did not receive any electrical stimulation through inactivated EA device or disconnected cables. The details of sham EA were described in Table S1.

The number of items complied with the criteria varied from 3/8 to 7/8 with the average of 5.2. All 94 studies declared randomization and 63 studies reported the details. Among them, 49 studies181922232425272935363738424344454749515559606162636465666770717281889091929396979899102106107108109110111 described a computer-generated randomization; 11 studies3240415052576973748586 were based on random number Table; 3 studies8387101 used the lot. Adequate allocation concealment was found in 43 studies18192023252930353637434445555758606163646566676971727679818486889091939899105106107109110111 with sequentially numbered, opaque, sealed envelopes or independent administrator. The remaining 51 studies did not provide the details on allocation concealment. Blinding of participant was described in all 94 studies. Among these, 23 studies22232534364552575960636572818897102103105106107108110 proved their success of blinding by credibility test, while one study66 failed in blinding of participant after testing by statistical analysis. No study mentioned blinding of acupuncturists. Ninety-two out of the 94 studies reported blinding of assessor, whereas one study54 did not contain any information on assessor blinding and another study66 was sorted as “no” due to its unsuccessful assessor blinding. Eighteen studies2530456065697980919293979899103105106107 conducted intention-to-treat analysis. Seventy-five studies18192021222425262728293031323334353637383940414344464749505152535455565758596061626364656667687071727374757677787981828384858687888990919294959697100101102103104105108109110111 were free of incomplete outcome data; eleven studies234245486980939899106107 assessed as “no” due to high dropout rate or statistically significant differences between groups in withdrawals from the treatment. The rest were unclear due to lack of information on this aspect. Sixteen studies36394142445963647579808691939899 were free of selective reporting; one study65 was sorted as “no” due to an incomplete outcome measurement report that had been registered in protocol; the others were unclear because such details were not found. Of the 91 trials that provided the information on other bias, 77 studies1819202122232425262728293031323334353637383940414243444546474849505152535455566061626364656667687072737475767980838485868788899192939495969799101103104105109111 were free of other bias; 14 studies575859697881829098102106107108110 were assessed as “no” due to the statistical differences in baseline variables regarding as the most important prognosis. The details on the risk of bias studies were summarized in Table 2.

Table 3

Summary of effect result within different type of sham electro-acupuncture methods and electro-acupuncture indications.

electro-acupuncture (EA) indications	The type of sham EA method										The NO. of reference included
	Sham EA type A 26 control groups	Sham EA type B 7 control groups	Sham EA type C 7 control groups	Sham EA type D 6 control groups	Sham EA type F 7 control groups	Sham EA type L 17 control groups	Sham EA type M 3 control groups	Sham EA type N 4 control groups	Sham EA type O 14 control groups	Sham EA type Q 10 control groups
Pain 32 RCTs	T > C 8 comparisons ND 7 comparisons	T > C 3 comparisons ND 1 comparison	T > C 1 comparison ND 3 comparisons	T > C 1 comparison		ND 2 comparisons	T > C 1 comparison	T > C 1 comparison	T > C 1 comparison ND 2 comparisons	ND 2 comparisons T > C? 1 comparison	T > C 16 ND 17 T > C? 1
Obesity 4 RCTs			T > C 1 comparison			T > C 2 comparisons			T > C 1 comparison ND 1 comparison		T > C 4 ND 1
Anesthesia 8 RCTs	T > C 2 comparisons ND 1 comparison T < C 1 comparison					ND 2 comparisons T > C? 1 comparison			T > C 1 comparison		T > C 3 ND 3 T < C 1 T > C? 1
Stroke 7 RCTs	ND 1 comparison	ND 1 comparison	T > C 1 comparison	T > C 1 comparison ND 1 comparison						T > C 1 comparison ND 1 comparison	T > C 3 ND 4
Depression 6 RCTs		T > C 2 comparisons ND 1 comparison				T > C 1 comparison	ND 1 comparison		ND 1 comparison		T > C 3 ND 3
Primary dysmenorrhea and (or) Menstrual Pain 4 RCTs					T > C 2 comparisons ND 2 comparisons	T > C 2 comparisons ND 2 comparisons					T > C 4 ND 4
Substance abuse 3 RCTs				T > C 1 comparison				T > C 1 comparison		ND 1 comparison	T > C 2 ND 1
Healthy 3 RCTs				T > C 1 comparison		T > C 1 comparison ND 2 comparisons					T > C 2 ND 2
Osteoarthritis 2 RCTs	T > C 2 comparisons										T > C 2
Migraine 2 RCTs					T > C 1 comparison	ND 1 comparison					T > C 1 ND 1
Nausea and Vomiting 2 RCTs					T > C 1 comparison				T > C 1 comparison	T > C 1 comparison	T > C 3
Postoperative ileus 2 RCTs									T > C 2 comparisons		T > C 2
Insomnia 2 RCTs	ND 1 comparison							ND 1 comparison		T > C 1 comparison	T > C 1 ND 2
benign prostate hyperplasia 2 RCTs						T > C 1 comparison			T > C 1 comparison		T > C 2
Diabetic mellitus 2 RCTs	ND 1 comparison							T > C 1 comparison			T > C 1 ND 1
Carpal tunnel syndrome 1 RCTs										T > C 2 comparisons	T > C 2
Rheumatoid arthritis 1 RCTs				ND 1 comparison							ND 1
Whiplash-associated disorders 1 RCTs							T > C 1 comparison				T > C 1
Constipation 1 RCTs						T > C 1 comparison					T > C 1
Multiple sclerosis 1 RCTs									T > C 1 comparison		T > C 1
Tinnitus 1 RCTs	ND 1 comparison										ND 1
Auditory hallucination 1 RCTs									T > C 1 comparison		T > C 1
ADHD (Attention deficit hyperactivity disorder) 1 RCTs			T > C 1 comparison								T > C 1
PCOS (Polycystic Ovary Syndrome) 1 RCTs										ND 1 comparison	ND 1
hot flushes in postmenopausal women 1 RCTs									ND 1 comparison		ND 1
Postpartum Insufficient Lactation 1 RCTs					T > C 1 comparison						T > C 1
Cardiac ischemia-reperfusion injury 1 RCTs			T > C 1 comparison								T > C 1
Stress-related symptoms 1 RCTs	T > C 1 comparison										T > C 1
The positive rate of efficacy result	T > C 13 ND 12 T < C 1 T > C? 0 53.8% (14/26 comparisons)	T > C 5 ND 3 62.5% (5/8 comparisons)	T > C 5 ND 3 62.5% (5/8 comparisons)	T > C 4 ND 2 66.7% (4/6 comparisons)	T > C 5 ND 2 71.4% (5/7 comparisons)	T > C 8 ND 9 T > C? 1 44.4% (8/18 comparisons)	T > C 2 ND 1 66.7% (2/3 comparisons)	T > C 3 ND 1 75% (3/4 comparisons)	T > C 9 ND 5 64.3% (9/14 comparisons)	T > C 5 ND 5 T > C? 1 45.5% (5/11 comparisons)	T > C 59 ND 43 T < C 1 T > C? 2 57.1% (60/105 comparisons)

NOTE: T > C, EA treatment group was significantly superior to sham EA control group; ND, no difference between EA and sham EA group; T < C, real EA group was significantly inferior to sham EA group; T > C?, the efficacy result of trial was reported as “T > C” without conducting the between-group analysis and with the original data not available.

Table 2

Risk of bias of included studies.

Reference (author, year and country)	A	B	C	D	E	F	G	H
Ntritsou et al. 2014 USA18	+	+	+	−	+	+	?	+
Chu et al. 2012 Hong Kong19	+	+	+	−	+	+	?	+
Wang et al. 2010 Denmark20	+	+	+	−	+	+	?	+
Zyloney et al. 2010 USA21	+	−	+	−	+	?	?	+
Jubb et al. 2008 UK22	+	−	+	−	+	+	?	+
Chung et al. 2012 Hong Kong23	+	+	+	−	+	−	?	+
Barlas et al. 2006 UK24	+	−	+	−	+	?	?	+
Wayne et al. 2005 US25	+	+	+	−	+	+	?	+
Sahin et al. 2010 Turkey26	+	+	+	−	+	+	?	+
Fanti et al. 2003 Italy27	+	−	+	−	+	+	?	+
Hsing et al. 2012 Brazil28	+	−	+	−	+	+	?	+
Gosman-Hedström et al. 1998 Sweden29	+	+	+	−	+	+	?	+
Jing et al. 2009 China30	+	+	+	−	+	+	?	+
Kong et al. 2009 USA31	+	−	+	−	+	?	?	+
Darbandi et al. 2013 Iran32	+	?	+	−	+	?	?	+
Yang et al. 2012 China33	+	?	+	−	+	+	?	+
Liu et al. 2013 China34	+	+	+	−	+	+	?	+
Zhang et al. 2014 China35	+	+	+	−	+	+	?	+
Mao et al.2014 USA36	+	+	+	−	+	+	+	+
Leung et al. 2011 Hong Kong37	+	+	+	−	+	+	?	+
Rusy et al. 2002 USA38	+	?	+	−	+	+	?	+
Yang et al. 2014 China39	+	?	+	−	+	+	+	+
Chen et al. 2013 China40	+	?	+	−	+	+	?	+
Quispe-Cabanillas et al. 2012 Brazil41	+	?	+	−	+	+	+	+
Aranha et al. 2015 Brazil42	+	?	+	−	+	−	+	+
Yu et al. 2013 Hong Kong43	+	+	+	?	+	+	?	+
Li et al. 2010 China44	+	+	+	−	+	+	+	+
Zheng et al. 2007 Australia45	+	+	+	−	+	−	?	+
Li et al. 2013 China46	+	?	+	?	+	+	?	+
Lin et al. 2002 Taiwan47	+	?	+	?	+	+	?	+
Chen et al. 2013 Taiwan48	+	?	+	−	+	−	?	+
Schukro et al. 2014 Austria49	+	?	+	?	+	+	?	+
Yu et al. 2014 China50	+	?	+	−	+	+	?	+
Xie et al. 2014 China51	+	?	+	?	+	+	?	+
Sim et al. 2002 Singapore52	+	?	+	−	+	+	?	+
Song et al. 2009 China53	+	?	+	?	+	?	?	+
Shafshak 1995 Egypt54	+	?	+	?	?	?	?	+
Franasiak et al. 2012 USA55	+	+	+	−	+	+	?	+
Naslund et al. 2002 Sweden56	+	?	+	−	+	?	?	+
Shen et al. 2000 US57	+	+	+	−	+	+	?	−
Wyon et al. 2004 Sweden58	+	+	+	−	+	+	?	−
Zhang et al. 2012 Hong Kong59	+	?	+	−	+	+	+	−
Zheng et al. 2010 Australia60	+	+	+	−	+	+	?	+
Ma et al. 2010 China61	+	+	+	−	+	+	?	+
Wang et al. 2014 Taiwan62	+	?	+	−	+	+	?	+
Yeung et al. 2011 Hong Kong63	+	+	+	−	+	+	+	+
Chan et al. 2014 Taiwan64	+	+	+	−	+	+	+	+
Man et al. 2014 Hong Kong65	+	+	+	−	+	+	−	+
Oh et al. 2013 Australia66	+	+	−	−	−	+	?	+
Wong et al. 2006 Hong Kong67	+	+	+	−	+	+	?	+
Song et al. 2007 USA68	+	?	+	?	+	?	?	+
Cameron et al. 2011 Australia69	+	+	+	−	+	−	?	−
Darbandi et al. 2014 Iran70	+	?	+	−	+	+	?	+
Wei et al. 2008 China71	+	+	+	?	?	+	?	?
Wang et al. 2007 USA72	+	+	+	−	+	+	?	+
Kvorning et al. 2003 Sweden73	+	?	+	−	+	+	?	+
Kvorning et al. 2003 Sweden74	+	?	+	−	+	+	?	+
Yang et al.2010 China75	+	?	+	−	+	+	+	+
Sahmeddini et al.2010 Iran76	+	+	+	?	+	+	?	+
Wang et al. 2007 China77	+	?	+	−	+	+	?	?
Jia et al. 2009 China78	+	?	+	?	+	+	?	−
Wang et al. 2013 China79	+	+	+	−	+	+	+	+
Andreescu et al. 2011 Canada80	+	?	+	−	+	+	+	+
Yeung et al. 2009 Hong Kong81	+	+	+	−	+	+	?	−
Chen et al. 2014 Taiwan82	+	?	+	?	+	+	?	+
Wang et al. 2008 China83	+	?	+	−	+	+	?	+
Meissner et al. 2004 Germany84	+	+	+	−	+	+	?	+
Zhou et al. 2012 China85	+	?	+	−	+	+	?	+
Yeh et al. 2012 Taiwan86	+	+	+	−	+	+	+	+
Yu et al. 2011 Taiwan87	+	?	+	−	+	+	?	+
Ma et al. 2011 China88	+	+	+	−	+	+	?	+
Li et al. 2013 China89	+	?	+	?	+	+	?	+
Deluze et al. 1992 Switzerland90	+	+	+	−	+	+	?	−
Ng et al. 2013 Hong Kong91	+	+	+	−	+	+	+	+
Lee and Lee 2009 Republic of Korea92	+	?	+	−	+	+	?	+
Tam et al. 2007 Hong Kong93	+	+	+	−	+	−	+	+
Kvorning and Akeson 2010 Sweden94	+	?	+	−	+	+	?	+
Waite and Clough 1998 UK95	+	?	+	?	+	+	?	+
Holzer et al. 2011 Austria96	+	?	+	−	+	+	?	+
Sator-Katzenschlager et al. 2006 Austria97	+	?	+	−	+	+	?	+
Liu et al. 2011 China98	+	+	+	−	+	+	+	−
Liu et al. 2014 China99	+	+	+	−	+	+	+	+
Maeda et al. 2013 USA100	+	?	+	−	+	+	?	?
Shi et al. 2011 China101	+	?	+	?	+	+	?	+
Dias et al. 2010 Brazil102	+	?	+	?	+	+	?	−
Zhang et al. 2013 Hong Kong103	+	?	+	?	+	+	?	+
Sangdee et al. 2002 Thailand104	+	?	+	−	+	+	?	+
Johansson et al. 2001 Sweden105	+	+	+	−	+	+	?	+
Hopwood et al. 2008 UK106	+	+	+	−	+	−	?	−
White et al. 1998 England107	+	+	+	−	+	−	?	−
Dias et al. 2014 Brazil108	+	?	+	−	+	+	?	−
Lin et al. 2013Taiwan109	+	+	+	−	+	+	?	+
Michalek-Sauberer et al. 2007 Austria110	+	+	+	−	+	+	?	−
Carlsson and Sjolund 2001 Sweden111	+	+	+	−	+	+	?	+

Note: A, Adequate sequence generation; B, Allocation Concealment; C, Blinding (participants); D, Blinding (personnel); E, Blinding (outcome assessor); F, Incomplete outcome data addressed; G, Free of selective reporting; H, Free of other bias. +, Yes; −, No; ?, Unclear.

Credibility of blinding

Only 24 out of the 94 studies reported the credibility of blinding in participants by conducting the creditability test in six types of sham EA methods. Twenty-three studies22232534364552575960636572818897102103105106107108110 proved to be successful and one study66 proved to be failure. All six types of sham EA methods were claimed to be successful in blinding. They are sham EA type A (10/24 with 1 failure)2260668197102103106108110, type B (5/24)23596588105, type Q (5/24)25343652107, sham EA type O (2/24)4557, sham EA type D (1/24)72, and sham EA type N (1/24)63.

Efficacy results of the included studies

All 94 studies involving 105 comparisons of real and sham EA groups provided the information for between-groups analyses. Among them, 59 real EA groups181921222428303334353738404142434448495153545759606162636465697071727375777879838486878890919294959799100104108 reported significant superiority over corresponding sham EA groups; forty-three real EA groups were not statistically better than sham EA groups; one study74 showed that sham EA group was superior to the real EA group; the remaining two studies50111 lacked original data for between-groups analyses and were stated as “T > C?”. The efficacy results of the studies are listed in Table 1 and summarized in Table 3 according to different types of sham EA methods and EA indications.

Compared with sham EA controls, EA therapy in about 56.2% (59/105 comparisons) of comparisons reported the specific effect. Correspondingly, the real EA was superior to sham EA for type N (75%, 3/4 comparisons), type F (71.4%, 5/7 comparisons), type D (66.7%, 4/6 comparisons) and type M (66.7%, 2/3 comparisons). The lowest percentage of positive efficacy result was 44.4% (8/18 comparisons) in sham EA type L. The positive rate of efficacy for the three most often used sham EA methods were 50% (13/26 comparisons) for sham EA type A, 44.4% (8/18 comparisons) for sham EA type L and 64.3% (9/14 comparisons) for sham EA type O.

The type of sham EA methods varied across different EA indications. The sham EA type A was most commonly used in RCTs for pain, anesthesia and osteoarthritis. The sham EA type D and sham EA type Q were applied mainly in stroke studies. The sham EA type B was commonly applied to RCTs on depression. The sham EA type L and sham EA type O were commonly performed in trials on obesity. The sham EA type F and sham EA type L were commonly used in studies on primary dysmenorrhea.

Discussion

To our knowledge, this is the first systematic analysis to address sham EA methods in RCTs. The numbers of publications and sham EA methods have been increasing every decade. We summarized seventeen kinds of sham EA methods according to three aspects as needle location, depth of needle insertion and electrical stimulation, whereas only ten types of sham EA methods were identified from 94 included RCTs involving 6134 participants. The three predominant types of sham EA methods used were sham EA type A, type L and type O ordinally. Only 24 out of 94 trials reported credibility test with the results of 23 success and 1 failure using six types of sham EA methods mainly as follows: sham EA type A (10/24 with 1 failure), type B (5/24) and type Q (5/24). The remaining 3 sham EA methods were only tested in 4 trials. About 56.2% of comparisons provided the evidence of specific effect of EA therapy, and the four types of Sham EA controls with highest positive rate of efficacy result were type N (75%, 3/4 comparisons), type F (71.4%, 5/7 comparisons), type D (66.7%, 4/6 comparisons) and type M (66.7%, 2/3 comparisons) ordinally. However, all types of Sham EA controls were used in a small number of trials. Thus, the evidence was insufficient to recommend any type of sham EA control despite of the high positive rate. The sham EA control was frequently used in RCTs for pain, anesthesia, stroke, depression, obesity and primary dysmenorrheal/menstrual pain, suggesting that these diseases are particularly worthy of further EA RCTs.

The ideal design of sham acupuncture method remains methodologically challenging112. Consequently, a great variety of emerging sham acupuncture methods have found their ways into present RCTs by using non-traditional Chinese medicine acupoint2627113, no or superficial penetration29 and no or suboptimal stimulation28. The sham procedures in acupuncture RCTs were previously summarized by He et al.9 as seven types. A previous review by Dincer et al.8 reported the classification of sham acupuncture as sham type I~V based on three respects as needle location, insertion and stimulation. In the present study, we focused on the sham EA methods according to three aspects as needle location, depth of needle insertion and electrical stimulation, and summarized seventeen types of sham EA methods. Ten types of sham EA methods were actually used in the included RCTs.

The main purpose of RCTs on EA is to evaluate its specific effect. An optimal sham acupuncture technique must be biologically inactive and psychologically credible114. A lot of practice has been done to make the sham components of EA less perceptually and operationally distinguishable from real EA intervention for the purpose of keeping the blinding status of the participants. Streitberger needles, blunted needles and verum needles were frequently used with foam, tape or tube for hiding acupuncture loci from subjects18192122. Furthermore, a pricking sensation was elicited by dull tips for concealing the perceptual differences2021235260636781. The sham EA device was often accompanied by indicator light or with sound signals for confusing the participants182237. In the present study, six types of sham EA method were tested as concealable control in terms of blinding of participants.

The top three types of sham EA methods used were sham EA type A, type L and type O. The most frequently applied sham EA method was type A, accounting for a popular belief in its inertness based on its absence of key EA components as needle stimulation and electrical stimulation as well as its indistinguishable manipulation on same therapeutic acupoints. In the present study, the validation of credible participant blinding of this sham EA type was reported by most credibility tests. The debate emerged over the past decades over the inertness of non-penetrating procedure since the slight acupressure effects and physiological activity might be evoked by the tactile stimulation from blunt needle tips even without skin penetration112115. Takayama et al.116 argued that non-penetrating placebo needle is at least clinically inert for pain alleviation based on their cross-over study reporting no analgesic effects of the skin-touch placebo needle over that of the no-touch placebo or that of the no-treatment control. However, conclusive evidence are out of our awareness up to now whether non-penetrating but skin-touch placebo needle plays a specific therapeutic role in other medical condition. Thus, the sham EA type A may be an promising candidate control for further RCTs on analgesic effects of EA and relative further researches are called for in aspect of any other conditions.

Sham EA type Q is deemed to be the most inert type of sham EA control because it avoids all therapeutic components, which also probably makes this sham method perceptually and operationally distinguishable from real EA intervention and to some extent results in problematic credibility of blinding in participant. In the present study, the credibility of participant blinding of this sham EA type was endorsed by five studies with credibility test25343652107. However, mechanical non-penetration can evoke brain responses in healthy subjects. Thus, controversy raised regarding whether this type of sham EA method is physiologically inert control117. Moreover, four-fifth of the studies were conducted on acupuncture naïve participants. There is a possibility that previous experience of acupuncture treatments might have an impact on present perception of verum and sham EA intervention, which should be rigorously controlled in EA RCTs to avoid bias from unblinding. With the informed consent lack of explicit information on the sham method, debates emerged over ethical acceptability of the study.

The second commonly used sham EA method was type L. It was found that the differential effects of real EA and sham EA, which were attributed to point location, was not consistent across studies and conditions within this sham EA type, suggesting that EA on non-acupoints might be efficacious as EA on therapeutic acupoints. Furthermore, the improvements from baseline were also observed by Sahin et al.26, Li et al.46 and Yu et al.43 The similar findings were previously presented by Moffet et al.118 showing that sham acupuncture at non-acupoints was as efficacious as true acupuncture. It seemed that in the above studies the specificity of acupoints does not exit and to some extent were in violation of traditional acupuncture theories. Li et al.46 stated that the specificity of acupoints was not present in EA treatment. However, Wang et al.79 argued for the specificity of acupoints in EA treatment based on the better effects of EA at acupoints than that at non-acupoints on certain clinical outcomes. From the heterogeneous evidence of acupoint specificity, no definitive conclusion could be drawn based on the paucity of available high-quality clinical trials119. The main issue in this sham EA type might lie in the accurate identification of non-acupoint rather than a rough location nearby traditional acupoint that might be responsible for specific effects, which rises the challenge of conducting appropriate sham control in EA clinical trials especially in the presence or absence of the mechanism of acupoint specificity and the consistency in finding actual point across the different practitioners. Nevertheless, it is unclear whether this type is a concealable control for participant blinding since this sham EA tested credibility in the present study. Therefore, it demonstrates that sham EA type L might not be adequately controlled from inert or concealable perspective only if the mechanisms of acupoints were explicitly explored or the validation of so-called non-acupoint was verified by further researches. Cautions should be taken for eliminating bias from this sham EA control type.

The third commonly used sham EA was type O. During the procedure, the shallow insertion was applied to simulate deep skin penetration and to ensure the blinding of participant. In the present study, the validation of participant blinding of this sham EA type was endorsed by two studies4557. As for the issue of inertness, a few studies reported that sham EA control improved baseline in certain clinical parameters compared with conventional group3246565758. Moffet et al.118 stated that shallow needling at non-acupoints might be as efficacious as real acupuncture. Lund et al.120 reported that minimal acupuncture based on superficial insertion was not a valid control from a physiological perspective. Hróbjartsson et al.121 held the view that sham EA type was not inert control from Chinese medicine perspective. A sham EA procedure with superficial needling at non-acupoints might have subliminal effects since the locations of points was nearby true acupoints or myotome. Moreover, it is likely that the superficial insertion was not consistently applied since the needling depth varied across differentacupoints on different body parts and the relatively deep insertions might be conducted for taking the weight of the attached electrodes. Thus, this sham EA type may be concealable control but far from inert control in RCTs for EA, unless the extent to which the sham procedure could be regarded as physiologically inert has been clarified.

In the present study, six types of sham EA method were reported as successful in blinding. However, further investigations are needed for confirmation, since half of the tested types of sham EA controls were reported in a small number of trials. It should be noted that studies included did not provide sufficient evidence of blinding in acupuncturist. Vase et al.122 stated that it was hard to get acupuncture intervention fully double-blinded. Although non-penetrating needle was previously reported as potential sham control to mask both participants and practitioners in acupuncture research123124125, a previous review demonstrated that the acupuncture intervention was not fully double-blinded122. New strategies should be implemented for the development of double-blind sham EA control in terms of both participants and acupuncturists.

RCTs are generally recognized as the gold standard for the efficacy of clinical interventions by excluding the non-specific effect via a placebo control126. However, one study reported that the effect of EA therapy was merely the non-specific effect63. In the present study, the number of real EA group with superiority to, no difference from and inferiority to corresponding sham EA group was 59, 43 and 1, respectively. Thus, more than half comparisons demonstrated that EA therapy existed specific effects. Within all types of sham EA methods, the highest effective rate were type N (75% 3/4 comparisons), type F (71.4%, 5/7 comparisons), type D (66.7%, 4/6 comparisons) and type M (66.7%, 2/3 comparisons) successively. Considering the small number of included studies within corresponding sham EA type, the evidence are still insufficient to recommend any type of sham EA control despite of the high positive rate.

In the present study, 43/105 comparisons reported that EA has no specific effects compared with sham EA controls202324252627293132363943444546475255565863666768768081828589939698101102103105106107109110. For a reason, the extent to which the individual component of EA intervention plays its therapeutic influences on the final outcomes is not clear during clinical treatment7. The debate consequently emerges regarding the therapeutic inactivity of sham EA control which is partially comprised of real EA components, such as suboptimal manual or electrical stimulation110. The probability in the specific effects of EA may be reduced by the potential activity produced by sham EA control. On the other hand, EA is a complex intervention method. Its therapeutic effects consist of specific effects from needling and stimulation components as well as moderately large nonspecific effects, which means that the efficacy results of RCTs for EA are more likely to be influenced by a variety of factors, such as patient/practitioner interaction and patient expectations127128. In the clinical use, EA may be more effective than manual acupuncture in some situations such as when strong, continued stimulation is required, and when treating pain, anesthesia, stroke, depression, obesity and primary dysmenorrhea/menstrual pain, suggesting that further RCTs with appropriate sham EA control are in need to verify the specific effects on above conditions.

There are several weaknesses in the present study. Firstly, the search and screen procedure were limited to randomized, parallel-controlled trials published in English. Thus, those trials with cross-over design or published in other than English language were omitted. Secondly, with the aim of evaluating the sham method in RCTs on EA, a generous criterion was established to select eligible studies. Therefore, it was not easy to examine the specific effect of EA by data synthesis from different outcomes and indications because of the heterogeneity of trials. Finally, the reported credibility test addressed blinding effects in participant rather than in both participant and acupuncturist. The credibility tests were not reported in all studies and the number of studies using sham EA types was small, and therefore the conclusion should be interpreted with cautions.

Conclusion

Ten types of sham EA methods were identified based on our scheme classification. Generally, sham EA type A, type L and type O were frequently used. Yet, further clinical trials are recommended to maintain standard methodology of concealable and inert placebo EA techniques. Only 24 out of 94 trials were reported as positive credibility test in six types of sham EA methods, where sham EA type A, type B and type Q were highly practiced. It is worthy to study further about the importance of concealable sham EA types. EA therapy in approximately, 56.2% of comparisons provided the specific effects. The four types of sham EA (N, F, D and M respectively) represented the highest positive rate of efficacy results. However, progressive evidences on specific effects are mandatory. The sham EA control was observed frequently in pain, anesthesia, stroke, depression, obesity and primary dysmenorrhea RCTs. Also, broader studies in these predominant diseases are advised.

Additional Information

How to cite this article: Chen, Z. et al. Sham Electroacupuncture Methods in Randomized Controlled Trials. Sci. Rep. 7, 40837; doi: 10.1038/srep40837 (2017).

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.