Article Text


Summaries and commentaries by Adrian White on a selection of recent acupuncture research studies

Statistics from

Clinical trials of effectiveness

Unreliable controlled trial for knee pain

This sham-controlled randomised controlled trial (RCT) of electroacupuncture (EA) (n=84 enrolled, n=40 analysed) tested the effect both on pain of osteoarthritis (OA) of the knee, and concurrent changes in plasma levels of β-endorphin and cortisol. The results are unreliable.


Patients (n=40) of either sex over 40 years old with primary OA knee were recruited into a patient blinded, sham-controlled study. They were allocated to two groups alternately by the order in which they were seen.

For EA, needles were inserted into ST34, ST35 and ST36, LR8, SP10 and distal point ST44. De qi was elicited, then EA at 3 Hz to the maximum tolerable intensity was given for 20–25 min. In the sham group needles were inserted at random points about 5 cm away from true acupuncture points to a depth of 5–10 mm; leads were attached but no current was passed. Both groups were treated daily for 10 days. Pretreatment and post-treatment Western Ontario and McMaster Universities (WOMAC) index of OA knee and Visual Analogue Scale for pain were recorded and blood samples were taken for the measurement of plasma cortisol and b-endorphin levels.


The change in WOMAC score for the sham group was a 0.7% reduction in the median score (IQR 0–4%), while the median reduction in the score was 72% (IQR 53–85%) for those receiving EA (p<0.001), shown in figure 1. Following EA treatment there was a rise in plasma β-endorphin, and a significant fall in plasma cortisol, and the difference between the groups for β-endorphin was just significant: p=0.046.


On the face of it, this study appears to be interesting and important, and in many ways the detail of the study is well described. It would be comforting to believe the results, there are several signs that they are unreliable.

It was not randomised—although the CONSORT diagram states ‘randomised’. In that diagram, the numbers do not add up. Fifty-eight patients were allocated to sham and 26 to acupuncture—and yet patients were described as being allocated alternately. This does not seem to fit. It looks as if additional participants may have been recruited to the sham group to restore the size of that arm. Thirty-eight dropped out of the sham arm—20 were ‘not satisfied’ and 18 did not attend all sessions. Only six were lost from the acupuncture group. This suggests that the control patients might have known they were not receiving active treatment, and the changes seen in the control group are surprisingly small. Many participants were illiterate, so the WOMAC scores were complete on their behalf—by the investigators, offering significant opportunity for bias.

Opioid peptides are known to have central nervous system effects but hardly cross the blood-brain barrier. The authors did not justify measuring β-endorphin the plasma as a marker of the mechanism.

The final surprise about this paper is that it was accepted by Pain.

Acupuncture for hot flushes in women with breast cancer

This RCT (n=50) compared acupuncture with venlafaxine for reducing vasomotor symptoms in women with breast cancer.


Fifty patients were randomly assigned to receive 12 weeks of acupuncture (n=25) or venlafaxine (n=25) treatment. The acupuncture was given twice weekly for 4 weeks, then weekly for 8 weeks. The venlafaxine group received 37.5 mg daily for 1 week, then 75 mg daily for the remaining 11 weeks. Health outcomes were measured for up to 1 year post-treatment.


Both groups exhibited significant decreases in hot flushes, depressive symptoms and other quality-of-life measures, including significant improvements in mental health from pretreatment to post-treatment. These changes were similar in both groups, indicating that acupuncture was as effective as venlafaxine. But from 2 weeks after the treatment period, the venlafaxine group experienced significant increases in hot flushes, whereas hot flushes in the acupuncture group remained at low levels. The venlafaxine group experienced 18 episodes of adverse effects (eg, nausea, dry mouth, dizziness, anxiety), whereas the acupuncture group experienced no negative adverse effects. Acupuncture had the additional benefit of increased sex drive in some women, and most reported an improvement in their energy, clarity of thought, and sense of wellbeing.


Vasomotor symptoms in this group of patients are often severe and difficult to treat. Venlafaxine is a selective serotonin reuptake inhibitor used as a non-hormonal alternative for vasomotor symptoms in women and men, though less effective than oestrogens. Acupuncture appears to be of equivalent effectiveness in women, and is a safe, effective and lasting treatment for vasomotor symptoms secondary to long-term anti-oestrogen hormone use in patients with breast cancer.

Reperfusion injury

This RCT (n=60) tested EA for prevention of reperfusion injury.


Sixty patients with acquired heart valve disease were randomly allocated to the EA pretreatment group or the control group. Patients in the EA group received EA stimulus at bilateral PC6, LU7 and LU2 for 30 min each day for five consecutive days before surgery. The EA was at frequency of 5/30 Hz at an intensity of 0.8–1.9 mA until the patient felt de qi. The control group had needles inserted but no EA stimulation.

The main outcome of interest was serum cardiac troponin I concentrations as an indicator of reperfusion injury after releasing the aortic clamp at the end of the operation. Hemodynamic data, mechanical ventilation time, inotropic drug use, morbidities and mortalities were compared between the two groups.


Concentrations of serum cardiac troponin I were significantly decreased in the EA group at 6 h, 12 h and 24 h after reperfusion (see figure 2) compared with the control group. There was no difference in the inotropic drugs administered to the two groups. EA pretreatment also shortened intensive care unit stay time (p<0.05).

Figure 2

Troponin concentrations, hours after surgery. EA, electroacupuncture.


Cardiac ischaemia-reperfusion injury after cardiopulmonary bypass contributes to postoperative morbidity and mortality, and is difficult to prevent. The idea for this study arose after observations in rats that EA alleviates cerebral and myocardial injury from ischaemia. The authors conclude that this simple and convenient treatment has the potential to be used in the clinic for reducing myocardial injury in patients with heart valve replacement surgery, and well worth further study.

Moxibustion for breech presentation

In this RCT (n=212) no effect of moxa for breech presentation was seen on rates of cephalic version.


This RCT was conducted in a Swiss university hospital maternity unit. The acupuncture point BL67 was stimulated by moxibustion daily for 2 weeks for 212 consenting women between 34 and 36 weeks of gestation with a single fetus in breech presentation. The intervention was given three times weekly in the hospital with a teaching session and an information leaflet on the technique for additional daily therapy at home. The control group received expectant management care. The availability of external cephalic version was maintained the same for both groups. The main outcome measure was the comparison of the proportion of women with cephalic presentation at delivery.


Baseline characteristics were similar between groups, except more nulliparous women were randomised to moxibustion. The percentage of versions performed was similar between groups: 18% in the moxibustion group compared with 16% in the control group (RR 1.12, 95% CI 0.62 to 2.03). Adjustment for the imbalance in parity did not change these results. The frequency of caesarean delivery was similar (64% compared with 58% in the moxibustion group and the control group respectively). Acceptability of the intervention and women's perceptions of moxibustion were favourable.


No beneficial effect was observed of moxibustion in converting breech to cephalic presentation. Despite this lack of proven effectiveness, women had positive opinions on the intervention.

Systematic reviews

Induction of labour


The usual English-language databases were searched for all studies published on the subject between 1970 and 2008, and 10 studies on labour induction were identified.


The points used included SP6, LI4, (the two most common) GV1, GB34, LR3, GV7, HT7, PC6, ST36 and CV4. The duration of labour as a result of acupuncture treatment ranged from 10 h 20 min to 29.1 h. All of the studies demonstrated labour induction by acupuncture treatment. However, because two RCTs reported that there was no statistically significant effect of acupuncture, these results are more suggestive than definitive. Furthermore, although the relationship between cervical ripening and interleukin 8 (IL-8), prostaglandin F(2α) (PGF(2α)) and β-endorphin is well documented in the literature, there is no evidence to suggest that acupuncture alters these mediators. Serum levels of IL-8, β-endorphin and PGF(2α) were not found to be significantly influenced by acupuncture.


This is an ‘old-fashioned’ narrative review, and thus gives a useful summary of the literature at an early stage of research into a topic. If they had done a formal meta-analysis they would simply have come up with a plainly ‘negative’ conclusion—the results do not overall show a significant effect. However, this would be an oversimplification: there were five RCTs and all showed at least a trend in favour of acupuncture. One showed a significant reduction in time to delivery and use of oxytocin. Unfortunately, the authors did not report whether manual or electrical stimulation was used in each study, nor the treatment schedules, so it is difficult from these data to recommend any particular design for another study—though there is definitely enough evidence to justify one.

The authors include some history of the subject. Electrical stimulation of first the cervix and then the abdominal wall was used in the 1950s to induce labour in women with postmature pregnancy. Perhaps that relates to the widely expressed anxiety about the risk of premature labour from acupuncture in early pregnancy—which is unfounded.



Nineteen electronic databases, including English, Korean, Japanese and Chinese databases, were systematically searched for RCTs investigating acupuncture for primary dysmenorrhoea up to July 2008 with no language restrictions.


Twenty-seven RCTs were systematically reviewed with 2960 participants. Twenty-three were published in Chinese, three in Korean and one in English. Various forms of acupuncture were included, such as auricular, heated needles, EA and acupuncture point injection. Many points were used but SP6 was common to most, and most started treatment about 7 days before the onset of menstruation. Most trials compared acupuncture with other treatments.

Compared with pharmacological treatment or herbal medicine, acupuncture was associated with a significant reduction in pain (see figure 3), but two RCTs did not find a significant difference between acupuncture and sham acupuncture.

Figure 3

Meta-analysis of acupuncture for dysmenorrhoea.


This review achieved some publicity, but the data extracted from the studies show remarkably high response rates, typically over 90% in the acupuncture group and nearly 80% in the controls. In view of the methodological problems in the studies, this evidence should not be regarded as convincing.


A standard type of systematic review of seven publications.


Searches of the usual English-language databases, and meta-analysis of all relevant outcomes for fibromyalgia namely pain, fatigue, sleep disturbances, reduced physical function and side effects at post-treatment.


Seven RCTs with a median treatment time of nine (range 6–25) sessions and 385 patients were included. Two RCTs included a median follow-up of 26 weeks. Standardised mean differences (SMDs, equivalent to effect size) comparing verum and control acupuncture were calculated. Strong evidence for the reduction of pain (SMD −0.25; 95% CI −0.49, −0.02; p=0.04) was found at the end of treatment, see figure 4.

Figure 4

Meta-analysis of acupuncture for fibromyalgia pain. SMD, standardised mean differences.

There was no evidence for the reduction of fatigue and sleep disturbances, or the improvement of physical function at post-treatment. Nor was there any evidence for the reduction of pain and improvement of physical function at the latest follow-up.

Subgroup analyses resulted in moderate evidence for a significant and small reduction of pain at post-treatment in two studies with electrostimulation with an effect size of 0.43 (CI 0.04 to 0.81) compared to manual acupuncture (effect size 0.19, CI 0.14 to 0.52). Individualised acupuncture was superior to standardised acupuncture. Stratifying the type of controls (penetrating vs non-penetrating control acupuncture) did not change the results. Significant reduction of pain was only present in studies with risk of bias. Side effects were inconsistently reported.

The authors concluded that acupuncture could not be recommended for fibromyalgia on the current evidence.


These authors use a little phrase that seems neat and correctly cautious—but actually condemns a positive finding. It is increasingly used. The small effect of acupuncture “was not clearly distinguishable from bias”. This is itself a biased statement, since it concentrates on possible overstatement of the effect (due to possible bias) but ignores the likely understatement of the effect, that is, the control procedure in these studies was an active treatment.

The authors make some general recommendations for further research, but no specific advice for example on EA.

Diagnostic study

This study is about trigger point sites and pain symptoms in men with chronic prostatitis/chronic pelvic pain syndrome.


A cohort of 72 men was randomly selected who underwent treatment with physiotherapy and relaxation training from 2005 to 2008. Patients self-reported up to seven pelvic pain sites before treatment and whether palpation of internal and external muscle trigger points reproduced the pain. Fisher's exact test was used to compare palpation responses, that is, referral pain, stratified by reported pain site.


Pain sensation at each anatomical site was reproduced by palpating at least two of 10 designated trigger points. Furthermore, five of seven painful sites could be reproduced at least 50% of the time (p<0.05). The most prevalent sites for the pain were the penis in 90.3%, the perineum in 77.8% and the rectum in 70.8%. Puborectalis/pubococcygeus and rectus abdominis trigger points reproduced penile pain in more than 75% of cases (p<0.01). External oblique muscle palpation elicited suprapubic, testicular and groin pain in at least 80% of the patients at the respective pain sites (p<0.01).


Identifying the site of clusters of trigger points inside and outside the pelvic floor may assist in understanding the role of muscles in this disorder and provide focused therapeutic approaches.

Basic research

Double-blind evaluation of acupuncture treatment has not previously been reported. We investigated the possible benefits of a new placebo needle. This study tested the analgesic effects of needling, using a doubleblind procedure.


This was a double-blind crossover study of penetrating and non-penetrating (placebo) acupuncture trials. Fifty-six healthy volunteers were needled at LI4 for 20 min with either real or double-blind control non-penetrating needles. Electrical stimulation pain rating in the forearm was tested using a numeric rating scale (0–150). This was done before and during needling, and up to 40 min after the removal of the needle. Pain from skin penetration and deep, dull pain (de qi) associated with needle application, which is considered essential for achieving successful acupuncture analgesia, were also recorded.


No significant difference was found in analgesic effects between the penetrating and non-penetrating needle trials. In addition, no significant correlation was found between analgesic effect and de qi. A significant analgesic effect was observed during needle application and immediately after needle removal for both the penetrating and non-penetrating needle trials when compared with before acupuncture control (p<0.001).


Needle penetration did not confer a specific analgesic advantage over non-penetrating (placebo) needle application.

View Abstract


  • Provenance and peer review Not commissioned; internally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.