Auricular acupuncture for postoperative pain control: a systematic review of randomised clinical trials


Dr Taras Usichenko


The number of publications on the peri-operative use of auricular acupuncture has rapidly increased within the last decade. The aim was to evaluate clinical evidence on the efficacy of auricular acupuncture for postoperative pain control. Electronic databases: Medline, MedPilot, DARE, Clinical Resource, Scopus and Biological Abstracts were searched from their inception to September 2007. All randomised clinical trials on the treatment of postoperative pain with auricular acupuncture were considered and their quality was evaluated using the Jadad scale. Pain intensity and analgesic requirements were defined as the primary outcome measures. Of 23 articles, nine fulfilled the inclusion criteria. Meta-analytic approach was not possible because of the heterogeneity of the primary studies. In eight of the trials, auricular acupuncture was superior to control conditions. Seven randomised clinical trials scored three or more points on the Jadad scale but none of them reached the maximum of 5 points. The evidence that auricular acupuncture reduces postoperative pain is promising but not compelling.

The debate whether acupuncture is an effective method of pain-control is continuing [1]. It is confounded by the fact that acupuncture is an umbrella term for a whole range of different forms of treatments. One such therapy is auricular acupuncture or auriculotherapy, which is used for a wide range of conditions, e.g. drug dependence, anxiety and pain [2, 3]. Auricular acupuncture (AA) is defined as ‘a form of acupuncture in which needles are placed in various positions of the ear to affect the person. It postulates body correlates on the ear, so a treatment performed upon the ear will have effects reflected on the body part’ [4]. At present, AA is heavily promoted and hugely popular; a Google search for AA (accessed 1 December 2007) generated 270 000 hits.

Postoperative pain seems to be a convenient model to test the efficacy of AA: there is no shortage of patients suffering from comparably painful conditions, investigators can use standardised instead of individualised acupuncture and clear common endpoints exist e.g. pain intensity and analgesic consumption. It is therefore hardly surprising that, in recent years, several clinical trials of AA have been published. The aim of this systematic review is to critically evaluate the totality of the data from rigorous trials with a view of generating a reliable verdict on the efficacy of AA for postoperative pain-control.


Data sources

Electronic databases were searched from their respective inception to September 2007 using the following databases: Medline, Med Pilot, DARE, Clinical Resource, Scopus and Biological Abstracts. The search terms used were: auricular acupuncture, postoperative pain, clinical trial, pain intensity, analgesic requirement. In addition, hand-searches were done of our own files.

Study selection

Prospective, controlled, clinical trials were included if they were randomised studies of AA that compared this approach with any other type of control intervention in human postoperative patients. We excluded trials, which were not randomised, or did not report sufficient methodological details for evaluation. No language restrictions were imposed.

Data extraction, quality and validity assessment

Hardcopies of all potential trials were obtained and read by all authors. Data from the articles were validated and extracted according to pre-defined criteria (Table 1). Their methodological quality was evaluated using the Jadad score [5], which is a tool for evaluating the likelihood of bias on a 5-point scale. Two of the authors (T.U. and C.L.) were the authors of three studies eligible for inclusion in the review. In order to avoid potential bias, the initial data extraction and analysis was performed by another reviewer (E.E.) and verified later by two former authors. The discrepancies concerning the inclusion of trials in the review were solved by discussion.

Table 1.   Key data of the included RCTs.
First author (year)SurgeryNo. of patientsVerum procedure
(no. of patients)
Control procedure
(no. of patients)
Primary outcomeResultsJadad scale
  1. AA, auricular acupuncture; EAA, electro auricular acupuncture; THA, total hip arthroplasty; ST, standard therapy; SA, sham acupuncture; LapCCE, laparoscopic cholecystectomy; AKA, ambulatory knee arthroscopy.

Wang 1988 [12]Thoracotomy36EAA (19)Standard therapy (17)Pain intensity
Opioid requirement
Pulmonary function
EAA better than control for all measured parameters1
Lewis 1990 [13]Burns11EAA (11)Placebo pill (11)Pain intensityEAA sessions reduced pain intensity3
Mann 1999 [14]THA100AA (50)ST (50)Analgesic requirement, pain intensityPain intensity & analgesic requirement decreased in AA vs ST2
Usichenko 2005 [15]AKA20AA (11)SA (9)Ibuprofen requirementIbuprofen requirement decreased in verum vs sham3
Usichenko 2005 [17]THA61AA (31)SA (30)Opioid requirementOpioid requirement decreased in verum vs sham4
Sator-Katzenschlager 2006 [18]Oocyte aspiration94EAA (32)
AA (32)
No acupuncture and no electrical stimulation (30)Opioid requirement
Pain intensity
Pain intensity & opioid requirement was highest in control (non-intervention) group4
Michalek-Sauberer 2007 [19]Molar extraction149EAA (76)
AA (37)
No acupuncture and no electrical stimulation (36)Opioid requirement
Pain intensity
No differences among groups4
Usichenko 2007 [16]AKA120AA (61)SA (59)Ibuprofen requirementIbuprofen requirement decreased in verum vs sham4
Likar 2007 [20]LapNE44EAA (21)AA without electrical stimulation (20)Opioid requirement
Pain intensity
Both parameters reduced in verum group vs sham group3


Pain intensity and analgesic requirements were defined as the primary endpoints. A meta-analytic assessment of the data was considered but had to be abandoned because of the clinical and statistical heterogeneity of the primary studies.


Our searches found 23 articles. Closer inspection of the abstracts showed that only 15 were clinical trials. Of these, six failed to meet our inclusion criteria [6–11] (Fig. 1). The key data of the remaining nine randomised clinical trials (RCTs) are summarised in Table 1. Additional information is provided below.

Figure 1.

 Flowchart of included studies.

Wang et al. reported a small study in which patients were randomised into two groups after thoracotomy [12]. The experimental group (n = 19) received electro AA at two points on each side (2–20 Hz) for 10 min followed by injections of vitamin B complex into these points. This was repeated every 12 h for 7 days. The control group (n = 17) received standard therapy alone. All endpoints, including pain intensity, opioid analgesic requirement and standard pulmonary function tests, favoured AA.

Lewis et al. conducted a crossover RCT in which 11 patients with burns received either bilateral electro AA followed by a placebo pill or the same treatments in reverse order [13]. Pain intensity, measured 15, 30 and 60 min after treatment, was lower after acupuncture condition vs placebo pill.

Mann reported in his MD thesis an RCT of 100 patients who were treated either with AA plus standard analgesics or analgesics alone after hip or knee arthroplasty [14]. The results demonstrated decreased analgesic requirement and pain intensity as well as increased satisfaction with postoperative pain therapy and quality of life in the AA compared to the control group.

Usichenko et al. reported a pilot RCT with 20 patients undergoing knee arthroscopy [15]. They received either AA (fixed indwelling acupuncture needles retained for 1 day) or a sham procedure before surgery. The sham acupuncture consisted of inserting needles in non-acupuncture points on the helix. Analgesic consumption was less in the experimental compared to the control group whereas pain intensities were similar.

Following their pilot study, the same research group performed an RCT with 120 patients undergoing ambulatory knee surgery [16]. The methodology was the same as used previously. Intention to treat analysis favoured real over sham treatment in terms of analgesic consumption (median 200 mg ibuprofen in AA group vs 600 mg in control group) while pain intensity was similar in both groups. Deblinding was assessed but there was no evidence that this had occurred.

Another RCT of Usichenko et al. included 61 patients undergoing total hip arthroplasty [17]. This time, four specific AA points were tested vs four non-acupuncture points of the helix, and fixed indwelling needles were left in situ for 3 days after surgery. Requirement of opioid, delivered via patient-controlled analgesia device during the first 36 h after surgery, was reduced by 36% with real acupuncture in comparison to sham procedure. Pain intensities were similar in both groups. Here, the investigators tried to limit therapist-patient interactions in an attempt to minimise the risk of patient-deblinding. They also asked patients what treatment they thought they had received and found no indication of significant levels of deblinding.

Sator-Katzenschlager et al. conducted an RCT with 94 women undergoing oocyte aspiration [18]. They received electro AA with (group 1) or without electrostimluation (group 2) or placebo acupuncture (group 3), which consisted of adhesive tapes applied to the ear without needles and without electrostimulation. Pain relief was more pronounced in group 1 compared to group 2 or group 3, while the consumption of rescue opioid was higher in group 3.

Michalek-Sauberer et al. reported a similar RCT with 149 patients undergoing third molar extractions [19]. Patients received either AA with electrostimulation, AA without electrostimulation, or a placebo procedure without stimulation. There were no inter-group differences in terms of patients’ pain experience or analgesic consumption.

Likar et al. conducted an RCT with 41 patients undergoing laparoscopic nephrectomies [20]. They received either electro AA or a sham treatment. Pain and postoperative morphine consumption favoured real over sham treatment. Deblinding was not assessed in this study.


Except for one RCT [19], all of the included studies suggest that AA is effective in reducing postoperative pain. Many of these RCTs are, however, far from rigorous. Prevalent problems include small sample size and inadequate control of non-specific-effects or deblinding. The Jadad score ranged from 1–4. Therapist-blinding was not incorporated in any of the RCTs, the reason why none of them attained the maximum score of 5. The more recent studies seem to be more rigorous than the older ones (Table 1). Of the four rigorous RCTs that include control for non-specific effects [16–19], one study suggests that AA is ineffective [19] while three favour real over sham AA.

Auricular acupuncture is used for many conditions other than pain; here the evidence is mixed. For cocaine dependence, two independent systematic reviews concluded that there is no evidence demonstrating the effectiveness of AA [21, 22]. For smoking cessation, a recent meta-analysis in favour of AA demonstrated that the effects of acupuncture may not depend on point location [23]. Another recent systematic review concluded that AA appears to be effective for treating insomnia [24]. Moreover, there is suggestive but not compelling evidence implying that AA may alleviate peri-operative anxiety [25–27], reduce anaesthetic requirement during general anaesthesia [28] and control cancer pain [29]. The history of AA is somewhat nebulous. Many authors believe that its roots are in Traditional Chinese Medicine (TCM) [1, 2]. The TCM concept does obviously include several acupuncture points in the ear. Yet AA is fundamentally different from TCM; it is based on Paul Nogier’s assumption of an inverse representation of the entire human body on the outer ear (head towards the ear lobe). This is often referred to as ‘somatotopic representation’ [30]. There is tentative support for this concept in relation to musculoskeletal problems [31] and heart disease [32], but not in relation to the treatment of chronic pain [33]. AA is ‘based on a self-contained model of thought’ [34]. Nogier states in his book ‘Man in the Ear’: ‘The auricle is not simply a piece of uneven cartilage. It has a close connection with the internal organs. When an internal organ has a disorder, a corresponding reaction will emerge in the auricle’ [34]. Many modern textbooks agree that AA is a recent development, which is largely unrelated to TCM [35, 36]. ‘There are no references to a concept of AA by the standard authors of Chinese medicine…’, and the findings of Nogier ‘…were not known about in China until the beginning of 1959’[30].

No anatomical pathways exist to directly connect inner organs with the ear. The innervation of the central part of auricle (triangular fossa and concha) comes from trigeminal, geniculate and superior vagal ganglions, whereas the peripheral regions receive their innervation mainly by spinal nerves [37–39]. The central parts of the neurons constituting the auricular branch of the vagal nerve are situated in the superior ganglion of the vagal nerve and nucleus tractus solitarii [40]. Functionally, the stimulation of the inferior concha induced a significant increase in parasympathetic activity [41]. In another study, the stimulation of the sympathetic AA point significantly decreased the stimulus-evoked electrodermal response compared with an AA stimulation to a non-specific point of the helix [42]. This resembles the design of studies performed by Usichenko et al. [15, 16], where stimulation of the points in the central regions of the auricle was better than sham acupuncture at the non-specific points of the helix for reduction of postoperative analgesic requirement.

Even if AA were based on a sound theory and imbedded in the history of TCM, a sceptical attitude about the validity of its underlying concepts would be in order [43]. With this background, although the results of those RCTs that suggested effectiveness of AA made an attempt of controlling for patients’ expectation and had a Jadad score of 4 [16–18], they should be viewed critically.

None of the three RCTs in question [16–18] blinded the therapist, although the patients blinding was tested and appeared to be appropriate in two of these studies [16, 17]. Therapist blinding is obviously difficult to achieve – but it is not impossible. For instance, a fully trained therapist could, according to a random code, mark the placebo or real points with a felt pen in the patients’ ears. Subsequently, a blinded research nurse who is taught how to place the needles but does not know the location of AA points, could insert the needles.

The use of a visual analogue scale for pain as a study endpoint seems problematic, yet it was used in two of the RCTs [17, 18]. As patients were using self-administered analgesia, one would expect pain to be well-controlled at all times. Subjective measures of pain are therefore only meaningful study endpoints when standard doses of analgesia are administered. In one RCT [18], the power calculation was based on this outcome measure, which may have introduced an error in the statistics of this study. The same RCT also failed to report the success of blinding the patients.

Given these caveats, the evidence for AA seems less than convincing. Future studies should overcome some of the shortfalls of the current RCTs. They should include therapist blinding to minimise potential sources of bias. Furthermore, independent replications are required; presently, the most reliable RCTs originate from only two different research groups.

In conclusion, the evidence that AA reduces postoperative pain is promising but not compelling. Further rigorous research and independent replications, which effectively exclude bias, seem warranted.


The authors thank Peggy Lietz and Danika Rimasch for their help in retrieving the literature for this review.