Acupuncture compared to placebo-acupuncture for postoperative nausea and vomiting prophylaxis: A randomised placebo-controlled patient and observer blind trial


K. Streitberger


This randomised, placebo-controlled, patient and observer blinded trial was conducted to determine whether acupuncture at the acupuncture point P6 is effective in preventing postoperative nausea and vomiting (PONV) compared to placebo acupuncture. Female patients (n = 220) scheduled for gynaecological or breast surgery were randomly assigned to two groups receiving either acupuncture (n = 109) or placebo acupuncture (n = 111). Each group was stratified for type of surgery and included two subgroups receiving intervention either before or after induction of anaesthesia. The incidence of PONV and/or antiemetic rescue medication within 24 h after surgery was the main outcome measure which showed no statistically significant difference between groups (43.7% acupuncture, 50.9% placebo, p = 0.27). The differences were more pronounced for patients having gynaecological surgery (48.9% acupuncture, 67.6% placebo, p = 0.07) than for those having breast surgery (38.7% acupuncture, 40.3% placebo, p = 0.86). The secondary outcome, vomiting, was significantly reduced by acupuncture from 39.6% to 24.8% (p = 0.03). Subgroup analysis showed no difference between applications of acupuncture before compared to after induction of anaesthesia.

Postoperative nausea and vomiting (PONV) is not only an unpleasant symptom for patients [1], but can also delay hospital discharge and increase use of resources [2]. The overall incidence of PONV is approximately 30%[3], increasing up to 79% in high risk patients [4]. Recommended strategies for minimising the incidence of PONV include identification of high risk patients, avoidance of emetogenic stimuli and multimodal drug therapy [5]. Among antiemetic drugs, a combination of serotonin antagonists and droperidol seems to be the most effective in preventing PONV [6]. According to a recent warning from the Food and Drug Administration (FDA) concerning the side-effects of droperidol, this drug is no longer recommended as first line therapy. Besides dexamethasone, total intravenous anaesthesia with propofol, supplemental oxygen and nonpharmacological techniques such as acupuncture have been discussed as part of a multimodal therapy.

According to the National Institute of Health conference on acupuncture in 1997, promising results have emerged showing the efficacy of acupuncture in adult PONV [7]. This was based on a review of 21 trials, of which 16 showed positive results in favour of acupuncture, electroacupuncture, transcutaneous electrical nerve stimulation or acupressure at the acupuncture point Pericardium 6 (P6) [8]. A more recent review [9] seemed to confirm the effectiveness of P6-stimulation to prevent PONV. However, the methodology, including number of patients, selection of control group, statistical analysis and the mode of stimulation of the acupuncture point, was very heterogeneous in these studies. Therefore, the authors stated that further randomised controlled trials using a better study methodology are needed. In many of the studies, multiple testing was performed, resulting in positive and negative results using different outcome criteria. Some existing negative outcomes are related to P6-stimulation during anaesthesia [10–12] or to studies including only children [13–16]. This implies that acupuncture might only work for conscious and adult patients and therefore the effect might be due to a psychological effect, which also could be achieved by intensive psychological preoperative preparation of the patient.

We designed a randomised, placebo-controlled, patient- and observer-blinded study to investigate whether acupuncture at P6 is better than noninvasive placebo acupuncture to prevent PONV without additional antiemetics within 24 h after surgery. A secondary objective was to determine whether there is a difference when acupuncture is applied before and when it is applied after induction of anaesthesia.


Following approval of the Ethics Committee of the University of Heidelberg, this study was conducted between January and August 2002 at the Department of Anaesthesiology in co-operation with the Department of Gynaecology, University of Heidelberg. We included women aged between 18 and 80 years, ASA (American Society of Anesthetists) classification I and II, scheduled for gynaecological or breast surgery under general anaesthesia, with signed written informed consent. Exclusion criteria were: acupuncture treatment during the last 6 months, pregnancy, nausea or vomiting during the past 24 h, lymphoedema of the upper limbs, eczematous skin changes at the location of the acupuncture point P6, and coagulopathy. The patients were informed in detail about the study design, including the use of penetrating and nonpenetrating needles and the possible risks of acupuncture treatment (haematoma, infection, fainting).

After each patient was entered in the study, the acupuncturist obtained randomisation allocation by phone from a member of the Coordination Centre for Clinical Trials, University of Heidelberg, who had no contact with study patients. An adequate concealment was thereby assured. The patients were randomly destributed by type of surgery (gynaecological or breast) to ensure balance between groups. Patients were randomly assigned to four groups, receiving acupuncture either before or after induction of anaesthesia or receiving placebo acupuncture either before or after induction of anaesthesia.

Only the acupuncturists knew the randomisation result. The patients, the observer of the endpoints, the nurses, the anaesthetists and all other staff members were not informed about the allocation. Blinding of the patients was ensured by using the placebo needle in the same therapeutic setting as acupuncture. The acupuncturists answered questions about acupuncture using an identical list of answers. To assess blinding, patients were asked what kind of needle they believe they had received.

The acupuncture point P6 (in Chinese called Neiguan) is located on the inside of the forearm, 2 cun (approximately 3 cm) above the midpoint of the transverse crease of the wrist, between the tendon of palmaris longus and flexor carpi radialis.

Twenty minutes before induction of anaesthesia, patients allocated to the acupuncture group were prepared for study treatment by marking the acupuncture point P6 at both forearms with a plastic ring covered by a sticking plaster. According to this procedure, a sham point was marked 1 cm proximal and 1 cm lateral to P6 on both forearms of the patients allocated to the placebo group. Patients in the acupuncture group received acupuncture through the plaster and plastic ring with a 0.32 × 30 mm stainless steel needle (asia med, Munich, Germany). The needles were placed until a dull needle sensation (deqi) occurred or to about 1 cm deep if patients were already anaesthetised. In the control group, patients received placebo-acupuncture with a blunted, telescopic placebo needle designed by the author and manufactured by asia med. This placebo needle simulates an acupuncture procedure without penetrating the skin (Fig. 1) [17].

Figure 1.

Placebo-needle when touching the skin (left) and after retraction of the needle into the handle (middle), real acupuncture needle (right). Copyright of The Lancet Ltd. 1998 [17].

Timing of acupuncture was in both groups according to subgroup allocation, either 20 min before or immediately after induction of anaesthesia. In all cases, the needle had to remain in place for 20 min. The interventions were performed by two trained acupuncturists (K.S., A.B.).

Anaesthesia was administered by different anaesthetists using a standardised technique. No antiemetic medication was given before or during surgery. All patients received midazolam 30 min before induction of anaesthesia. Anaesthesia was induced with intravenous propofol and fentanyl or alfentanil, and maintained with inhaled isoflurane 0.5–1.5% and nitrous oxide 60–70% in oxygen. Neuromuscular block was provided by intravenous vecuronium. If required in the recovery room, intravenous dimenhydrinate and dolasetron were administered as rescue antiemetics. Postoperative pain was treated with the pyrazolone derivate metamizol, intravenously or the nonsteroidal anti-inflammatory drug diclofenac, rectally. In cases with severe pain the μ-receptor opioid agonist piritramide, or alternatively pethidine, was administered intravenously.

Outcome measurement was evaluated by a blinded observer on the ward 24 h after the operation. In a structured interview, all patients were asked whether they experienced nausea or vomiting, including time of first occurrence. If patients had already been discharged, they were contacted by phone. Additional information was collected from the patients' charts.

Primary outcome measure was the number of patients with PONV within 24 h after surgery. PONV was defined as the occurrence of nausea or vomiting or the need for any antiemetic drugs.

Secondary outcome criteria were early outcomes in the recovery room and within 12 h after surgery, differentiation between nausea on a 4-point scale (none = 0, mild = 1, moderate = 2, severe = 3), the occurrence of vomiting and an additional requirement of antiemetic drugs, as well as the occurrence of adverse events related to acupuncture and questions concerning the credibility of the therapeutic setting.

The study was monitored by an independent clinical monitor of the Coordination Centre for Clinical Trials, University of Heidelberg, who did source data verification and ensured that the study was conducted according to good clinical practice.

Statistical analysis

The analysis of the primary outcome variable was carried out using the stratified Mantel–Haenszel test (α = 0.05, two-sided), stratified by time of intervention (before or after induction of anaesthesia). The primary null-hypothesis was that invasive acupuncture at P6 is not better than placebo acupuncture to prevent PONV without additional antiemetics during the first 24 h after gynaecological or breast surgery. Secondary null-hypothesis was that there is no difference if acupuncture is applied before or after induction of anaesthesia. Assumptions for the calculation of the sample size were based on a pilot study with 11 patients and on results of previously published studies [18–20]. A decrease of PONV from 55% to 35% was judged as clinically relevant. Based on the primary endpoint, 106 patients per group had to be included to detect this difference with a power of 80% (Chi squared-test, α = 0.05, two-sided). The use of the stratified Mantel–Haenszel test (stratification for time of intervention) was not taken into account for the sample size calculation as it only increased the power of the study.

Secondary outcome variables were analysed in the form of a descriptive comparison using Fisher's exact test for categorical variables, mean score test for ordered categorical variables and the Mann–Whitney U-test for continuous variables.

According to the intention-to-treat-principle, all randomly assigned patients who received surgery were included in the analysis. If assessment was not possible, the patient was counted as a therapeutic failure. Patients who were withdrawn from surgery by the gynaecologist after randomisation were assessed but not included in the intention-to-treat analyses because the gynaecologist's decision not to operate could not be influenced by the acupuncturist or the result of randomisation.


Between January and August 2002, 325 women having gynaecological or breast surgery were asked to participate in the study, of whom 67 did not fulfil the inclusion criteria, and 38 did not consent to participation. Of 220 randomly assigned patients, eight did not receive an operation, and one patient withdrew consent after randomisation; 109 patients were assigned to the acupuncture group, and 111 to the placebo group (Fig. 2). Baseline characteristics revealed no relevant differences between the two groups (Table 1).

Figure 2.

Patient flow chart.

Table 1.  Characteristics of patients (n = 211). Values are mean (SD) [range] or number (proportion).
CharacteristicsAcupuncture n = 105Placebo n = 106
  • *

    l.n. = lymphatic node. §n = 105.

Age; years44.3 (13.4) [18.3–79.4]46.5 (13.4) [22.4–74.5]
History of
 Migraine31 (29.5%)33 (31.1%)
History of PONV
 Nausea41 (39.0%)38 (36.2%)§
 Vomiting35 (33.3%)38 (36.2%)§
Anxiety concerning
 Result of surgery62 (59.0%)58 (55.2%)§
 PONV32 (30.5%)34 (32.4%)§
 Awareness40 (38.1%)31 (29.5%)§
 Not awaking53 (50.5%)42 (40.0%)§
 Postoperative pain46 (43.8%)32 (30.5%)§
Type of surgery
 Without l.n. dissection*3323
 With l.n. dissection*1316
 Other 2 3

The study was strictly conducted according to the protocol and all violations are reported. Of 212 patients included in the intention-to-treat analysis, 18 did not receive the allocated treatment intervention due to organisational reasons (for details see Fig. 2). Due to clinical requirements, anaesthetists sometimes needed to diverge from the standardised anaesthetic, mainly concerning drug dosage. There was no significant difference between groups. One of the patients received anaesthesia by mask and one patient in the acupuncture group was ASA III.

Duration of acupuncture was less than 15 min in 12 patients (6%), but was always more than 5 min without difference between groups. Duration of anaesthesia tended to be shorter in the acupuncture group but the difference was not statistically significant. Postoperative use of analgesics was similar in both groups (Table 2).

Table 2.  Characteristics of perioperative management (n = 211). Values are mean (SD) [range] or number (proportion).
CharacteristicsAcupuncture n = 105Placebo n = 106Difference (95% CI)
  • *Four risk factors for PONV according to Apfel et al.[4]: Woman, previous PONV, nonsmoker, postoperative opioids.

  • §

    Mean score test.

Duration of acupuncture, min18.8 (2.5) [8–22]18.8 (2.5) [7–21]0 (0.0; 0.0)
Duration of anaesthesia, min107.7 (63.2) [25–345]122.2 (69.6) [30–340]−14.5 (−25.0; 0.0)
Postoperative analgesics 61 (58.1%) 67 (63.2%)−5.1% (−16.8%; 8.6%)
 Opioids 44 (41.9%) 51 (48.1%)−6.2% (−19.9%; 7.8%)
 Metamizol 21 (20.0%) 19 (17.9%)2.1% (−9.2%; 13.3%)
Diclofenac 24 (22.9%) 36 (34.0%)11.1% (−23.5%; 1.7%)
Number of risk factors*
 1 0 0p = 0.5§
 2 15 (14.3%) 15 (14.2%) 
 3 34 (32.4%) 25 (23.6%) 
 4 39 (37.1%) 51 (48.1%) 
 5 17 (16.2%) 15 (14.2%) 

The intention-to-treat analyses included all 212 patients who received surgery (Table 3). Assessment of the primary and secondary endpoints was possible for all patients, except for the patient who withdrew consent. According to the protocol, her outcome was counted as treatment failure, i.e. occurrence of PONV. The quality of documentation was excellent.

Table 3.  Outcome measures (n = 212).
 Acupuncture n = 106Placebo n = 106Difference (95% CI)p-value
  • *


  • **

    **Fisher's exact test.

  • ‡Mean score test.

  • §

    Four patients received gynaecological and breast surgery at once.

  • #:

    n = 211 (acupuncture n = 105).

24-h follow up
PONV and/or antiemetics46 (43.4%)54 (50.9%)−7.5% (−21.2%; 6.5%)0.27*
 Before induction (n = 103)25 (48.1%)28 (54.9%)−6.8% (−26.3%; 13.4%)0.56**
 After induction (n = 109)21 (38.9%)26 (47.3%)−8.4% (−27.1%; 11.2%)0.44**
PONV and/or antiemetics
 Gynaecological surgery§ (n = 91)22 (48.9%)31 (67.4%)−18.5% (−38.1%; 3.2%)0.07*
 Breast surgery§ (n = 124)24 (38.7%)25 (40.3%) −1.6% (−19.4%; 16.3%)0.86*
 None61 (58.1%)60 (56.6%) 0.53
 Mild31 (29.5%)24 (22.6%)  
 Moderate 9 (8.6%)18 (17.0%)  
 Severe 4 (3.8%) 4 (3.8%)  
Vomiting#26 (24.8%)42 (39.6%)−14.8% (−27.5%; −1.6%)0.03**
Vomiting and/or antiemetics#34 (32.4%)49 (46.2%)−13.8% (−27.0%; 0.0%)0.05**
12-h follow-up
 PONV and/or antiemetics42 (39.6%)52 (49.1%) −9.5% (−22.9%; 4.5%)0.16*
Recovery room follow-up
PONV and/or antiemetics
 1 according nurses' protocol30 (28.6%)44 (41.5%)−12.9% (−25.8%; 0.6%)0.06**
 2 according patient's report14 (13.2%)16 (15.1%)−1.9% (−12.1%; 8.3%)0.69*

According to the definition of the primary outcome variable, PONV occurred within 24 h in 46 patients (43.4%) in the acupuncture group and in 54 patients (50.9%) in the control group. The absolute difference of −7.5% (95% CI: −21.2% to 6.5%) was not statistically significant (p = 0.27; Mantel–Haenszel test). This result was confirmed by complete case analyses, excluding the patient who withdrew consent (p = 0.23), as well as the analysis of all patients including the eight nonoperated patients treated as failures (p = 0.22) and the as-treated-analyses of the 199 patients who actually received treatment (p = 0.25).

Subgroup analysis showed no difference between application before and that after induction of anaesthesia in the acupuncture group (p = 0.43) as well as in the placebo group (p = 0.45) (Table 3). Comparing time of application regardless of placebo or acupuncture treatment, 53 patients (51.5%) had PONV if the intervention was applied before induction of anaesthesia compared to 47 (43.1%) if it was applied after induction (difference of 8.4%; p = 0.27; 95% CI: −5.7% to 21.9%).

Reduction of PONV seemed to be greater in patients having gynaecological surgery (n = 91, reduction of 18.5%, p = 0.07; 95% CI: −38.1% to 3.2%) than in those having breast surgery (n = 124, reduction of 1.6%, p = 0.86; 95% CI: −19.4% to 16.3%). In the largest subgroup of gynaecological surgery, 44 patients having gynaecological laparoscopy, a reduction of −17.5%, from 55% to 37.5%, was achieved (p = 0.36; 95% CI: −45.3% to 14.1%).

Analysis of secondary outcome criteria confirmed the tendency for better results with acupuncture (Table 3). The incidence of vomiting within 24 h was significantly lower in the acupuncture group (24.8%) than in the placebo group (39.6%) with an absolute difference of −14.8% (p = 0.03; 95% CI: −27.5% to −1.6%). Blinding proved to be successful; no patient seemed to have discovered his/her intervention. There was no difference in the guesses of any patients as to which intervention they believed they had received. Also there was no difference when asked about the intervention they received, whether it helped to tolerate the anaesthesia, whether it would be recommended to friends, whether it might work by psychological or physiological effects, or whether it might be helpful in other situations.

As to acupuncture-related adverse events, haematomas were reported by one patient in the acupuncture group and by two patients in the placebo group. Five patients in each group claimed an allergic reaction to the sticking plaster. No severe adverse events were reported.


This study was conducted with particular focus on adequate blinding of the patients, appropriate methodology and high-quality conduct according to good clinical practice. Performing acupuncture under anaesthesia would be the best method of blinding the patients. Because previous studies, however, have suggested that acupuncture does not work under anaesthesia [10–12], we decided to include subgroups receiving the intervention before or after induction of anaesthesia. To ensure blinding, all patients in the placebo group received the intervention with a placebo needle, which simulates penetration of the skin. Using the placebo needle, we avoided any invasive stimulation. To avoid acupuncture point specific acupressure effects, we applied the placebo needle not directly at P6 but at a sham point. The main outcome criteria ‘PONV and/or rescue antiemetics within 24 h’ was chosen because of its reliability and its major clinical relevance [21].

Randomisation was stratified by surgery, because we expected a higher incidence of PONV in patients having gynaecological surgery than in those having breast surgery. However, we expected similar effects of acupuncture in both groups. Calculation of the sample size was based on a clinically relevant reduction of the PONV rate of 20% by acupuncture compared to placebo acupuncture (from 55% to 35%).

In our study, we could not prove our hypothesis that acupuncture is more effective than placebo acupuncture in the prevention of PONV. The 95% confidence interval of the achieved reduction from 51% to 43% of 8% (95% CI: −21% to 7%) does just include the proposed 20% reduction but also would be compatible with a better result from placebo acupuncture. A separate exploratory analysis of the patients having gynaecological surgery and of those having breast surgery revealed unexpected findings. In contrast to our expectation, there was almost no difference in the occurrence of PONV between acupuncture (38.7%) and placebo (40.3%) in the subgroup of patients having breast surgery. However, PONV was reduced by acupuncture about 18.5% (from 67.4% to 48.9%) in the subgroup of patients with gynaecological surgery, which was close to our expectation of a reduction of 20%. Therefore, our study might suggest that acupuncture is effective for PONV prophylaxis in patients having gynaecological surgery but not in those having breast surgery. However, our study had insufficient power for this question because it was not designed to prove this new hypothesis. It might be of importance for further practice, as well as research, to investigate whether the effect of acupuncture depends on the type of surgery. Most of the positive results in previous studies have been reported in gynaecological surgery. One of the most rigorous smaller studies showed a significant reduction of PONV from 54% to 30% when acupuncture was applied at P6 under anaesthesia compared to no intervention in 81 patients having gynaecological laparoscopy [18]. This is similar to the reduction from 55% to 37.5% in our study in the subgroup of the 44 patients having a gynaecological laparoscopy.

Concerning timing of stimulation, we had unexpected results, too. In contrast to the hypothesis that acupuncture might only work in aware patients, no difference could be shown when the intervention was applied before and when it was applied after induction of anaesthesia. Surprisingly, there was even a tendency to better results if the intervention was applied under anaesthesia.

A statistically significant result was achieved in the secondary outcome criteria of vomiting within 24 h after surgery. The conclusion that acupuncture is more effective in preventing vomiting than nausea might be obvious, but has to be viewed with caution. Positive results in secondary endpoints or subgroup analysis might be due to multiple testing. In a previous smaller study with acupressure, the secondary outcome criterion vomiting also showed a significant difference, whereas the primary outcome of complete response showed no statistically significant difference [22]. Most of the other studies showed effects on nausea but not on vomiting [9].

No acupuncture specific adverse events beside some minor haematomas were reported in our study. This confirms the safety of acupuncture [23].

We used a simple standardised form of acupuncture with needling P6 once before surgery. The efficacy of stimulation of P6 as antiemetic prophylaxis may be different if applied more than once, with continuous stimulation, with transcutaneous electrical nerve stimulation, acupoint injections or by acupressure. Recent studies with acupressure at P6 showed contradictory results. Acupressure seemed to be effective in laparoscopic surgery [24, 25], but not in urological endoscopic surgery [26]. Positive results of stimulation by acupoint injection in paediatric surgery were limited to early PONV [27].

For further studies, it might be advisable not only to focus on P6, but also to investigate other acupuncture points with possible antiemetic effects, like ST 36 [28] or paravertebral acupuncture points [29]. However, these techniques are not as easy to apply as P6 stimulation and in the clinical setting there might not be enough time to place the needles appropriately. To introduce acupuncture or acupressure as a method for nonpharmacologic prevention of PONV into clinical routine, it is important to identify the most practical and effective technique of acupuncture as well as choosing subgroups with the largest potential benefit. In the context of previous studies, the results of our study indicate that acupuncture at P6 might be effective in patients having gynaecological surgery, but not in patients having breast surgery. It might be more practicable, and as effective, to stimulate patients after induction of anaesthesia. It is necessary to conduct further large high-quality clinical trials to confirm these hypotheses.


The authors thank Gunther Bastert (Director, Department of Gynaecology, University of Heidelberg), Sandra Ehrenpfordt (Monitoring), Evelyn Kludt (Datamanagement), Carmen Bauer (Datamanagement, Programming) and the nurses of the recovery room for support in conducting this study. The authors thank Andrew Vickers and Jürgen Windeler for reviewing the manuscript.

Funding/support: This study was supported by a grant of the University of Heidelberg (F.203583).