Efficacy and safety of electrical acupoint stimulation for postoperative nausea and vomiting: A systematic review and meta-analysis

Background Postoperative nausea and vomiting are typical postsurgical complications. Drug therapy is only partially effective. The goal of our meta-analysis is to systematically evaluate the efficacy and safety of electrical acupoint stimulation for postoperative nausea and vomiting and to score the quality of evidence supporting this concept. Methods PubMed, Embase, Cochrane Library, Web of Science, and ClinicalTrials.gov were searched from inception to March 19, 2020. Results Twenty-six studies (2064 patients) were included. Compared with control treatment, electrical acupoint stimulation reduced the incidence of postoperative nausea and vomiting (RR 0.49, 95% CI 0.41 to 0.57, P < 0.001), postoperative nausea (RR 0.55, 95% CI 0.47 to 0.64, P < 0.001) and postoperative vomiting (RR 0.56, 95% CI 0.45 to 0.70, P < 0.001). Electrical acupoint stimulation also reduced the number of patients requiring antiemetic rescue (RR 0.60, 95% CI 0.43 to 0.85, P = 0.004). No differences in adverse events were observed. Subgroup analysis showed that both electroacupuncture (RR 0.58, 95% CI 0.46 to 0.74, P < 0.001) and transcutaneous electrical acupoint stimulation (RR 0.44, 95% CI 0.34 to 0.58, P < 0.001) had significant effects. Electrical acupoint stimulation was effective whether administered preoperatively (RR 0.40, 95% CI 0.27 to 0.60, P < 0.001), postoperatively (RR 0.59, 95% CI 0.46 to 0.76, P < 0.001), or perioperatively (RR 0.50, 95% CI 0.37 to 0.67, P < 0.001). The quality of evidence was moderate to low. Conclusions Electrical acupoint stimulation probably reduce the incidence of postoperative nausea and vomiting, postoperative nausea, postoperative vomiting, and reduce the number of patients requiring antiemetic rescue, with few adverse events.


Introduction
Reporting Items for Systematic Reviews and Meta-Analyses [32] and Assessing the methodological quality of systematic reviews (AMSTAR) guidelines [33].
The control treatment (no acupuncture + active device, no acupuncture + inactive device, gel electrodes + inactive device, or usual care) as the control group: 1. No acupuncture + active device means the needles were bent to lay flat against the skin, or no needles were applied. Insulated wires from the activated stimulator box with a normal current output were attached to the needles or the inside of the arm covers. 2. No acupuncture + inactive device means the needles were bent to lay flat against the skin, or no needles were applied. Insulated wires from the inactivated stimulator box with no current output were attached to the needles or the inside of the arm covers. 3. Gel electrodes + inactive device means gel electrodes were placed on the acupoints and attached to a stimulator box with no current output.

Inclusion and exclusion criteria
The inclusion criteria were specified by the Population, Intervention, Control, Outcomes, and Study design (PICOS). The inclusion criteria were: 1. The review included emergency and elective surgery patients that underwent general anesthesia, regardless of age, sex, ethnicity, or surgery type. 2. The intervention measure in the treatment group was EAS (EA or TEAS). 3. the intervention measures in the control group were sham acupuncture, sham acupoint, sham electrical stimulation, or preoperative routine nursing. 4. Primary outcomes were the incidence of PONV, postoperative nausea (PON), or postoperative vomiting (POV). Secondary outcome measures were the numbers of patients requiring antiemetic rescue and adverse events. 5. The study was a randomized controlled trial (RCT) (blinded or non-blinded).
The exclusion criteria were as follows: 1. Patients undergoing cesarean section or abortion or those in the control group who received any electrical stimulation. 2. American Society of Anesthesiologists (ASA) � III. 3. Literature with incorrect data or inaccessible data.

Study selection and data collection
Two researchers independently used EndNote (version X8.1, Clarivate Analytics, Philadelphia, United States) reference management software for literature classification, preparation, and removal of duplicates. Then, we excluded nonrelevant studies that did not meet the inclusion criteria after reading the title and abstract of each article. The articles that could not be excluded based on the title and abstract were retrieved for a full-text screening. If necessary, we acquired additional information from the trial authors by email or telephone.
A standardized data extraction list in an Excel spreadsheet (Microsoft Corporation, Redmond, Washington, United States) was used to collect information from the included studies: author, year, country, study design, sample size, age, types of surgery, intervention measures, duration and outcome indicators. This process was performed by two researchers independently and cross-checked after completion. When opinions differed, a third reviewer was consulted, and the case was reviewed for consensus.

Outcomes
The primary outcomes were the incidence of PONV, PON, or POV. The secondary outcome measures were the numbers of patients requiring antiemetic rescue and adverse events.

Risk of bias assessment
The risk of bias of the clinical trial was independently assessed by two reviewers using "The Cochrane Collaboration's tool" [34], and any disagreements were solved by discussion. Furthermore, publication bias was evaluated by funnel plot analyses if sufficient studies were included.

The Grades of Recommendation Assessment Development and Evaluation
The quality of the evidence was assessed by the Grades of Recommendation Assessment Development and Evaluation (GRADE) system [35]. The GRADE system includes the risk of bias, inconsistency, indirectness, inaccuracy, and publication bias [36]. Bias risks included inappropriate randomization methods, allocation concealment, blindness, and excessive loss of follow-up data. The inconsistencies mainly involved different intervention or evaluation metrics. Indirectness covered direct and indirect comparisons of results between two groups. The inaccuracy was mainly judged by the width of the confidence intervals. Publication bias was due to unpublished studies (usually a negative result) by the investigator.
For each outcome, we initially awarded four points to each RCT and then downgraded the point total for defects regarding the 5 aspects. The quality of evidence was classified as A (high quality), B (medium quality), C (low quality), or D (very low quality) [37].

Statistical analysis
Statistical analyses were carried out using RevMan 5.4 (RevMan, the Cochrane Collaboration, Oxford, United Kingdom) software. We examined 5 outcomes, each as dichotomous variables. The differences in categorical outcomes between the EAS and control groups were reported as the relative risk (RR) with a 95% confidence interval (CI). Significant heterogeneity was considered when I 2 > 50% or P < 0.1, and a random-effects model was used for analysis [38]. Otherwise, a fixed-effects model was used.
If I 2 > 50% in a group of studies, suggesting there was high heterogeneity, then we used sensitivity analysis or subgroup analysis to investigate the possible reasons from both clinical and methodological perspectives [39].

Study selection
A total of 864 articles were retrieved from the database search. By screening the article titles, abstracts, or both and removing duplicates, 794 records were excluded. Seventy full-text articles were included for rescreening. We excluded 3 reports published as abstracts, 15 articles lacking relevant data, 11 articles showing improper patients, 7 studies showing improper control groups, 2 studies without descriptions of acupoints, and 6 articles that were protocols. Finally, we included 26 studies . A flowchart of the literature screening process is outlined in Fig 1.

Description of the included studies
Twenty-six studies  met our inclusion criteria and included work performed in 8 countries (United States of America, The United Kingdom, China, India, Malaysia, Turkey, Iran, and Denmark) with a total of 2,064 patients. Among them, 1,051 patients received EAS, and 1,013 patients served as controls. The surgical type included laparoscopy, lithotripsy, tonsillectomy, thyroidectomy, gynecological, thoracotomy, sinusotomy, craniotomy, and plastic surgery. Among the included RCTs that used EAS as an intervention measure, 8 trials [46-50, 54, 56, 57] including 545 patients performed EA and 18 trials [40-45, 51-53, 55, 58-65] including

Risk of bias within studies
All of the included trials mentioned randomization but 2 articles [43,44] failed to report the method used to generate random sequences and were rated as unclear risk. For allocation concealment, 7 studies [48,54,57,[60][61][62][63] used a sealed envelope and 2 studies [52,53] used central allocation and were considered low risk, whereas the others did not mention allocation concealment and were rated as unclear risk. For blinding, 16 studies [40-47, 49-51, 55, 56, 58, 59, 64] provided insufficient information or used methods that could allow the patients to be aware of their assigned groups and were rated as unclear risk. For blinding of the outcome measurers, 5 studies [43,53,57,64,65] were rated as unclear risk, 3 [53,57,64] of which were not blinded. Considering the specificity of EAS treatment, blinding might not be done for operators. The remaining 2 studies [43,65] did not describe blinding in their design. In addition, due to incomplete outcome data reporting, 6 studies [43,44,48,50,56,64] did not describe the loss rate of the follow-up population, and 2 studies [42,55] did not describe the reasons for the loss in the follow-up. These were, therefore, classified as unclear risk. One study [41] was judged as high risk due to later changes in random enrollment personnel. For selective reporting, 1 trial [47] did not describe adverse events of therapy, but when asked, the authors indicated that no obvious adverse effects were noted during the study. In terms of other biases, for the purpose of treatment, 3 studies [41,42,45] were rated as high risk. Among them, 1 article [42] was rated as high risk due to the differences between the groups. Another article [45] was rated as high risk due to the inconsistency of the data between figures and text. The last article [41] was rated as high risk due to the late supplementation of patients (see S1 Fig).

Incidence of PONV
Sixteen of the studies [40, 41, 45-49, 52, 53, 55, 56, 58, 59, 61, 62, 65] included 1372 patients and measured and recorded the results of PONV after EAS. A fixed-effects model was used (P = 0.48; I 2 = 0%). The results indicated that participants who received EAS exhibited a significantly lower incidence of PONV than those in the control group (RR, 0.49; 95% CI, 0.41 to 0.57; P < 0.001) (Fig 2A). Sensitivity analysis of the outcomes showed that the heterogeneity was not significantly lower after any study was excluded (S3 Table).
The RCTs that recorded the incidence of PONV included several types of surgical procedures. To clarify the impact of the surgical procedure, we performed a subgroup analysis. In 25% (4/16) of the included studies, the control patients did not receive placebo targeting acupuncture points or electrical stimulation. Thus, we conducted a subgroup analysis. The results showed that with placebo (no acupuncture + active device, no acupuncture + inactive device, or gel electrodes + inactive device) (RR, 0.48; 95% CI, 0.40 to 0.57; P < 0.001; I 2 = 12%) or without placebo (usual care only) (RR, 0.53; 95% CI, 0.33 to 0.84; P = 0.007; I 2 = 0%), EAS reduced the incidence of PONV (P < 0.001, S2B Fig).
PONV within 24 h contained the maximum number of studies [40, 41, 45-47, 49, 56, 58, 59, 61, 62, 65]. We further used these studies to perform subgroup analysis based on different  Fig 3A). However, there was no significant difference between the two types of treatment (RR, 0.51; 95% CI, 0.43 to 0.60; P = 0.13). According to the starting and ending times of the first intervention, the studies were divided into three subgroups: preoperative intervention, postoperative intervention, and
When the trial by zheng [64] was excluded, the heterogeneity decreased to 7%, with no change in the results. The remaining trials revealed that EAS could reduce PON compared to the control group. The sources of heterogeneity that remained may have been due to the different types of surgery and durations of the interventions (S3 Table).
Most of the heterogeneity was caused by one study [47]. After exclusion, heterogeneity decreased to I 2 = 33%, which may have been due to the type of surgery, the method, and the duration of the intervention (S3 Table).

Publication bias
No obvious release bias was found through the inverted funnel chart (S5 Fig).

PLOS ONE
Efficacy of electrical acupoint stimulation on postoperative nausea and vomiting: Meta-analysis Sensitivity analysis S3

GRADE evaluation
Based on the principles of the GRADE evaluation, we evaluated the quality of the evidence provided via the PONV, 24 h PONV, PON, POV, rescue antiemetic, and adverse events assessments. Table 2 shows that, except for rescue antiemetic assessments, which were classified as low-quality, the others were evaluated as moderate in quality.

Discussion
The results of this meta-analysis indicate that EAS significantly decreases the incidence of PONV, PON, and POV. Our results were consistent with the previous studies. A meta-analysis first published in 2004 was updated in 2015 and found that the effect of acupoint stimulation is comparable to antiemetics in preventing PONV [23]. But, this research only focused on evaluating the effect of PC6. Another meta-analysis indicated that additional effective meridian points in the treatment of PONV included BL10-11, BL18-26, SP4, SP6, ST34, ST36, ST44, and others [66], and thus PC6 may not be the only acupoint effective in treating PONV. We

PLOS ONE
Efficacy of electrical acupoint stimulation on postoperative nausea and vomiting: Meta-analysis included studies looking at many acupuncture points, which are more comprehensive and practical for studying the efficacy of EAS. The acupoints should be selected for each patient according to their presenting symptoms and characteristics.
Subgroup analysis showed that EAS, whether applied as a preoperative intervention, postoperative intervention, or perioperative intervention, had a significant effect. To the best of our knowledge, no meta-analysis has been conducted on this topic. However, one RCT showed that EA on PC6 is effective in the prevention of PONV, and pre-operative acupuncture is more effective than post-operative acupuncture. There are two possible reasons for the disparity in results. Firstly, meta-analysis can only show that the three types of treatment timing are effective because they were not directly compared in the included studies by us. Secondly, the previous meta-analysis indicated that the antiemetic effect of acupuncture require treatment of awake rather than anesthetized patients [67]. Third, during surgery, the needle cannot be well protected, and the overall effect may be minimized. For safety reasons, it is ideal to provide acupuncture to conscious patients before surgery because most of the acupuncture points are located around the nerves, and deep punctures on unconscious patients may damage nerves.
Another subgroup analysis demonstrated that both EA and TEAS had good therapeutic effects compared with the control group. As far as we know, no similar study comparing the efficacy of EA and TEAS for PONV. However, a meta-analysis indicated that acupuncture therapy can reduce the risk of PONV in abdominal operation. Another meta-analysis suggested that TEAS was effective in preventing PONV [68]. Those results were consistent with Note 1 11 studies were rated as unclear risk in allocation concealment; 2 studies were rated as hight-risk and 8 were rated as unclear in terms of other biases 2 Funnel plot showed publication bias 3 10 studies were rated as unclear risk in allocation concealment; 2 studies were rated as hight-risk and 5 were rated as unclear in terms of other biases 4 2 studies were rated as unclear risk in random sequence generate; 8 studies were rated as unclear risk in allocation concealment; 4 studies applied a single-blind study design in blinding; 1 study was rated as hight-risk and 5 were rated as unclear in terms of other biases 5 2 studies were rated as unclear risk in random sequence generate; 7 studies were rated as unclear risk in allocation concealment; 4 studies applied a single-blind study design in blinding; 1 study was rated as hight-risk and 4 were rated as unclear in terms of other biases 6 1 study was rated as unclear risk in random sequence generate; 5 studies were rated as unclear risk in allocation concealment; 2 studies applied a single-blind study design in blinding; 2 study were rated as hight-risk and 2 were rated as unclear in terms of other biases 7 I square = 51% 8 1 study was rated as unclear risk in random sequence generate; all studies were rated as unclear risk in allocation concealment; 2 studies applied a single-blind study design in blinding; 2 studies were rated as hight-risk and 3 were rated as unclear in terms of other biases. https://doi.org/10.1371/journal.pone.0285943.t002 the results obtained in our study. The addition of EAS treatment can reduce the workload of doctors, allowing doctors to treat multiple patients at once, and mass production also makes the parameters used in treatment more accurate. EAS is divided into EA and TEAS. EA combines a pulsating electrical current with acupuncture to enhance acupoint stimulation, which is a more effective method for administering acupuncture [28]. The advantages of EA are that it preserves the therapeutic effect of traditional acupuncture based on increasing EAS and combines electro with physical stimulation generated by acupuncture at acupuncture points. However, these advantages also mean that this process of acupuncture requires professional acupuncturists to only use invasive needles. In addition, some patients who are sensitive to needles may experience syncope during treatment and may have complications, such as subcutaneous bleeding, persistent acid bloating, and a burning sensation at the acupuncture site. TEAS has advantages: there is no intrusive behavior and patients' acceptance is higher. TEAS also has disadvantages: only electrical stimulation can be applied and there is a lack of physical stimulation of needle penetration into the skin. Additionally, TEAS may cause minor allergies, such as skin flushing. Compared with EA, TEAS treatment is safer, more effective, and worthy of clinical promotion. We have found that for laparoscopy, thyroidectomy, and tonsillectomy, EAS reduced the incidence of PONV. Similar findings have also been reported. Studies on the preventive and therapeutic effects of acupuncture on PONV for specific types of surgeries, and the findings all indicate that acupuncture is an effective and safe therapy for PONV [69,70].
In our meta-analysis, we included more patients and RCTs. In addition to the inclusion of TEAS, RCTs evaluating the effectiveness of EA in the treatment of PONV were also added. Thus, our study provides a more comprehensive analysis than the 2020 TEAS meta-analysis [68]. Furthermore, we used the Quality Evaluation Tool approach to rate the quality of evidence to provide more reliable conclusions. In addition, we conducted a subgroup analysis to compare the pros and cons of EA and TEAS and distinguish the role of EAS in different types of procedures.
The quality of the RCTs included in this review was assessed by the GRADE system [35]. GRADE clearly differentiates the quality of each RCT from the overall strength of the evidence, making the accuracy of the analysis results clearer [71,72]. Grading the strength of the evidence using the GRADE approach is becoming an important, recommended step in a comprehensive evidence evaluation and could increase the transparency of the clinical decisionmaking processes, especially when the quality of the evidence is poor or unclear [71,73]. AMSTAR is a measurement tool created to assess the methodological quality of systematic reviews [74]. It is a reliable, valid, and critical assessment tool developed by AMSTAR in 2017 [75]. According to the GRADE criteria and AMSTAR 2 system, the quality of evidence was moderate for all outcomes, except the rescue antiemetic requirement (low-quality evidence). Therefore, the results lower our confidence in clinical decisions. Fortunately, the methodological quality was high. Incorporating all the evaluation results reported in the study, some studies may have a certain degree of bias. However, no obvious release bias was found through the inverted funnel chart.
Moreover, regarding adverse events, there was no significant difference between the EAS group and the control group. This result indicated that EAS is not related to any serious adverse events, although some patients may present minor hemorrhage or pain at the insertion site of the needle, or redness and itching may appear on the skin where the surface electrode was applied. The safety data available were limited, as 14 (53.85%) studies did not report any side effects of EAS.
Apfel and his colleagues developed a data-based assessment tool to predict the risk of PONV [76]. The tool assigns one point for each known risk factor (gender, smoking status, history of PONV or motion sickness, and use of opioids for postoperative pain) [77]. They found that the presence of 1, 2, 3, and 4 of these risk factors increased the incidence of PONV by 21%, 39%, 61%, and 79%, respectively [76]. The tool classifies patients with 0-1, 2, or 3-plus risk factors into "low," "medium," and "high" risk categories, respectively. In 2020, the Fourth Consensus Guidelines for the Management of PONV concluded that "multimodal PONV prophylaxis in patients with 1 or 2 risk factors, in an attempt to reduce risk of inadequate prophylaxis" [14]. Two antiemetics are recommended for PONV prophylaxis in patients at medium risk and 3-4 antiemetics in patients at high risk [14]. In order to evaluate whether EAS is useful for patients with "low," "medium," or "high" risk of PONV, RCTs in which participants were stratified by Apfel score should be done.However, only one article mentioned the Apfel scoring system during patient inclusion among the included studies. It specifies the inclusion of patients with an Apfel score � 2 including patients with a medium-high risk for PONV [40]. While none of the other 25 studies had that inclusion criterion. Thus, we could not conduct a subgroup analysis based on Apfel scores. We agree that the Apfel scoring system is very important. If these RCTs included Apfel scores in their inclusion criteria or stratified randomized groups of patients based on the Apfel scoring system, higher-quality results would have been obtained.
Opioids have been identified as an independent risk factor for the development of PONV [77]. The mechanism of action of EAS may involve the regulation of endogenous opioid release and other neurotransmitters [78]. Evidence has shown that EA treatment might lead to better analgesia and reduce opioid use [79]. This is a possible mechanism of how EAS reduces PONV.
We excluded patients who underwent cesarean section or abortion, because the abortion patients may have had pregnancy vomiting. Similarly, cesarean patients might be administered oxytocin. Oxytocin causes emesis. We only included patients who underwent general anesthesia, whereas most cesarean patients do not use this type of anesthesia.
Our review has several limitations. First, most of the included studies had small sample sizes, and the event rates of several outcomes were low. This limitation may lead to imprecise evidence. Additionally, using GRADE, we judged the quality of the evidence in the review to be only moderate or low. The GRADE evidence quality of the pooled results lowers our confidence in the utility of the evidence to guide clinical decisions. Third, some design flaws are obvious. For example, there are many types of surgery included in the literature. The results of the analysis of all surgical studies are more likely to lead to bias. And some results that were significant for the evaluation of PONV were not included (the frequency of PONV). Cause only one of the 26 studies included in our meta-analysis recorded the frequency of PONV. Fourth, no clear comparison with already recommended pharmacological treatment group. Fifth, no obvious risk assessment or use of the scoring system to stratify the included studies. In addition, a certain degree of heterogeneity was observed in this meta-analysis. We tried to reduce heterogeneity through subgroup and sensitivity analyses, but it has not yet been completely resolved.

Conclusion
Our analysis showed that EAS can decrease the incidence of PONV, PON, and POV, as well as the number of patients requiring antiemetic rescue, and does not increase adverse events in patients undergoing elective surgery under general anesthesia. According to the GRADE criteria, the quality of evidence was moderate for all outcomes, except the rescue antiemetic requirement (low-quality evidence). These findings suggest that EAS may be considered an effective and safe treatment for PONV and that the EAS approach may be promising to promote the recovery of patients after surgery. The reliability of these results for PONV needs to be further explored, and the quantity and quality of the included studies need to be improved.