Acupuncture-Point Stimulation for Postoperative Pain Control: A Systematic Review and Meta-Analysis of Randomized Controlled Trials

The purpose of this study was to evaluate the effectiveness of Acupuncture-point stimulation (APS) in postoperative pain control compared with sham/placebo acupuncture or standard treatments (usual care or no treatment). Only randomized controlled trials (RCTs) were included. Meta-analysis results indicated that APS interventions improved VAS scores significantly and also reduced total morphine consumption. No serious APS-related adverse effects (AEs) were reported. There is Level I evidence for the effectiveness of body points plaster therapy and Level II evidence for body points electroacupuncture (EA), body points acupressure, body points APS for abdominal surgery patients, auricular points seed embedding, manual auricular acupuncture, and auricular EA. We obtained Level III evidence for body points APS in patients who underwent cardiac surgery and cesarean section and for auricular-point stimulation in patients who underwent abdominal surgery. There is insufficient evidence to conclude that APS is an effective postoperative pain therapy in surgical patients, although the evidence does support the conclusion that APS can reduce analgesic requirements without AEs. The best level of evidence was not adequate in most subgroups. Some limitations of this study may have affected the results, possibly leading to an overestimation of APS effects.


Introduction
Nearly 86% of surgery patients experience moderate to severe postoperative pain [1]. Depending on surgery type, as many as half of these patients go on to experience chronic postoperative pain [2]. Unsatisfactory pain control can limit patients' physical activities, prolong recovery time, and contribute to poor quality of life [3,4]. Pain may also increase postoperative complications, such as postoperative morbidity, and may extend the length of hospitalization, increasing health care costs [5,6].
Administration of standard analgesics, which are considered generally safe and effective, remains the primary approach to postoperative pain management [7]. However, systemic analgesic administration can induce some adverse effects (AEs), such as nausea, vomiting, depressive symptoms, pruritus, urinary retention, gastrointestinal motility, and ileus [8,9]. AEs can impair physical and psychological wellbeing and, more seriously, may result in significant morbidity or even mortality [8][9][10]. To achieve better postoperative pain relief and reduce the requirement for analgesic medication, various nonpharmacological approaches, including educational intervention, relaxation, and acupuncture-point stimulation (APS), have been employed. APS has been lauded as a promising alternative method for achieving postoperative pain relief [11,12].
APS is a widely used component of traditional Chinese medicine (TCM) together with full-body and auricular approaches [11]. In addition to the most popular methods of manual acupuncture and acupressure, APS can be achieved using modalities such as electrical acupuncture (EA) or laser stimulation and acupoint massage [12]. According to TCM  philosophy, the stimulation of target acupoints along meridians produces positive effects by rebalancing qi circulation in the body [13]. However, the existence of the qi meridian system, as described in TCM, has never been demonstrated scientifically [13]. Nevertheless, the management of various forms of pain remains a key purported benefit of APS [14,15].
Many animal experiments and clinical studies have examined the therapeutic effects of APS [16]. Early studies showed that APS provided postoperative pain relief in comparison with control groups [17,18]. Recently, several small trials [3,19] demonstrated that APS can relieve pain and reduce analgesic requirements associated with hysterectomy and inguinal surgery. However, Sakurai et al. [20] failed to identify any significant change in pain intensity or morphine requirement in surgical patients undergoing acupressure. A prior systematic review found that acupuncture and related techniques aided postoperative pain control, but the quality of evidence was low due to the quality and quantity of included trials, and no subgroup analysis according to acupuncture type was performed [21]. The results of another systematic review conducted by Usichenko et al. [22] suggested that auricular acupuncture was a promising method of postoperative pain reduction, but the heterogeneity of primary studies precluded data synthesis and the evidence was insufficient to draw a definitive conclusion about the treatment's effectiveness. Following the 2008 publication of these reviews, several clinical trials were conducted to evaluate the efficacy of APS for postoperative pain management, generating new evidence on this topic [23][24][25].
The present systematic review and meta-analysis was conducted to evaluate the effectiveness of APS for pain control following surgical procedures. Therefore, in this study, current evidence generated by randomized controlled trials (RCTs) on the use of APS interventions for postoperative pain management was reviewed and analyzed. Data from patients receiving APS were compared with those from control groups receiving sham/placebo acupuncture, usual care, or no treatment. Compared with the previous literature, this systematic review and meta-analysis incorporates new evidence not previously synthesized and distinguishes between multiple types of APS for postoperative pain control. Figure 1 We also used PubMed's "related articles" function to identify additional potentially relevant studies. The electronic search had no language restriction. In the case that there were multiple publications from the same RCT, overlapping results were extracted from one publication.

Study Selection. As summarized in
Two reviewers assessed all potentially relevant articles independently. Disagreements regarding study selection were resolved by discussion, with strict adherence to the inclusion criteria. Studies were selected for inclusion based on the following criteria: (1) RCT; (2) adult (age ≥ 18 years) participants with pain following any surgical procedure; (3) APS intervention (including full-body or auricular manual acupuncture or EA, acupressure, seed embedding, and plaster therapy) conducted by an acupuncturist, TCM practitioner, or other health care providers with qualification and/or training in acupuncture therapy; (4) control group receiving standard treatment (e.g., active pain control approach normally provided to surgical patients, including analgesia medication, nursing guidance, and other usual cares), sham/placebo APS (faked APS intervention), or no treatment (provision of usual postoperative care not involving active analgesic interventions); (5) primary outcome of pain intensity, measured by a valid self-reported instrument such as a visual analog scale (VAS), numerical rating scale (NRS), or verbal reporting; and (6) secondary outcomes of analgesic consumption and APS-related AEs (i.e., any adverse events resulting from APS intervention, minor (e.g., needling site pain), intermediate (e.g., bleeding and hematoma), or serious (pneumothorax and cardiac tamponade)).
In the study selection process, acupuncture was defined as the stimulation of specific acupuncture points along the skin of the body by using thin needles, with or without the application of heat, pressure, or laser light to these same points [12]. EA is similar to acupuncture but involves the use of devices (e.g., a wristwatch-like device and surface electrodes attached to a transcutaneous electrical nerve stimulation device) on acupoint [14]. Seed embedding was defined as an auricular acupressure process involving the embedding of magnetic beads or other seeds within skin-colored adhesive tape, which is placed on the auricular acupoints and retained in situ for several days [6]. In this systematic review, plaster therapy mainly referred to the use of capsicum plaster as an alternative to acupuncture [23].

Quality Assessment.
Two reviewers conducted independent assessments of the methodological quality and risk of bias of each RCT using Cochrane Collaboration's risk of bias tool [26]. This tool provides for the assessment of seven domains: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and "other issues." Items were scored as showing low, high, or unclear risk of bias [26]. All disagreements on scoring were resolved by discussion. When a sufficient number of studies were available and a meaningful assessment of publication bias could be carried out, a funnel plot was constructed.
Adequate allocation concealment and blinding of outcome assessors were designated as key domains for this assessment, where key domains are not only more likely to influence bias magnitude and direction but also more likely to impact study results. Domain-based evaluation was employed as described in the Cochrane Handbook for Systematic Reviews of Interventions 5.0.2 (updated September 2009). The overall risk of bias was categorized as follows. An overall low risk of bias (plausible bias unlikely to alter the results) was identified when all key domains were assessed as having a low risk of bias. An overall unclear risk of bias (plausible bias that raises some doubt about the results) was identified when one or more key domains were assessed as having an unclear risk of bias. An overall high risk of bias (plausible bias that seriously weakens confidence in the results) was identified when one or more key domains were assessed as having a high risk of bias. Small studies have been shown to overestimate treatment effects and to be at increased risk of bias, allowing critical criteria such as blinding to be compromised. Studies were considered to be at low risk of bias if they had at least 200 participants, at unknown risk if they had 50 to 200 participants, and at high risk if they had fewer than 50 participants.

Data Extraction and Management.
Two reviewers independently extracted the following data from included studies using a predefined form: authors, study design, setting, population and participant demographics, intervention characteristics (e.g., acupuncture type, acupoints used, and treatment duration), comparators, outcome measures and instruments, follow-up, and some numeric data (mainly the results of pain intensity and analgesic consumption). We contacted RCT authors by email to obtain data necessary for effect size estimation when such data were missing from publications (e.g., due to aggregated data reporting). When authors did not reply, outcome data presented only in figures and/or graphs were extracted when possible; these data were included in the analysis only when the two reviewers independently obtained the same results.
When a study reported multiple group comparisons (e.g., high electrical stimulation versus low electrical stimulation or usual care and preoperative acupuncture versus postoperative acupuncture or usual care), only data from the treatment group that received the more intensive and comprehensive postoperative intervention were included in the analysis. These data were compared with those from the control group.

Subgroup Analysis.
When data were sufficient, subgroup analyses of different types of APS, surgery, and control groups were conducted. Analyses of APS type compared the use of acupoints on the body (EA, manual acupuncture, acupressure, and plaster therapy) and/or auricular points (EA, manual acupuncture, and seed embedding). A subgroup analysis of EA studies compared the use of different devices (e.g., a wristwatch-like device and surface electrodes attached to a transcutaneous electrical nerve stimulation device). Analyses of control group types compared APS with standard treatment (usual care and no treatment) or placebo/sham therapies. On the basis of whether body acupoints, auricular 4 Evidence-Based Complementary and Alternative Medicine In the presence of significant heterogeneity ( 2 test, < 0.1), random-effects model was used. Otherwise, fixed-effects model was applied. Descriptive analysis was used when data could not be converted or pooled. Potential sources of heterogeneity in the outcomes examined are differences in the tool used to measure pain, population differences (e.g., surgery type, age, and sex), and differences in the comparator used (e.g., sham/placebo acupuncture, usual care, and no treatment). We assessed heterogeneity using the 2 statistic, which describes the percentage of total variation across trials (low, 0-40%; moderate, 30-60%; substantial, 50-90%; considerable, 75-100%; Chapter 9: Analysing Data and Undertaking Meta-Analyses; The Cochrane Collaboration 2011, available from http://www .cochrane-handbook.org/). To assess which RCTs affected the overall results, sensitivity analyses were performed for the entire sample and subgroups with significant heterogeneity. If heterogeneity was considerable, even with the random-effects model, best evidence synthesis was also used. The evidence was synthesized based on each subgroup. We employed a qualitative modified approach to grading of evidence, as summarized in Table 1.  Tables 2-4). Nine publications were in Chinese, and the remaining 50 publications were in English. The included studies were conducted in mainland China, Hong Kong, Taiwan, the United States, Germany, Austria, South Korea, Japan, Iran, the United Kingdom, Brazil, Sweden, Singapore, Italy, and Turkey. Sixteen of these RCTs had three arms and three trials had four arms.

Characteristics of Included
RCTs included in the analysis involved a total of 4,578 randomized patients, 4,402 of whom completed the respective studies (APS groups, = 2, 097; control groups, = 2, 305; 96.16% completion rate). The average sample size was 73 (range, . Standardized anesthetic and postoperative analgesia regimens were used in all studies. Follow-up duration ranged widely from 7 days [60,61] to 4 months [41]. Twenty-one RCTs stated that an intention-to-treat analysis was used.
Five types of APS were used: low-and/or high-intensity EA, manual needle acupuncture, seed embedding, acupressure, and plaster therapy. Preoperative and postoperative APS were used in two RCTs, and sham/placebo control was used in 36 studies. Chinese herbs were used as a control in a single study.

Methodological Quality and Risk of Bias of Included
Trials. The methodological quality of included studies is characterized in Figures 2 and 3. Eighteen (30.5%) publications specifically stated that the outcome assessor was blinded and 23 (38.98%) studies used adequate allocation concealment together with full methodological description. Twelve (20.33%) studies were rated highly in both of these domains and were deemed to be at low risk of overall bias. Forty-two studies (71.19%) were deemed to be at unclear risk of overall bias. Finally, five studies (8.5%) were deemed to be at high overall risk of bias. Allocation concealment was not reported or was described poorly in 24 (40.67%) studies. According to the number of participants, no studies were considered to be at low risk of bias (≥200 participants), 42 studies (71.19%) were at unknown risk of bias (50-200 participants), and 17 studies (28.81%) were at high risk of bias Evidence-Based Complementary and Alternative Medicine 5         14 Evidence-Based Complementary and Alternative Medicine   (<50 participants). Visual inspection of funnel plots revealed some substantial asymmetry in comparisons ( Figure 4).

Postoperative Pain.
The results of the meta-analysis are reported in Table 5 [53,74] that used NRS scores reported a significant difference between groups, but these data could not be included in the meta-analysis due to clinical and statistical heterogeneity. One of two RCTs [34,75] that used BPI scores reported a significant difference between the intervention and control groups, but these data could not be included in the meta-analysis due to different data modes. Three trials [58,60,61] using a four-point scale reported that the intervention reduced pain intensity, but one of these studies [58] reported that total or partial pain relief did not differ significantly between the groups. Two trials [60,61] reported that the APS intervention increased the duration of postoperative painfree status compared with that of the control groups. Thirty-eight RCTs used body points for stimulation. Subgroup analyses according to control treatment and meta-analysis of 20 RCTs indicated that APS interventions improved VAS scores significantly in comparison with standard treatment and sham/placebo control (both < 0.00001; Table 5). Similarly, pooled results from 24 trials showed that body APS significantly improved VAS scores in comparison with all control groups ( < 0.00001), and subgroup analyses revealed similar improvement compared with standard treatment ( < 0.00001) and sham/placebo control ( < 0.0001; Table 5). The evidence for body points APS reducing postoperative pain intensity in surgery patients was determined to be of Level I quality based on six overall high quality RCTs [14,23,47,49,52,58]. A meta-analysis of pooled results and subgroup analyses of body EA, as well as invasive and noninvasive forms of this treatment, yielded similar results ( Table 5). The evidence for body points EA reducing postoperative pain intensity in surgery patients was determined to be of Level II quality based on one overall high quality RCT [14]. High-frequency EA was found to be more effective than low-frequency EA [42,45]. Pooled results from three RCTs examining acupressure [48,50,51] showed a significant difference in VAS scores between intervention and control groups ( = 0.01; Table 5), although a fourth study [20] not included in the meta-analysis showed no such difference. The evidence for body points acupressure reducing postoperative pain intensity in surgery patients was determined to be of Level II quality based on three moderate quality RCTs [20,48,50] and one low quality RCT [51]. Synthesis of data from two RCTs [23,47] examining plaster therapy showed a significant reduction in pain intensity compared with standard treatment ( < 0.00001) and sham controls ( < 0.0001; Table 5), and one other study [49] examining this treatment obtained similar results. The evidence for body points plaster therapy reducing postoperative pain intensity in surgery patients was determined to be of Level I quality based on three overall high quality RCTs [23,47,49].
In contrast, meta-analysis including three studies [55][56][57] revealed no significant effect of manual acupuncture on VAS score. Four [55,57,59,60] of 10 RCTs examining manual acupuncture reported no difference in pain score between the intervention and control groups, whereas the remaining six studies found that this treatment reduced postoperative pain intensity ( < 0.05).
Twelve RCTs used body point stimulation for patients with abdominal surgery. The pooled results from eight trials [29,35,40,42,43,45,47,51] showed that body APS significantly improved VAS scores in these patients ( = 0.0006). The evidence for body points APS reducing postoperative pain intensity in patients who had undergone abdominal surgery was determined to be of Level II quality based on one overall high quality RCT [47]. Five [31,32,50,57,59] RCTs used body point stimulation for patients with knee surgery. Pooled results from four trials [31,32,50,57] showed that body APS did not significantly improve VAS scores for these patients ( = 0.16). Each of two RCTs used body point stimulation for patients with oral surgery [60,61], cardiac surgery [33,37], hemorrhoid operation [30,53], or cesarean section [48,55]. Pooled results from two trials showed that body APS significantly improved VAS scores for patients undergoing cardiac surgery [33,37] ( = 0.002) or cesarean section [48,55] ( < 0.00001). The evidence for body points APS reducing pain intensity in patients who underwent cardiac surgery and cesarean section surgery was determined to be of Level III quality based on two moderate quality RCTs [33,37]. Other studies could not be included in the meta-analyses due to insufficient data and the different types of surgery.
Fourteen [6,19,24,[62][63][64][65][66][67][68][69][70][71][72] RCTs used auricular points for stimulation. Data synthesis from 12 studies showed significantly lower VAS scores in intervention groups than in all types of control group ( = 0.001), and similar results were obtained in comparison with standard treatment ( = 0.04) and sham/placebo control ( = 0.02) groups ( Table 5). The evidence for auricular points APS reducing postoperative pain intensity was determined to be of Level I quality based on six overall high quality RCTs [6,24,68,69,71]. Metaanalysis of data from five studies [6,24,62,64,65] examining seed embedding also showed a significant effect on VAS score in comparison with all control groups ( = 0.02; Table 5). The evidence for auricular points seed embedding reducing postoperative pain intensity was determined to be of Level II quality based on two overall high quality RCTs [6,24]. Two studies [6,65] of this auricular APS technique found a gradual reduction in pain, but no significant difference, according to VAS and Short-Form McGill Pain Questionnaire scores. One study [63] of manual auricular acupuncture data reported a significant difference in VAS score, and another study [72] showed a positive trend toward less pain in the intervention group, but meta-analysis of data from four studies [63,68,70,71] showed that this auricular APS type was not associated with significant pain reduction. The evidence for manual auricular acupuncture reducing postoperative pain intensity in surgery patients was determined to be of Level II quality based on two overall high quality RCTs [68,71]. Metaanalysis of auricular EA data from three studies [19,66,69] showed a significant reduction in VAS scores (including those reflecting pain at rest and on huffing and coughing; < 0.0001), although two of the four RCTs examining this treatment found no significant difference due to low pain intensity in intervention groups. The evidence for auricular EA reducing postoperative pain intensity in surgery patients was determined to be of Level II quality based on one overall high quality RCT [69].
Five [6,24,63,68,70] RCTs used auricular point stimulation for patients with knee surgery. Pooled results from five trials [6,24,63,68,70] showed that auricular point APS did not significantly improve VAS scores for these patients ( = 0.20). Two [19,62] RCTs used auricular point stimulation for patients with abdominal surgery. Pooled results from both trials [19,62] showed that auricular point APS significantly improved VAS scores for these patients ( = 0.01). The evidence for auricular point stimulation reducing postoperative pain intensity in abdominal surgery patients was determined to be of Level III quality based on two moderate quality RCTs [19,62].

Analgesic Requirement.
Forty-three RCTs measured analgesic use, and most studies documented a lesser analgesic requirement in APS intervention groups than in control groups. Meta-analysis of data from six RCTs [3,25,34,42,43,55] showed a significant reduction in total morphine consumption in intervention groups compared to the control groups ( = 0.0001). Similar results were obtained in the comparison of intervention and sham/placebo control groups ( < 0.00001; Table 5). In addition, Lin et al. [42] reported that the morphine requirement after high-frequency EA was decreased by 31% compared with that after low-frequency EA. The evidence for APS reducing analgesic requirement in surgery patients was determined to be of Level I quality based on multiple overall high quality RCTs.

AEs.
No serious AEs were associated with APS, and patients were reported to tolerate the intervention well in    the 21 RCTs that reported on this outcome. Reported minor AEs included temporary increased pain [74], localized pain or discomfort at insertion sites [60,71], minor bruising or bleeding [74], constitutional symptoms [74], and a mild burning sensation with erythema [23,47,49]. Michalek-Sauberer et al. [67] stated that 38% of patients reported minimal side effects of acupuncture, most commonly fatigue (16%) and ear pain (10%).

Sensitivity and Heterogeneity.
Given the detection of obvious heterogeneity ( 2 > 50%) in meta-analyses, we conducted a sensitivity analysis to remove studies with a greater risk of bias. The results are presented in Table 6. 2 values were decreased substantially by the removal of such trials in most comparisons.

Discussion
In this review, it was determined that there is insufficient evidence thus far to conclude that APS is an effective nonpharmacological approach to the reduction of postoperative pain intensity for surgery patients, although the evidence did show a reduced analgesic requirement with no significant adverse effects in surgery patients. The results may have been affected by some limitations of this study, such as the wide variability of interventions and participants, absence of follow-up evaluation in most included trials, and the often mediocre methodological quality of the included studies. These factors contributed to the high heterogeneity of the data, which limits the strength of the evidence. No studies were considered to be at low risk of bias (≥200 participants) based on the number of participants. These factors may have led to overestimations of APS efficacy. Given the intensity of surgical trauma, postoperative pain is inevitable and it is deemed to be a serious problem. If this pain is not managed effectively, it can contribute to several clinical risks and affect patients' physical and psychological wellbeing; potential effects include emotional distress, infection, increased myocardial oxygen consumption, and prolonged hospitalization. Associated pathological changes can harm organs and lead to abnormal function [31,33]. Reduction of postoperative pain is therefore essential.
Our meta-analysis of overall effects from 39 trials showed that interventions involving stimulation of body or auricular points significantly reduced postoperative pain, as measured by VAS scores. Data from studies using integrative APS or manual acupuncture showed uncertain outcomes or no significant change. In one of these studies, Deng et al. [75] suggested that these results may be due to the insufficient strength of APS to produce analgesic effects.
Among body APS studies, the largest subgroup analyzed, all intervention types except manual acupuncture significantly reduced postoperative pain. The precise analgesic mechanism of body APS remains unclear. However, it has been found to facilitate central nervous system release of met-enkephalin and dynorphins into the spinal fluid, causing synergistic pain relief with exogenous opioid medication and production of pain-producing substances, such as potassium and lactic acid [31,34,39,79]. The finding that highfrequency EA at body points was more effective than lowfrequency EA may be due to differences in opioid peptide release [33].
Similarly, auricular APS therapies were found to significantly reduce postoperative pain, with the exception of manual acupuncture. The most commonly used auricular point is Shen Men, which generates analgesic, sedative, and anti-inflammatory effects [6]. It also increases endorphin secretion and serotonin production, thereby suppressing the transmission of pain messages and thus pain perception [80]. The results for integrative (auricular and body) APS are less clear; this treatment was found to significantly reduce NRS and VAS scores. Thus, the existing evidence neither supports nor refutes the effectiveness of integrative APS for postoperative pain control.
We also undertook subgroup analyses of surgery types, including abdominal, knee, oral, cesarean, and cardiac surgeries. The meta-analysis results showed that body point acupuncture stimulation and auricular therapy had no significant change on VAS scores for patients undergoing knee surgery. The same trend was observed for patients receiving integrative acupoint stimulation and undergoing oral surgery. Short-term APS stimulation may have been insufficient to reduce patients' pain intensity after knee or oral surgery, or the postoperative rehabilitation program may have affected the results of APS interventions [57]. Rigorously designed large-scale RCTs are needed to identify the effects of APS for these kinds of patients.
This analysis also showed that APS significantly reduces patients' postoperative analgesic requirement. Given the dose-response relationship between analgesics and related adverse effects [81], any nonpharmacological method that reduces the use of analgesic medication is likely to be beneficial. Lin et al. 's [42] finding of reduced morphine requirement after high-frequency EA compared with that after lowfrequency EA demonstrates the existence of a dose-response relationship in this treatment as well. However, analgesic requirements are controlled by the health care staff and directly affected by the surgery type and patient's economic condition. Thus, analgesic medication use is not a particularly reliable indicator for the effects of APS.
No APS study reported the occurrence of a serious adverse event, although some minor (mild and transient) side effects were reported. To prevent such effects, APS should be carried out by experienced, well-trained health care professionals who understand the theories underlying this therapy and take necessary precautions.
APS may produce strong placebo effects; for example, sham acupuncture did not affect analgesic-related side effects but did exert a moderate pain-relieving effect [42]. The use of sham/placebo control groups, as in 36 of the examined RCTs, enables clear distinction between true and placebo effects. This meta-analysis showed that the true effects of APS were much stronger than placebo effects. Short-term APS and placebo interventions have shown similar effects, but long-term APS treatment causes beneficial changes in specific brain areas [82].
A small sample size can distort the results of metaanalyses, by overestimating treatment effects, probably due to methodological weaknesses [83]. In our review, no studies were considered to be at low risk of bias (≥200 participants) on the basis of sample size. Forty-two studies (71.19%) were at an unknown risk of bias (50-200 participants), and 17 studies (28.81%) were at a high risk of bias (<50 participants).
In this review, statistical heterogeneity was considerable, even with use of the random-effects model. The best level of evidence was not found for most forms of APS, suggesting that there is, thus far, insufficient evidence to conclude that APS is an effective method for reducing pain intensity in postoperative patients. Within the available body of evidence, there is Level I evidence supporting the effectiveness of body points plaster therapy. Additionally, there is Level II evidence supporting the use of body points EA, body points acupressure, and body points APS in abdominal surgery patients specifically, as well as Level II evidence supporting the use of auricular points seed embedding, manual auricular acupuncture, and auricular EA in surgery patients. Meanwhile, there is only Level III evidence for the use of body points APS in patients who have undergone cardiac surgery and a cesarean section and Level III evidence for the use of auricular point stimulation for pain reduction after abdominal surgery. The main reason that better levels of evidence were not achieved was the methodological quality of the included studies, with only 13 (22.03%) studies meeting at least five of the seven Cochrane review criteria and only 12 (20.33%) studies that were rated highly in key domains being considered at low risk of overall bias.
Two systematic reviews [21,22] with objectives similar to those of the present study were published in 2008, but overall they produced low quality evidence due to the insufficient quality of included trials. A number of the clinical trials included in the present analysis also had some methodological problems that may have affected their efficacy results. However, we examined all types of APS, with combined and separate analyses of body, auricular, and integrative APS. Rigorously designed large-scale RCTs are needed to identify an optimal standard APS program.

Study Limitations.
Some limitations of this study may have affected the results. For example, the wide variability in APS and surgery types, populations, intervention durations, and timing of outcome measurement may be the main factors underlying the observed heterogeneity, which limits the strength of the study results. The small samples and absence of follow-up evaluation in most included trials may have led to overestimation of the effects of APS. Methods of randomization, blinding, and allocation concealment were not reported or were poorly described in some trials, making quality assessment difficult. In addition, visual inspection 22 Evidence-Based Complementary and Alternative Medicine of the funnel plots revealed some substantial asymmetry in comparisons; thus, the possibility of publication bias (i.e., preference for publication of significant over nonsignificant results) cannot be excluded. In addition, the end-points of included studies varied. End-points in Gilbertson et al. [41] and Chen et al. [32] were 4 and 3 months, respectively. When removing these two studies, the 2 values were decreased markedly (Table 6). Therefore, the end-points of included studies have important biases in this review. Future studies of APS should be designed rigorously to ensure a high level of methodological quality.

Implications for Practice and Research.
The major advantages of APS are related to its clinical safety, favorable effects in postoperative pain relief, and low complication rate following surgery [58]. Clinical nurses and other health care providers should thus be encouraged to learn and implement this simple, convenient, and economical method of postoperative pain control in routine clinical care [34].
Our findings have implications for research on the precise mechanism of APS in postoperative pain relief. Optimal acupoint selection, session duration, stimulation intensity, and application frequency have not been established. A standardized APS program for postoperative pain management should be designed using an evidence-based method. Because available evidence for integrative APS and manual acupuncture is inconclusive, further studies should focus on further assessing the effects of these treatments on postoperative pain control. Moreover, the best APS type for the reduction or elimination of long-term opioid use and the long-term effects of APS therapies remain unknown. Thus, large-scale multicenter RCTs with long-term followup periods should be conducted to verify the short-and long-term effects of APS on postoperative pain control. Furthermore, more attention should be paid to the economic effects of APS in health care systems.
In conclusion, this study indicates that, thus far, there is still insufficient evidence to conclude that APS is an effective method for controlling postoperative pain in surgery patients, although the evidence does suggest that APS can reduce patients' analgesic requirement with no significant adverse effects. The best level of evidence was not adequate in most subgroups. Some limitations of this study may have affected the results, leading to an overestimation of the effects of APS. Rigorously designed large-scale RCTs are needed to identify the effects of APS.

A. Searching Strategies
A. 1