FormalPara Key Summary Points

Why carry out this study?

– Epidural analgesia (EA) is the most common analgesic intervention to relieve labor pain through the pharmacological action of local anesthetics and opioid analgesics, but it cannot be performed when contraindications exist.

− Moreover, its side effects may include hypotension, motor blockade, fever, and others. However, transcutaneous electrical acupoint stimulation (TEAS) offers increased efficacy without invasive stimulation or side effects of narcotic drugs. Additionally, studies have shown TEAS to relieve labor pain, but no systematic evidence has yet been presented.

− Therefore, this study aimed to compare the efficacy and safety of TEAS and EA for parturients.

What was learned from the study?

− No intervention was superior to the control in terms of standard mean difference, odds ratio, or 95% confidence interval. The surface under the cumulative ranking curve score all ranked TEAS interventions first in the domains of the visual analog score, the failure rate of natural delivery, and adverse events outcomes.

− Compared to EA, TEAS could be considered an alternative analgesic for labor pain in terms of efficacy and safety.

Introduction

Labor pain is recognized as the most intense and challenging pain that many women experience in their lives [1]. This experience results in substantial physiological stimulation in parturients [2, 3], with increases in oxygen consumption, blood pressure, and cardiac output, and effects on psychodynamic behaviors [3, 4]. Additionally, uncontrolled pain during childbirth is associated with an increased rate of Cesarean sections [5] and adverse consequences on newborn well-being, including fetal hypoxemia incidence [2].

In epidural analgesia (EA), a combination of local anesthetics and an opioid analgesic drug are injected into the lumbar epidural space, acting on the spinal nerve roots, blocking sensory nerves, and decreasing endogenous catecholamines [6]. With a high level of satisfaction from mothers [7], EA is considered the most effective [8] and widely used intervention [9] for labor pain relief. However, this treatment is difficult to perform in patients with contraindications such as spinal deformity, unstable hemodynamics, and allergy to anesthetics; moreover, it is sometimes refused by parturients [10]. Additionally, some concerns remain regarding possible EA-related side effects including hypotension, motor blockade, fever, pruritus, and urinary retention [9, 11]. Therefore, finding alternative effective, safe, and non-pharmacological therapies is essential.

Acupuncture is a traditional Chinese medical therapy that originated 2000 years ago and has been increasing in popularity among doctors and patients worldwide [12]. A systematic review published in 2020 [13] indicated that compared to sham-acupuncture, acupuncture may increase pain relief satisfaction of parturients; however, the efficacy of acupuncture is closely related to the skills of acupuncturists. In fact, a considerable challenge for physicians arises from the requirement that physicians must maintain hand-manipulation of the needle throughout the process of delivery to maintain a sense of “Deqi” in parturients.

As an accepted noninvasive and non-narcotic pain relief treatment [14], transcutaneous electric nerve stimulation (TENS) has been successfully used in controlling pain during labor and delivery [15]. Based on the traditional theory of acupuncture and moxibustion, transcutaneous electrical acupoint stimulation (TEAS), also named acupuncture-like TENS or needleless acupuncture, involves the placement of electro tabs or pens with regulated stimulation modes on specific acupoints to create a two-fold improvement in efficacy over acupoint therapy and TENS. The likely mechanisms responsible for the analgesic effect of TEAS may include inhibition of pain-causing endogenous substance release, promotion of opioid peptide production [16], and disturbance of the normal transduction of the mitogen-activated protein kinase signaling pathway [17]. One pilot study [18] indicated that TEAS may be considered a complementary intervention for pain following abortion. In fact, TEAS has been shown to significantly decrease the visual analog scale (VAS) scores of patients in the treatment of postoperative pain [19]. Although TENS has been suggested as an alternative method for reducing labor pain [20], no systematic review has studied the efficacy and safety of TEAS in pain relief thus far. However, several randomized controlled trials (RCTs) have found that TEAS helps relieve labor pain [21,22,23,24,25,26].

Because of its convenient operation, noninvasiveness, and lack of side effects from drugs, this study aimed to evaluate whether the relatively new TEAS could serve as an alternative analgesic intervention for labor pain and to compare its effectiveness with the already mature EA. Based on insufficient clinical evidence comparing TEAS with EA directly, we designed this network meta-analysis (NMA) using the principle of indirect comparisons [27]. All available direct and indirect evidence regarding the efficacy and safety of TEAS and EA has been analyzed in this review.

Methods

This NMA is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Protocol and Registration

This NMA was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist. We registered the process of our review on the INPLASY website (https://inplasy.com/) under registration number INPLASY202250050.

Data Sources and Searches

Articles published on TEAS, EA, and labor pain in electronic databases including PubMed, EMBASE, Web of Science, and Cochrane CENTRAL were searched from database inception until September 4, 2022. The search strategy (Table S1–S4) comprised the use of MESH or Emtree terms along with their free-text words. Boolean operators AND and OR were used to combine search terms. Ongoing trials or unpublished studies were also searched through Clinical Trials.gov. Two reviewers (Yan and Kan) read titles and abstracts independently for preliminary screening and examined full texts for eligibility. Any disagreement was resolved after discussion or consultation with a third researcher (Ma).

Study Selection

Inclusion Criteria

The inclusion criteria were as follows.

  1. (1)

    The study included a control group of healthy laboring parturients planning natural birth without using TEAS, TENS, EA, acupuncture, or other analgesic treatments before beginning the study intervention.

  2. (2)

    The study used an intervention group of patients who underwent TEAS. Patients received electrical stimulation provided by an electrical stimulator through electrode tabs or acupuncture pens on the target acupoints. Studies using normal electroacupuncture were not included owing to the invasiveness of this procedure. The electrical stimulator was set at certain modes, frequencies, and intensities. EA or TEAS combined with other anesthetic anesthesia was used. There was no restriction on the time of the initiation of intervention, duration of stimulation, acupoints, frequency, waveform, mode, intensity, and pulse duration of treatment.

  3. (3)

    The study used a comparator such as a sham TEAS group that received very low electrical stimulation (< 5 mA), saline placebo, blank control, routine care, or some other form of analgesia combined with other treatment interventions used in the intervention group or the abovementioned interventions. However, if the comparison group received different frequencies, waveforms, modes, intervention times, or other forms of TEAS compared with the intervention group, results were excluded.

  4. (4)

    The main outcomes included indicators of pain intensity VAS; failure to progress natural delivery, including parturients who underwent cesarean section and instrumental deliveries such as forceps and vacuum extraction; and adverse events of parturients. The secondary outcome was a neonatal Apgar score.

  5. (5)

    The study design was RCTs; there was no language restriction.

Exclusion Criteria

The following were excluded.

  1. (1)

    Non-RCTs, quasi-randomized trials, crossover trials, cohort studies, retrospective studies, case reports, protocols, conference summaries, reviews, animal experimental research studies, and ongoing trials without result.

  2. (2)

    Studies that did not identify or had an unclear design.

  3. (3)

    Duplicate publications.

  4. (4)

    Studies with vague descriptions of intervention and comparison methods.

  5. (5)

    Studies without full text or from which it was difficult to extract outcomes.

  6. (6)

    Studies with irrelative interventions and outcomes were excluded.

  7. (7)

    Studies with participants who had received previous oxytocin injections.

Data Extraction and Quality

Two reviewers (Yan and Kan) conducted the data extraction process based on the eligibility criteria and any disagreements were resolved by reaching a consensus. The data collected from all eligible studies included: publication year, first author, sample size, age and gestational age of participants, and intervention measures along with their specific operations and duration. If the information provided was incomplete, we attempted to contact the corresponding author via e-mail to obtain the missing data.

The two reviewers (Yan and Kan) independently reviewed the risk of bias of all included RCTs using the Cochrane Handbook for Systematic Reviews of Interventions tool containing seven items [28]. Each item was classified as low, high, or unclear risk. Disagreements in this process were resolved by reaching a consensus or through a discussion with a third reviewer (Yin).

Data Synthesis and Analysis

Network plots were created by STATA (version SE15.0, Stata Corporation, College Station, TX, USA) to present the connection of the different interventions. Additionally, we used R software (version 3.6.1, R Project for Statistical Computing, https://www.r-project.org/) for heterogeneity analysis [29]. An I2 value between 0 and 40% was considered to indicate no heterogeneity, whereas an I2 value indicated heterogeneity. In this NMA, we used a random effects model to assess the reliability of data without considering its heterogeneity [30]. The NMA analysis was performed using ADDIS (version 1.16.8, Aggregate Data Drug Information System, http://www.drugis.org). First, the consistency model was applied when the P value was > 0.05 in the node-splitting analysis [31], then the potential scale reduction factor (PSRF) analysis determined model convergence. A PSRF value of 1 following stimulations indicated that the model had approximate convergence, meaning that the results were stable [32]. In addition, the surface under the curve ranking area (SUCRA) was used to rank interventions, with a higher score indicating a more positive effect on outcomes [33].

Results

Study Selection

We identified 2901 articles during our initial search. After 1519 duplicates were automatically removed, 1350 articles were eliminated during the screening of titles and abstracts. After reviewing the full text, 22 articles were excluded, of which four were deleted because we were unable to locate the original paper or contact the corresponding author to obtain the unavailable data. Consequently, ten RCTs met the inclusion criteria and were included in the NMA. The PRISMA flow chart shows the process of study selection (Fig. 1).

Fig. 1
figure 1

Flow chart of study selection

Full size image

Study Characteristics

All qualified articles were published in English. There were 605 participants in the intervention groups and 609 in the comparison groups, with no significant differences in age and gestational age between both groups. Six trials [21,22,23,24,25,26] compared TEAS and control interventions, three [7, 34, 35] compared EA and controls, and one [36] compared TEAS and EA. In the comparison groups of RCTs in which TEAS served as an intervention, one group [21] adopted sham-TEAS, one [26] adopted routine care, and four [22,23,24,25] adopted blank control. Seven acupoints were used in TEAS measures, including Hegu (LI4), Shenmen (HT7), Sanyinjiao (SP6), Neiguan (PC6), Ciliao (BL32), Jiaji (EX-B2), and an extra meridian point Neimadian (EX-LE29), with Hegu (LI4) and Sanyinjiao (SP6) being the most commonly used. The dilatational waveform 2/100-Hz electrical frequency of transcutaneous electrical acupoint stimulator was the most used. Among the studies comprising EA, two [34, 35] adopted a saline placebo as a control measure, while one adopted no analgesic intervention [7]. Five types of drugs—bupivacaine (Marcaine), ropivacaine, lidocaine, fentanyl, and sufentanil—were injected during EA in three included studies. The duration of intervention times differed between studies, and the duration of the active phase was the most frequent. The basic characteristics of the included trials are presented in Table 1.

Table 1 Basic characteristics of included studies
Full size table

Quality Assessment

The risk of bias of ten eligible RCTs is shown in Fig. 2. The colors green, yellow, and red represent low risk, unclear risk, and high risk, respectively. All studies were considered as having low risk in terms of random sequence generation as they reported how randomization was performed when dividing groups, including a random number table, balanced block randomization, or software-generated randomized number. In terms of allocation concealment, ten studies were assessed as having low risk as they all described how participants were assigned, including the use of sealed envelopes. Four RCTs were recognized as having low risk as participants and personnel were blinded to the information of the study [21, 22, 34, 35]; however, this was not discussed in the remaining studies. One study was considered as having high risk in terms of blinding participants and personnel because the author mentioned that blinding was unable to be performed owing to the different operations between TEAS and EA [36]. Regarding the blinding of outcome measures, three studies were assessed as having low risk [7, 25, 35], while those that were remaining had unclear risk as no details could be found in their articles. Only one study performed a rigorous triple-blinding method, in which parturients, investigators, obstetricians, and midwives were unable to assume the group to which they were assigned as the authors labeled their containers of injection with the study number only [35]. Ten RCTs were considered as having low risk in terms of incomplete outcome data as either no cases had been shed off from trials or the number of dropout cases was balanced in two groups and the operators recorded the reasons for dropout in detail. All studies reported their planned outcome data completely; accordingly, low risk was attributed to selective reporting. As no description of possible other bias had been reported, nine articles were considered to have unclear risk in terms of final other bias [7, 21,22,23,24,25,26, 34, 36]; however, one study was considered to have low risk as the obstetricians and midwives did not participate in the process of data collection to prevent potential bias [35]. In general, one study was considered to have low risk in all assessment terms [35], five had five terms considered to be low risk [7, 21, 22, 25, 34], and the remaining four had four low-risk terms [23, 24, 26, 36].

Fig. 2
figure 2

Summary of risks of bias

Full size image

Outcomes

VAS

Among ten articles, one did not include pain intensity as its outcome [7], two were excluded as they only provided VAS scores in the form of graphs without available data being extracted [34, 36], and two reported pain intensity at the time according to the degree of cervical dilation [22, 26]. The remaining RCTs recorded the VAS score before and after interventions. In the remaining five studies, the VAS score was assessed 1 h after intervention on a ten-point scale, with one considered to indicate no pain and ten considered to indicate the worst pain possible [21, 23,24,25, 35]. Therefore, these five studies encompassing three interventions or control treatments were included to compare the analgesic effects between groups [21, 23,24,25, 35]. In the network plot, four RCTs compared TEAS with a control treatment [21, 23,24,25], while one compared EA with a control treatment (Fig. 3a1) [35]. We used a random effects model with each node representing either an intervention or control treatment. The data related to the VAS score after 1 h of intervention showed no heterogeneity (I2 = 2.33%, < 40%). After 50,000 simulations, the PSRF value was 1.00. The standard mean difference (SMD) and 95% CrI (confidence interval) indicated that no intervention was superior to the control treatment (Fig. 3a2). As a lower VAS score indicates better analgesia effects, the SUCRA score revealed that the potential best intervention for labor pain relief was TEAS (SUCRA score, 0.71; SMD, − 1.26; 95% CrI, − 3.28–0.95). The funnel plot indicated no publication bias (Fig. S5).

Fig. 3
figure 3

The evidence network graphs and forest plots of main outcomes. a Visual analog scale score; b failure rate of natural delivery; c adverse events. Network graphs: the size of circle represents sample size; the width of the line is proportional to the number of included RCTs. Forest plots: I2 value is used to assess heterogeneity. EA epidural analgesia, TEAS transcutaneous electrical acupoint stimulation, RCT randomized controlled trial, SMD standard mean difference, OR odds ratio, CrI confidence interval

Full size image

Failure Rate of Natural Delivery

Five articles reported the rate of participants who underwent operative delivery methods such as forceps, vacuum extraction, and cesarean section to help further labor as they could not undergo natural vaginal delivery owing to cephalopelvic disproportion, precipitate labor, or unbearable pain [7, 21, 24, 34, 35]. Two studies explored the difference between TEAS and controls [21, 24],while three others compared EA with controls (Fig. 3b1) [7, 34, 35]. Therefore, we selected these five RCTs comprising three interventions or controls to explore the failure rate of natural delivery under two different interventions. Each node represented an intervention or control, and we used a random effects model for operation. No obvious heterogeneity (I2 = 1.05%, < 40%) was revealed. The PSRF value was 1.00 after 50,000 simulations. Moreover, the odds ratio (OR) and 95% CrI revealed that no intervention was superior to the control (Fig. 3b2). Furthermore, the SUCRA score indicated that parturients under TEAS intervention had the lowest relative natural delivery failure rate (SUCRA score, 0.50; OR, 0.96; 95% CrI, 0.18–4.84). There was no publication bias (Fig. S6).

Adverse Events

Four RCTs recorded the number and rate of adverse events including pruritus, nausea, emesis, urinary retention, fever, hypotension, and headache in each group [7, 24, 34, 36]. One study compared TEAS and control treatment [24], two compared EA and control treatment [7, 34], and one compared TEAS and EA (Fig. 3c1) [36]. These studies encompassing TEAS, EA, and controls were included to compare the safety of the different interventions. With each node representing an intervention or control, a random effects model was performed. The data related to adverse events revealed no heterogeneity (I2 = 2.21%, < 40%), and after 50,000 simulations, the PSRF value was 1.00. While no intervention was considered superior to control treatments based on OR and 95% CrI (Fig. 3c2), the SUCRA score revealed that the possible safer treatment for pregnant women was TEAS (SUCRA score, 0.92; OR, 0.15; 95% CrI, 0.01–2.68). No publication bias was found in the funnel plot (Fig. S7).

Apgar Scores

Two studies were excluded as they only reported the number of babies with Apgar scores < 9 or > 7 [34, 35], while other articles recorded this outcome in the form of continuous variables. Four studies compared TEAS with control treatment [21, 22, 24, 25] and one compared EA with control treatment [7] and TEAS [36] separately (Fig. 4a). Therefore, six RCTs encompassing three interventions or control treatments were included [7, 21, 22, 24, 25, 36]. Each node indicated an intervention or control treatment, and there was no heterogeneity (I2 = 36.25%, < 40%) after analysis. Using a random effects model, after 50,000 stimulations the PSRF value was 1.00. Moreover, the SMD and 95% CrI revealed that no intervention was superior to control treatments (Fig. 4b). The SUCRA score indicated that parturients who underwent TEAS tended to have the potential best newborn status 1-minute following delivery (SUCRA score, 0.43; SMD, 0.02; 95% CrI, − 0.15–0.19), and the funnel plot indicated no publication bias (Fig. S8).

Fig. 4
figure 4

The evidence network graph and forest plot of Apgar scores. a Evidence network graph; b forest plot

Full size image

Discussion

As labor pain is the most unbearable physical pain that most women have to endure [37], severe pain may cause postpartum depression [38]. Therefore, it is essential to explore a safe and effective intervention to relieve pain during delivery. In this study, we conducted an NMA to compare the efficacy and safety between two interventions, TEAS, a convenient and noninvasive analgesia therapy and EA, the most common therapy for relieving labor pain. Following strict screening, we identified ten RCTs eligible for our analysis [7, 21,22,23,24,25,26, 34,35,36]. These studies comprised 1214 participants without any significant difference in baseline characteristics. Six trials explored the difference between TEAS and controls [21,22,23,24,25,26], three compared EA with controls [7, 34, 35], and one compared TEAS with EA directly [36].

Based on the abovementioned results, there was no significant difference between TEAS and EA regarding the VAS score, failure rate of natural delivery, or Apgar scores, suggesting that TEAS showed a relatively equal effect of EA in the aspects of reducing labor pain, decreasing failure rate of natural delivery, and improving newborn status. As the SUCRA score of TEAS, all ranked first in three interventions or control, we proposed that TEAS may be a better choice for parturients owing to its relative sufficient analgesia efficacy, higher rate of natural delivery, and Apgar scores.

In terms of adverse events for parturients, our results also suggested that the risks after TEAS were comparable to those after EA as no significant difference was found. The SUCRA score of TEAS also was highest among the three interventions and controls, indicating its possible superior safety compared with EA. Two RCTs involved in TEAS intervention reported the outcome of adverse events [24, 36], one reported no adverse events being found in parturients between TEAS and control groups [24], and the other reported the number of parturients [36], proving significant differences between TEAS and EA. A review [4] indicated that nonpharmacologic approaches including acupuncture and TENS are promising, safe, and innocuous, but require further study. Given the absence of specific adverse events that happened in TEAS, it is unreasonable to conclude that TEAS is the safest treatment without any risks for parturients and neonates. As TEAS is a nonpharmacological intervention, it avoids the drawbacks of possible hemodynamic instability, intoxication [39], subtle depression, and neonatal respiratory depression brought by narcotic drugs [40]; thus, it provided another potential alternative for parturients and doctors under these special circumstances, which may be more acceptable for parturients to accept.

Above all, TEAS showed the better application possibilities in terms of both analgesic efficacy and safety. Not only it could produce equal sufficient analgesia as effective as EA, but it also reduces the fear and pain of needle injection along with a reduction of possible side effects of narcotic drugs. What is more, this treatment is relatively easy to administer as the midwives may place electro pads on the maternal acupoints by themselves following simple training. Therefore, we proposed that TEAS could be considered as a potential alternative effective analgesia intervention used in delivery.

Few studies have examined the causality or difference between TEAS and EA in delivery. Only one trial compared EA with TEAS in labor pain directly [36], one explored its curative effects for fever caused by EA [41], and one researched the efficacy of TEAS combined with EA in postpartum depression [42]. Although there was one NMA exploring the effects of non-pharmacological interventions in labor pain relief [43], we found no NMA comparing TEAS with EA in the databases searched. Therefore, to the best of our knowledge, this NMA was the first to analyze the efficacy and safety of TEAS in labor pain.

However, this NMA also has some limitations. Our network did not cover all common analgesic interventions used in labor pain relief and thus did not reveal the rank of TEAS among all interventions. Moreover, as the pain intensity of multipara and primiparous women are different, we considered dividing our parturients into two groups to analyze the efficacy of analgesia in our further research. Additionally, as there was limited evidence reporting specific side effects of TEAS, its assessment of safety is not comprehensive. More robust studies are needed to perform to verify the proposal.

Conclusions

Overall, our NMA indicated that, compared with EA, the simple intervention TEAS may be a potential choice for parturients in terms of analgesic efficacy and safety to both mothers and neonates, providing novel evidence to support TEAS as a non-invasive and non-narcotic intervention in labor pain under specific circumstances such as hemodynamic instability and allergy to anesthetics. Given the lack of evidence studying specific adverse events in TEAS, the current assessment of safety is not fully comprehensive. More RCTs are required to answer this important question.