Effectiveness and Safety of Acupuncture and Moxibustion for Primary Dysmenorrhea: An Overview of Systematic Reviews and Meta-Analyses

Background Acupuncture and moxibustion have been accepted as treatment options for primary dysmenorrhea (PD). So far, several systematic reviews (SRs) and meta-analyses (MAs) have reported on the efficacy and safety of acupuncture and moxibustion in treating PD. Objectives The aim of this study was to critically summarize the evidence from relevant SRs and MAs reporting on the efficacy and safety of acupuncture and moxibustion in treatment of PD. Materials and Methods Seven electronic databases, including Cochrane Database of Systematic Reviews, EMBASE, PubMed, SinoMed, China National Knowledge Infrastructure (CNKI), Chinese Science and Technology Periodical Database (VIP), and Wanfang database, were systematically searched. SRs or MAs about acupuncture for PD published up to May 2019 were included in the analysis. More than two authors independently assessed the quality of the evidence by AMSTAR2, PRISMA, PRISMA-A, and GRADE approach. Results A total of 28 SRs and MAs, 281 original studies, reporting on 26,459 female patients were analyzed. The majority of the SRs were of moderate reporting quality and poor methodological quality. Moderate-quality evidence suggested that acupuncture and moxibustion were more effective compared to indomethacin or Fenbid in treating PD. Low-quality evidence suggested that, compared to NSAIDs, acupuncture and moxibustion could relieve pain with less adverse effects. Conclusion Acupuncture and moxibustion seem to be effective and safe approaches in treatment of PD; yet, the methodological quality of most of the studies and the quality of evidence were low. Thus, additional studies are required to further confirm these results.


Introduction
Primary dysmenorrhea (PD) is a common gynecological disorder, mainly characterized by cramping pain in the lower abdomen that occurs before or during menstruation. Headache, nausea, vomiting, fatigue, irritability, diarrhea, and an overall feeling of discomfort are the common symptoms accompanying PD [1]. e disorder can significantly affect women's physical health and life quality [2]. In the USA, PD is responsible for the loss of 600 million work hours and two million dollars each year [3]. In China, the prevalence of dysmenorrhea is 30%∼80%, among which 53% is from PD and 15% from severe dysmenorrhea [4].
Nonsteroidal anti-inflammatory drugs (NSAIDs), oral contraceptive pills, or acupuncture are commonly used to alleviate the menstrual pain. Yet, drug therapy may lead to some adverse events, such as digestive disorders, headache, and drowsiness. In addition, 20%-25% of women do not respond well to these medications [5]. As nonpharmaceutical therapy, acupuncture stimulates the nervous system and release of endogenous substances, such as opioid peptides and serotonin, to improve symptoms [6,7]. Moreover, moxibustion can regulate the levels of reproductive hormones to reduce the pain of dysmenorrhea [8]. Some studies have reported that the combination of acupuncture and moxibustion at SP6 could effectively improve the uterine artery hemodynamics and hemorheology in patients, as well as regulate blood viscosity and erythrocyte aggregation degree to relieve the pain [9][10][11][12].
Systematic reviews (SRs), especially those combined with meta-analyses (MAs), are of essential importance in evaluating clinical efficacy and formulating clinical guidelines [13,14]. In their SR, Smith et al. have reported that current evidence cannot support the effectiveness of acupuncture in treating PD [15]. Moreover, Zhang et al. carried an overview to assess the effect of acupuncture and acupressure on PD using AMSTAR2 (A Measure Tool to Assess Systematic Reviews 2) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) tools [16][17][18]. ey concentrated on acupuncture and acupressure, without referring to the moxibustion and quality of the available evidence. In addition, the results from SRs are conflicting, and the conclusions are limited. To overcome the limitations of an individual SR and to provide comprehensive evidence, an overview of multiple SRs, which is a new approach designed to synthesize the available results, is needed.
We enlarged the research source from seven databases and also considered the intervention of moxibustion so as to provide comprehensive evidence. erefore, we conducted an overview to synthesize and critically appraise the available evidence on the effectiveness and safety of acupuncture and moxibustion for PD by using AMSTAR2, PRISMA, and GRADE (Grading of Recommendations Assessment, Development and Evaluation) approach [19].

Types of Study.
is SRs-MAs evaluate research articles written in Chinese and English reporting on effectiveness and safety of acupuncture and moxibustion for PD. Review articles, letters, conference papers, abstracts, protocols, and network meta analyses were excluded.

Types of Participants.
We included female patients of reproductive age suffering from PD. e definition of PD was based on cyclic pelvic pain during menstruation without any gynecological pathology, such as endometriosis, adenomyosis, or uterine myoma. Patients with secondary dysmenorrhea or serious medical conditions were excluded.

Criteria for Intervention.
e interventions included needle acupuncture, electro-acupuncture, auricular acupuncture, moxibustion, acupressure, point injection, or any combination of the above.

Criteria for Comparison.
We included western medicine, placebo, sham acupuncture, no treatment, or any combination of these.

Primary Outcome.
e total effective rate [20] was selected as a primary outcome. It was calculated based on the ratio of the total number of those who were effectively cured and the total number of sick people [20].
2.2.6. Secondary Outcomes. Secondary outcomes were the following: clinical effective rate, visual analogue scale (VAS), adverse effects, quality of life, and symptom of dysmenorrhea. Clinical effective rate was calculated based on the ratio of the total number of people who responded well to therapy and the total number of sick people who continued to be sick [20]. VAS was analyzed according to a previously described approach [21]. Adverse effects were measured as incidence of side effects and types of side effects. Quality of life [22] was measured using a validation scale, for example, the Short Form (SF) 36. Symptoms of the dysmenorrhea were analyzed according to a previously described approach [23].

Study Selection and Data Extraction.
Two reviewers (JY and TY) separately searched the aforementioned databases and listed the titles of all articles. According to the inclusion criteria, by looking through the title and abstract, they excluded papers that were not eligible. Next, they screened the contents of the unclear articles further. If articles contained insufficient information to make a decision on eligibility, authors of the original reports were contacted so as to obtain further details. Finally, investigators (TY and XW) independently extracted data on the first author's name, year, studies/participants, intervention, comparison, main outcomes, and adverse effects from the full text, all of which were recorded by WPS 2019. Any disagreements were resolved by discussion or consulting with a third reviewer (XJ), until reaching a consensus.

Quality Assessment.
On the basis of the first edition (AMSTAR), the newly developed high-quality evaluation tool of systematic review methodology (AMSTAR 2) has good consistency and practicability for estimators. We integrated the preferred reporting items for systematic reviews and Meta-Analyses-Abstract (PRISMA-A) and PRISMA to evaluate the reporting quality of the studies. e methodological quality and reporting quality of the included reviews were respectively assessed by AMSTAR2 and PRISMA. e investigators systematically studied the relevant data and known evaluation methods. Two reviewers (YJ and YT) independently assessed the quality of the research; disagreements were solved by discussion or consulting with a third reviewer (JX). We calculated the number and 95% confidence intervals of 3 levels ("Yes," "No," or "Partial Yes") for the AMSTAR-2 and PRISMA items. e 3 levels were scored as 1, 0.5, or 0 points separately for statistical analysis purposes.
e methodological quality of each study was evaluated using the system evaluation credibility rating of AMSTAR 2.

Quality of Evidence.
e GRADE approach was used to assess the quality of evidence for main outcomes. e rating included four levels: high, moderate, low, and very low, according to the quality of the evidence. Two reviewers (TY and XW) separately conducted the assessment process, any disagreement was resolved through discussion and consultation with a third author (JX) until a consensus was reached. An overview table, similar to a "summary of findings" table, was prepared with the help of GRADEPro software as per the GRADE approach. e summary table of the evidence for different SRs and MAs was prepared.

Strategy for Data Synthesis.
We performed a re-metaanalysis of the data where two or more reviews reported on the same or similar intervention for outcomes relevant to our review. Given the overlap of some of SRs and MAs, two reviewers listed RCTs of each SRs and MAs and then excluded those that were overlapping. Risk indices (RRs) or odds ratios (ORs) were standardized for dichotomous outcomes; mean difference (MD) or standard mean difference (SMD) was used for continuous outcomes by using equations published in the Cochrane Handbook for Systematic Reviews of Interventions. RevMan5.3.5 software was used to calculate the standardized effect. According to the heterogeneity levels of the included SRs and meta-analyses, the random-effects model (I 2 ≥ 50%) or fixed-effects model (I 2 < 50%) was properly selected.

Results on Literature Search and Selection.
We obtained 115 relevant citations from seven electronic databases and manual searches. Before screening, 62 duplicates were excluded. After reading the title and abstract, seven records were rejected, including one article that has been published two times, three papers published in different languages, two conference papers, and 2 network metaanalyses. Full texts of the remaining 39 citations were retrieved for further assessment, and 18 citations were eliminated. Finally, 28 articles were included in this study. Exclusion list is described in additional file 1. e flowchart of literature selection is represented in Figure 1.  3.3. Methodological Assessment. We adopted AMSTAR 2 to assess the methodological quality of included studies (Table 3; additional file 2). e mean score was 8.8, ranging from 6 to 14. AMSTAR-2 score showed that the key factors affecting the quality of the literature included item 2 (2 studies explained their review methods before conducting the review), item 4 (1 study provided an comprehensive literature search strategy), item 7 (4 studies provided a list of excluded studies and justified the exclusions), item 9 (19 studies used a satisfactory technique for assessing RoB), item 11 (24 researches applied meta-analytical methods appropriately and gave explaining reasons; 2 studies did not provide the explanation of the heterogeneity), item 13 (25 studies accounted for RoB in individual studies in the results), and item 15 (22 studies took funnel plots or Egger's test and Begger's test to investigate the publication bias, and 9 of the studies gave explanation to discuss the potential impact on the results of the review).

Characteristics of Included
Considering that the common problem of the included studies was lack of protocol and list of excluded studies, we adjusted the items 2 and 7 as the second line of the key factors in the process of the assessment. Our results revealed that more than half of the studies were graded as of critically low quality, 9 of low, 3 of moderate, and 1 of high quality.
Based on the quality results, most reviews followed the principle of PICO to carry on research and build framework. More than 2 reviewers performed study selection and extraction in duplicate. Reviewers evaluated the risk of bias Records excluded (n = 7) an article has been published more than once (1); one paper has been published in different language (2); conference (2); network meta analysis (2) Full-text articles assessed for eligibility (n = 39) Full-text articles excluded, with reasons (n = 18) not the research of interest (13); no access to get the full-text (3); not the intervention of interest (2) Studies included (n = 28) acupuncture and moxibustion (12)  of the included and providing satisfactory explanation for the results, while less attention was paid to the protocol and explanation for selection design, exclusion, or heterogeneity.

Reporting
Quality. PRISMA and PRISMA-A were used to assess the reporting quality of studies (Table 4; additional  file 3). e mean score was 28.45, ranging from 21.5 to 33. We found that most included reviews were of high reporting quality, with the part of the title, information sources, data collection process, risk of bias, and conclusion all being well reported. Some of the weaknesses of the reporting included lack to provide proper report in included studies, synthesis of the results, funding, and registration. In the method section, more than half of the studies provided search strategy for one database, while only one study provided a comprehensive literature search strategy. irteen studies did not make additional analysis, and twelve reviews did not refer to the funding.  Figure 6). In addition, 5 SRs   Figure 9).

Quality of
Evidence. e quality of evidence for 3 outcomes (total effective rate, VAS and adverse events) is shown in Table 5. e results showed that the quality of the evidence was low and all the outcomes were biased in allocation concealment or inadequate blinding; the outcomes of the VAS and adverse events were inconsistent, which was caused by course or treatment of the patient. e funnel plot of the total effective rate (acupuncture and moxibustion vs. indomethacin and acupuncture and moxibustion vs. Somiton) and VAS (acupuncture and moxibustion vs. no treatment) was dissymmetrical.

Summary of Main Findings.
is overview provided a comprehensive overview of the evidence on the effectiveness and safety of acupuncture and moxibustion for PD. Evidence of moderate quality suggested that acupuncture and moxibustion had a positive effect on indomethacin or Fenbid in treating PD. Low-quality evidence showed that compared to NSAIDs, acupuncture and moxibustion could relieve PD related pain with less adverse effects, which needs to be further researched. e adverse effects related to the Evidence-Based Complementary and Alternative Medicine acupuncture and moxibustion were mild, and they included dizziness, fainting, or minimal bleeding after acupuncture. e majority of the SRs were of moderate reporting quality and poor methodological quality. Most of the studies followed the principle of PICO to carry on research and build framework, select proper assessment tool or appropriate methods for statistical combination of results, while they fail to provide registration, and assess the potential impact of individual ROB studies on the results of the meta-analysis or other evidence synthesis. Most of the RCTs did not explain the treatment allocation concealed and blinding. e quality of the reporting of the SRs was limited by lack of data on registration and funding, comprehensive search strategy, and explanation of the heterogeneity. With reference to the abstract, although many studies reported structured abstract, they failed to fully report the synthesis of results, the risk of basis, funding, and registration.

Strengths and Limitation.
Following is the brief summary of the present research: (1) comprehensive search strategies were applied to seven databases to ensure that all relevant reviews were identified; (2) before assessment, we   Evidence-Based Complementary and Alternative Medicine have systematically learned some related courses on methodology and reporting evaluation and consulted relevant methodological experts, professors so as to gain deep understanding, and ensure the accuracy of the evaluation process; (3) during the process, we adopted AMSTAR 2 and PRISMA to evaluate the methodological and reporting quality of the qualified studies, and we combined PRISMA-A to the part of abstract in PRISMA, thus making the evaluation more precise; (4) we excluded the overlapping RCTs and conducted a quantitative analysis of the primary RCTs, with the help of the GRADE approach so as to evaluate the quality of the outcomes with different comparisons; (5) Cochrane Collaboration guidelines were followed for data synthesis. More than two reviewers were engaged so as to minimize potential bias in the overview process.
ere are some limitations in the present study: (1) e methodological quality of both included SRs and primary RCTs was not high, and the quality of evidence for the outcomes was unsatisfactory; thus, the conclusions from this          applied to all kinds of RCTs, and they include some characteristic guidelines such as STRICTA (Standards for Reporting Interventions in Controlled Trials of Acupuncture) [50] and STRICTOM ( e Standards for Reporting Interventions in Clinical Trials Of Moxibustion) [51] for acupuncture and moxibustion, independently. PRISMA is used for SR and MA. (3) We also recommend GRADE approach to assess the evidence quality of the more SRs in the future. (4) e primary RCTs should give more attention to blinding, allocation concealment, and registration, which could result in more reliable evidence.

Conclusion
In conclusion, the current evidence suggests that acupuncture and moxibustion is more effective than ibuprofen or Fenbid in the treatment of PD. While there is no enough evidence to support that acupuncture and moxibustion are safe methods to relieve pain and improve the VAS, future studies should place more emphasis on the safety of acupuncture for PD. Also, more efforts are required to improve the study quality of RCTs and SRs, and researchers should strictly adhere to the CONSORT and PRISMA guidelines.