Is acupuncture an effective treatment for temporomandibular disorder?: A systematic review and meta-analysis of randomized controlled trials

Abstract

Background:

Acupuncture is used for treating various disorders, but its effects on temporomandibular disorder (TMD) remain unclear. This study aimed to assess the effectiveness and safety of acupuncture for TMD via a systematic review of randomized clinical trials.

Methods:

A total of 11 Korean and worldwide databases were searched to identify acupuncture studies in adults with TMD. A Cochrane risk of bias assessment was performed on all articles; a meta-analysis, which involved the categorization according to the type of control used (inactive control, active control, or add-on), was subsequently performed. The quality of evidence was assessed using the Grading of Recommendations, Assessment, Development and Evaluation methodology.

Results:

The qualitative analysis of randomized clinical trials with acupuncture as the intervention included 32 articles, 22 of which were included in the quantitative analysis (471 participants). Acupuncture significantly improved outcomes (effect rate, relative risk [RR]: 7.00, 95% confidence interval [CI]: 1.91, 25, 62; visual analog scale, standardized mean difference: 0.49, 95% CI: 0.24, 0.73) versus active controls (effect rate, RR: 1.19; 95% CI: 1.12, 1.27). In the analysis of add-ons, acupuncture significantly improved the effect rate and pain intensity (effect rate, RR: 1.36; 95% CI: 1.04, 1.77; visual analog scale, mean difference: −1.23; 95% CI −1.79, −0.67). However, the quality of evidence was determined to range from low to very low.

Conclusion:

Acupuncture in TMD significantly improved outcomes versus active controls and when add-on treatments were applied. However, as the quality of evidence was determined to be low, well-designed clinical trials should be conducted in the future.

1. Introduction

Temporomandibular disorder (TMD) is defined as a type of craniofacial pain involving musculoskeletal problems of the temporomandibular joint (TMJ), masticatory muscle, and related head and neck area. Primary TMD symptoms include pain, limited or asymmetrical jaw movement, and TMJ crepitation.^[1] The TMD worldwide prevalence has been reported to be ≥5%. Reportedly, 6% to 12% of the US population has experienced TMD.^[2,3] Depending on the study, the percentage of individuals within a population experiencing at least one of the primary symptoms or signs of TMD ranges between 40% and 75%, with an especially high frequency of TMD reported among those aged 20 to 59 years.^[1,2,4]

When admitted to the hospital, most TMD patients have a principal complaint of pain.^[5] Pain due to TMD that recurs over a long period is closely associated with a diminished quality of life, which negatively impacts both the mental health and social life of individuals.^[6,7] Additionally, previous studies confirmed that TMD is associated with various chronic diseases,^[8] including metabolic disorders such as type II diabetes and abdominal obesity.^[9,10] From a long-term perspective, TMD, which occurs mostly in middle-aged individuals, impacts public health and negatively affects broad aspects of life.^[11]

Conservative treatments typically used for TMD include physical therapy (ultrasound, ultrashort wave, laser, heat, electro-stimulation, relaxation exercises, active exercises, stretching, and biofeedback); use of a splint or brace; and pharmacological therapy. When conservative treatments are ineffective, surgery (arthrocentesis or arthroscopy) may be considered.^[12–15] However, only a small percentage of TMD patients qualify for surgical treatment. Due to adverse events risk, patients are recommended to consider surgical treatment only when their symptoms cannot be treated using non-surgical therapies.^[16] Drugs used as conservative treatments are often effective. However, their use must be carefully controlled because long-term use may lead to drug resistance, and is accompanied by misuse and abuse risk.^[17,18] Though splints are often applied, the physiological mechanism by which they produce a therapeutic effect has not been elucidated.^[17,19]

Due to the limitations of commonly used treatment strategies, a growing number of patients has turned to Korean medicine (KM) to treat musculoskeletal and neurological disorders.^[20,21] In particular, acupuncture has been applied in treating musculoskeletal disorders.^[22] A previous systematic review confirmed that the use of acupuncture improves various joint disorders, including those of shoulder, elbow, wrist, and ankle joints.^[23–27] Acupuncture has also been often used for treating TMD and is associated with a high degree of patient satisfaction.^[28,29]

To date, several systematic reviews of the effectiveness of acupuncture in the treatment of TMD have been conducted.^[28,30–37] However, a meta-analysis of all articles has not been performed.^{[28,30,32,35,37]} Further, most published articles exclusively use inactive controls.^[28,30–37] No reviews or analyses have classified articles based on their use of controls to determine whether an accurate assessment of the therapeutic effect of acupuncture alone or combined with other treatments has been reported. Accordingly, we aim to conduct a systematic review of randomized clinical trials (RCTs) that assessed the effectiveness and safety of acupuncture in treating TMD following the reporting guidelines of preferred reporting items for systematic reviews and meta-analysis (PRISMA).

2. Materials and methods

The protocol of this study has been registered in PROSPERO (PROSPERO registration number: CRD42020190730, Aug 15, 2020) and waived by the Institutional Review Board of Jaseng Hospital of Korean Medicine because it was not human subject research.

2.1. Data sources and literature search

The literature search was performed to identify RCTs that assessed the use of acupuncture in the treatment of TMD. The international search engines used included 4 English (Ovid Medline, Ovid-EMBASE, Cochrane Central Register of Controlled Trials, and AMED); 1 Chinese (China National Knowledge Infrastructure); and 1 Japanese database (database [DB]; Citation Information by NII). Korean search engines included 5 Korean DBs (KoreaMed, Korean Medical Database; Oriental Medicine Advanced Searching Integrated System; National Discovery for Science Leaders; and Korean Studies Information Service System). A total of 11 DBs were searched, and hand searching was also conducted for additional search. The literature search was performed on March 20, 2020.

Search terms used to search for articles in DBs with high sensitivity included those associated with subjects/participants, interventions, and the study design. Search terms associated with each of the aforementioned categories were as follows: subjects/participants “TMJ,” “TMD,” “CMD,” “temporomandibular,” and “craniomandibular”; intervention, “acupuncture,” “needling,” “electroacupuncture,” and “electro acupuncture”; and study design, “randomized controlled trial,” “randomized controlled trial,” “RCT,” “randomized,” and “randomized.” To search in non-English DBs, corresponding keywords in the language of the DB were used. Detailed search strategies were presented in Supplementary Table 1, http://links.lww.com/MD/J687.

2.2. Study selection

2.2.1. Study types.

All RCTs using acupuncture as an intervention for TMD were included. Non-randomized studies, qualitative studies, observational studies, animal studies, and letters were not included in the meta-analysis.

2.2.2. Participants.

Only studies including adults with TMD were selected. Studies including patients with TMD, dysfunction, myofascial pain, and myofascial syndrome were included. Studies including patients with TMD caused by neurological, psychological, and other factors that were not of musculoskeletal etiology; recent onset acute traumatic TMD; or accompanying inflammatory and neoplastic conditions were excluded from the analysis. Moreover, studies that did not specify adult status as an inclusion criterion, and those that included adolescents were also excluded.

2.2.3. Types of interventions.

The following intervention types were included in the meta-analysis: acupuncture, electroacupuncture, dry needling (DN), warm-needle acupuncture, traditional acupuncture (KM acupuncture points), and contemporary acupuncture (A-shi points and trigger points). Other acupuncture-related stimulation methods that did not clearly specify acupuncture or DN (e.g., laser, seed, or acupressure) use were excluded from the analysis. Further, clinical studies that compared the effects of different types of acupuncture methods were excluded.

2.2.4. Control groups assessed.

Placebo groups such as non-treatment and sham acupuncture were considered inactive controls (e.g., non-penetrating needling, percutaneous needling, and sham laser acupuncture). Active controls included conventional care strategies, such as physical therapies, injection therapy, splinting, and drug therapy including analgesia. Studies with add-on active controls were included only when the same background treatment was performed on both the control and acupuncture groups to allow researchers to distinguish the acupuncture effects.

2.2.5. Outcome assessment.

Methods used to assess outcomes included pain intensity and dysfunction measures of TMJ and quality of life. Pain intensity was reported using pain measurement indicators, such as the visual analog scale (VAS) or the McGill Pain Questionnaire, while dysfunction was assessed using maximum mouth opening (MMO) values and various TMJ movement measurements. Moreover, the effective rate, which assesses the overall effect of treatment, and the number of adverse events, a means to assess safety, were also considered. Studies that did not use an objective assessment tool or those whose reliability and validity had not been established were excluded.

2.3. Data extraction

Articles were independently selected by 2 researchers. First, after excluding duplicate articles, the preliminary selection of articles was performed by reviewing article titles and abstracts using the inclusion/exclusion criteria. Full texts of preliminarily selected articles were then reviewed. Disagreements during the selection process were resolved via discussions, and those that were difficult to resolve were mediated by a third person. Inclusion and exclusion criteria were based on PRISMA guidelines.

2.4. Risk of bias assessment (RoB)

The quality of selected articles was independently assessed by 2 researchers. Disagreements between researchers were resolved via discussion, and mediation by a third person. To assess the quality of the articles ultimately selected for the study, the 6 domain Cochrane risk of bias (RoB) assessment tool was used. Each domain containing one or more assessment items,^[38] which were as follows: whether randomization generation methods were clearly stated (selection bias); whether randomization order was properly concealed (selection bias); whether appropriate blinding of participants and researchers was clearly reported (performance bias); whether appropriate blinding for outcome assessment was clearly reported; whether incomplete outcome data were reported appropriately (attrition bias); whether the study selectively reported results (reporting bias); and whether critical bias not included in other domains was observed (other bias). Each RoB item was evaluated using the following ratings: high, low, or unclear.

2.5. Statistical analysis

A meta-analysis of selected articles was performed to quantitatively assess the effectiveness of acupuncture. Results were extracted as mean values with standard deviation or incidence (number of cases observed vs total number considered). Continuous variables were assessed using inverse variance analysis. Mean difference (MD) or standardized MD values were determined by comparing intervention and control groups. To assess dichotomous variables, relative risk (RR) and 95% confidence intervals were presented. The meta-analysis was performed using a fixed inverse variance method with Review Manager (RevMan) Version 5.4 (Cochrane Collaboration, 2020) software.

Heterogeneity between extracted studies was assessed using the chi-square test (χ²) and I² statistic at a significance level of P < .10. When heterogeneity was identified, subgroup analyses were performed to identify and resolve the cause of heterogeneity.

2.6. Quality of evidence

The quality of evidence provided in studies included in the meta-analysis was assessed by reviewing each control and outcome variable in accordance with grading of recommendations, assessment, development and evaluation (GRADE) methodology. The quality of evidence was determined based on the assessment of GRADE criteria, including RoB, publication bias, inconsistency, indirectness, imprecision, dose-response, and large magnitude of effect. The quality of each parameter was rated as high, moderate, low, or very low.^[39] There were fewer than 10 studies included in each analysis; therefore, publication bias was not assessed. This is because a funnel plot and statistical analyses for asymmetry require at least 10 studies to evaluate publication bias.^[40]

3. Results

3.1. Selection of articles

A literature search of all DBs considered identified a total of 2518 articles. The following DBs contained articles included in the meta-analysis: Ovid Medline (n = 728), Ovid-EMBASE (n = 711), Cochrane Central Register of Controlled Trials (n = 246), AMED (n = 375), China National Knowledge Infrastructure (n = 90), Citation Information by NII (n = 1), KoreaMed (n = 1), KoreaMed, Korean Medical Database (n = 77), Oriental Medicine Advanced Searching Integrated System (n = 88), National Discovery for Science Leaders (n = 112), and Korean Studies Information Service System (n = 87). Additionally, 2 articles were included after hand searching. After excluding duplicates, the titles and abstracts of 2389 articles were reviewed. After exclusion criteria were applied, 190 articles were identified for secondary assessment, where additional articles were excluded for the following reasons: failure to meet PICO criteria (n = 59), were non-RCTs (n = 90), or were duplicates (n = 9). Ultimately, 32 articles were included in the qualitative analysis;^[41–72] 22 of them were included in the quantitative analysis (meta-analysis)^{[41,42,44–47,49,51–54,56,58–62,64,68,69,71,72]} (Fig. 1). The other 11 articles were excluded because outcome indicators were presented as medians and no mean values were given. Figure 1 includes a flow chart of steps performed in the PRISMA-based literature search.

Figure 1.

Preferred reporting items for systematic reviews and meta-analysis (PRISMA) flow diagram of the literature search.

3.2. Study analysis

3.2.1. Study characteristics.

A breakdown by country of the 32 articles ultimately selected revealed that there were 9 from China,^{[58–62,64,68,71,72]} 3 from Turkey,^[41,66,67] 1 from Japan,^[43] 5 from Brazil,^{[49,55,56,63,69]} 3 from the US,^[42,46,47] 2 from the UK,^[45,53] 4 from Spain,^{[50,51,57,70]} 2 from Sweden,^[52,54] 1 from Poland,^[65] and 2 from Austria.^[44,48] Participants with TMD included in the studies presented with various TMJ-related clinical features, and various diagnostic criteria were included. Study populations varied between 12 and 240, while the number of participants included in acupuncture or control group varied between 6 and 80 (Supplementary Tables 2–4, http://links.lww.com/MD/J688, http://links.lww.com/MD/J689, http://links.lww.com/MD/J690).

Study designs for RCTs included in the analysis were as follows: comparisons of acupuncture with an inactive control (n = 16),^[41–56] comparison of acupuncture with a conventionally used active control (n = 19),^{[50,52–54,56–70]} and comparison of effects of an active control with the same treatment plus acupuncture use as an add-on treatment (n = 4).^{[65,66,71,72]} Among these, 5 articles contained both inactive and active controls,^{[50,52–54,56]} and 2 articles used both active controls and add-on treatments with active controls^[65,66] (Supplementary Tables 2–4, http://links.lww.com/MD/J688, http://links.lww.com/MD/J689, http://links.lww.com/MD/J690).

Inactive controls used included the following: sham acupuncture,^{[41–43,45–47,49,50,55]} sham laser acupuncture,^[44,48,56] sham DN,^[51,53] and no treatment (wait-list)^[52,54] (Supplementary Table 2, http://links.lww.com/MD/J688). Traditional active controls identified were as follows: physical therapy including ultrasound and ultrashort-wave therapy,^[50,58–62] laser therapy,^[56] drug therapy,^{[57,64,65,67]} injection therapy,^{[53,63,66–68]} and the use of a splint^{[52,54,69,70]} (Supplementary Table 3, http://links.lww.com/MD/J689). Active controls used as an add-on included the following: the use of a splint,^[65] exercise and training,^[66] magnetic therapy,^[71] and ultrashort-wave therapy^[72] (Supplementary Table 4, http://links.lww.com/MD/J690).

Acupuncture interventions used included various forms of traditional and contemporary acupuncture, including acupuncture point stimulation;^{[42,44–49,52,54–56,58,59,61,62,64,68–72]} electroacupuncture;^[60,68,72] warm needling;^[58,61,64] and stimulation of trigger points^{[41,43,50,51,53,56,57,63,65–67]} or A-shi points.^{[58,60,64,68]} Various proximal (GB2, GB20, GB21, ST6, ST6, SI18, SI19, EX-HN5, etc) and distal (GB34, LI4, ST36, SI2, SI3, etc) acupuncture points were used as sites of acupuncture stimulation. Among these, acupuncture points used in at least 5 studies were ST6,^{[42,49,54,55,58,61,62,68,70–72]} ST7,^{[45,49,54,55,59,61,62,68,71,72]} SI19,^{[54,55,58,59,69,72]} GB20,^{[49,54,55,59,69]} and LI4,^{[42,44,46–49,52,54,55,58,59,62,64,68–72]} with LI4 used in the greatest number of studies.^[18] The number of treatment sessions provided varied between 1 and 15 sessions, while treatment periods ranged from 1 day to 8 weeks. Moreover, follow-up periods considered also varied widely, from 0 days to 12 months (Supplementary Tables 2–4, http://links.lww.com/MD/J688, http://links.lww.com/MD/J689, http://links.lww.com/MD/J690).

To assess outcomes, various outcome indicators for assessing TMD were used in the studies, including pain severity in the TMJ area (VAS,^{[41–50,52–57,63–67,69,70,72]} pressure pain threshold,^{[41,51,53,54,56,63,69]} and McGill Pain Questionnaire score^[55,60]), dysfunction (MMO,^{[43–45,49,51,56,68,69,71,72]} mouth movement evaluations,^{[41,45,57,66,68,70,71]} range of motion^[57,67] and effect rate^{[52,58–62,64,68,71]}). Among these, the outcome indicators used in the meta-analysis were effect rate,^{[52,58–62,64,68,71]} VAS,^{[41,42,44–47,49,53,54,69,72]} and MMO.^{[41,44,45,49,51,69]}

3.2.2. Risk of bias evaluation.

Most of the articles included in the present study had unclear RoB. With respect to RoB, results revealed that studies that used inactive controls tended to have low RoB when compared with studies that used active or active add-on controls (Fig. 2).

Figure 2.

Risk of bias of included studies.

3.2.3. Random sequence generation.

Among all studies that were included, 16 out of 32 studies^{[41–45,48–51,53,55,57,60,61,64,67]} were assessed to have low RoB and these reported on appropriate randomization order methods, such as table of random numbers^[42] and computer random number generator.^{[41,43–45,48–51,53,55,57,60,61,64,67]}

3.2.3. Allocation concealment.

Among the studies with inactive control for the control group, 3 out of 16 studies^[44,45,55] had low RoB, while only 1 study^[47] was determined to have high RoB. Among studies that used active or active add-on controls, 2 out of 21^[64,65] were determined to have low RoB. In studies with low RoB, nontransparent sealed envelopes^{[44,45,55,64,65]} were used for allocation concealment.

3.2.4. Blinding.

With respect to the blinding of study participants, 12 out of 16 studies^{[41–45,47–51,53,55]} with inactive control as the control group were assessed to have low RoB. However, only 3 out of 21 studies^[50,53,67] with active or active add-on control as the control group were assessed to have low RoB, which could be attributed to the difficulty in blinding the participants when an active control is used as the comparator. In all 32 studies, unclear RoB was reported for blinding of researchers.

With respect to outcome blinding, 15 out of 32 studies^{[41,42,44–49,51–53,55,62,65,67]} had low RoB based on the assessment of treatment outcomes by an independent evaluator.

3.2.5. Incomplete outcome data.

The use of incomplete data was also considered. Low RoB was reported in 10 out of 32 studies.^{[41,43–47,49–51,55]} These studies reported that few to no participants were lost throughout the study and follow-up. Authors explained the reasons for dropouts, which were similar between the groups. One study^[57] had high RoB, with differences in the dropout rate of the groups observed.

3.2.6. Selective reporting.

Selective reporting was evaluated. Among the 32 studies considered, 7^{[43,44,50,55,57,67,68]} reported the onset of adverse events.^{[43,44,50,55,57,67,68]} Protocols^[43,50,55] were determined to have low RoB.

3.3. Quantitative data analysis

3.3.1. Acupuncture versus inactive control.

A meta-analysis of 11 out of 16 RCTs^{[41,42,44–47,49,51–54]} included in the qualitative analysis was performed, which assessed the use of acupuncture versus an inactive control. Among these, 9 RCTs included in the meta-analysis^{[41,42,44–47,49,53,54]} used VAS to assess TMD pain improvement due to acupuncture. Meta-analysis results regarding pain VAS showed that acupuncture was significantly more effective than inactive control use for pain improvement, and no heterogeneity was observed (standardized MD: 0.49; 95% CI: 0.24, 0.73; heterogeneity: P = .53, I² = 0%; Fig. 3A). Five RCTs^{[41,44,45,49,51]} assessed TMJ functional improvement due to acupuncture using MMO as an outcome indicator. Meta-analysis results regarding MMO showed that acupuncture significantly improved MMO when treatment groups were compared with inactive control groups. However, heterogeneity was identified (MD: 4.18; 95% CI: 2.87, 5.48; heterogeneity: P < .00001, I² = 88%; Fig. 3B). Among the 11 RCTs that compared acupuncture groups with inactive controls, only 1^[52] assessed the effect rate of acupuncture in TMD treatment. A findings meta-analysis suggested that acupuncture significantly affected effect rate (RR: 7.00; 95% CI: 1.91, 25.62; Fig. 3C).

Figure 3.

Meta-analysis of acupuncture versus inactive control.

3.3.2. Acupuncture versus active control treatment.

A meta-analysis of 11 out of 19 RCTs^{[52,54,56,58–62,64,68,69]} included in the qualitative analysis was performed to evaluate the effectiveness of acupuncture versus that of active controls. Among the 11 RCTs that compared the use of acupuncture versus a conventional active control, 9 studies^{[52,54,58–62,64,68]} determined the effect rate of acupuncture for TMD. Results of the meta-analysis showed that acupuncture significantly affected the effect rate (RR: 1.19; 95% CI: 1.12, 1.27; Fig. 4A), however, heterogeneity was observed (heterogeneity: P < .002, I² = 66%; Fig. 4A). A subgroup analysis was performed to resolve heterogeneity, which showed that acupuncture significantly affected the effect rate compared with the use of conventional active controls, including Western medicine,^[64,68] physical therapy,^[58–62] and splinting^[52,54] (medicine, RR: 1.14; 95% CI: 1.02, 1.29; physical therapy, RR: 1.20; 95% CI: 1.10, 1.31; splint; RR: 1.28; 95% CI: 1.08, 1.51; Fig. 4A), but the heterogeneity problem remained unresolved in all sub-groups (Heterogeneity: medicine; P = .04, I² = 77%, physical therapy; P = .006, I² = 72%; splint, P = .05, I² = 73%; Fig. 4A).

Figure 4.

Meta-analysis of acupuncture versus active control.

Among 13 RCTs that were included in the quantitative analysis, 3 RCTs^[54,56,69] used VAS outcome indicators to assess TMD pain improvement associated with acupuncture versus active controls. A meta-analysis revealed that acupuncture did not statistically differ from active treatments. However, heterogeneity was identified (MD: −0.41; 95% CI: −0.91, 0.10; heterogeneity: P = .19, I² = 40%; Fig. 4B). A subgroup analysis was performed to determine whether treatment effect differences between control and acupuncture groups showed that acupuncture significantly improved pain compared to laser therapy^[56] (MD: −2.31; 95% CI: −4.52, −0.10; Fig. 4B), while there were no differences between acupuncture and splinting.^[54,69] This assessment eliminated heterogeneity (MD: −0.30; 95% CI: −0.82, 0.23; heterogeneity: P = .59, I² = 0%; Fig. 4B).

Among 13 RCTs that compared acupuncture and active treatment methods, only 1^[69] used the MMO as an outcome indicator of TMJ functional improvement. The study used splinting as an active control. Results of a meta-analysis revealed that no significant difference between MMO values of acupuncture and active control groups was observed (MD: 1.05; 95% CI: −2.36, 4.46; Fig. 4C).

3.3.3. Acupuncture as an add-on to active control treatment.

A meta-analysis was performed on 2 of 4 RCTs^[71,72] included in the qualitative analysis, which assessed use of acupuncture as an add-on therapy. Only 1 of 2 RCTs^[72] compared TMD pain improvement in active treatment and active treatment plus add-on (acupuncture and active treatment) groups. A meta-analysis of VAS outcome indicators revealed that the add-on treatment (acupuncture and active treatment) group had VAS values that were significantly improved when compared with the active treatment group (MD: −1.23; 95% CI: −1.79, −0.67; Fig. 5A). Among these, only 1 RCT^[71] assessed the effect rate for TMD therapy in active treatment versus acupuncture and the same active treatment (acupuncture and active treatment) groups; the use of acupuncture and magnetic therapy, a form of physical therapy, was assessed. Results of the meta-analysis showed that the add-on treatment (acupuncture plus magnetic therapy) significantly affected the effect rate (RR: 1.36; 95% CI: 1.04, 1.77; Fig. 5B).

Figure 5.

Meta-analysis of acupuncture with active control versus active control.

3.4. Adverse events

Among studies that compared acupuncture therapy with inactive control treatments, 4^{[43,44,50,55]} included treatment safety data. Itoh et al reported that 1 participant in the acupuncture group dropped out due to symptom worsening.^[43] Lopez-Martos et al reported that no adverse events occurred in both the DN and sham groups.^[50] Salles-Neto et al reported that 4 participants complained of pain, slight swelling, and irritability after acupuncture,^[55] while Schmid-Schwap et al mentioned that adverse events occurred neither in treatment nor in control groups.^[44]

Among studies that compared acupuncture therapies with active controls, 4^{[50,57,67,68]} reported safety data. Gonzalez et al reported no acupuncture-related adverse events, while 41% of patients receiving drug therapy (active control group) reported experiencing drowsiness. Wang et al reported fewer adverse reactions reported by the treatment group patients (electroacupuncture). Studies by Eroglu et al and Lopez-Martos et al reported no intervention-associated adverse events for any groups. Finally, no studies that assessed acupuncture as an add-on therapy (acupuncture plus active control vs active control alone) reported the occurrence of adverse events.

3.5. Quality of evidence

The quality of evidence of articles included in this meta-analysis was assessed according to GRADE methodology-based outcome indicators. In the comparison of acupuncture and inactive control groups, VAS and effect rate quality was determined to be low due to RoB and imprecision, while the quality of MMO evidence was determined to be very low due to RoB, imprecision, and inconsistency. In the comparison of acupuncture and conventional active treatment methods, the quality of VAS and MMO was low due to RoB and imprecision. Effect rate data quality was determined to be low due to RoB and inconsistency. In the comparison of active controls with active control plus add-on therapy (acupuncture plus a conventional active control), the effect rate and VAS quality were determined to be low due to RoB and imprecision.

4. Discussion

The objective of the present study was to assess the effectiveness and safety of acupuncture use in TMD treatment. To determine the effectiveness of acupuncture use in clinical practice, the present study included both traditional (stimulation of KM acupuncture points) and contemporary acupuncture (trigger points stimulation, electroacupuncture, and warm needling). Moreover, a systematic review and meta-analysis that included a subgroup analysis based on the type of control used was performed to determine effects inactive and active controls have on various indicators of TMD.

Acupuncture significantly affected the effect rate when the treatment method was compared to active controls. However, with regard to pain and mouth opening, which could be effectively used to compare treatment groups with inactive and add-on controls, acupuncture did not significantly affect outcomes compared with active controls. Because a greater number of articles compared effect rate than VAS and MMO, it is difficult to draw conclusions from these data. While results confirmed that acupuncture is effective to a degree similar to that of traditional clinical care, its superiority with regard to the effect rate was not determined. Additional studies may be needed to determine whether or not acupuncture is more effective than other traditional therapies. Conversely, acupuncture was significantly more effective than inactive controls. Results also confirmed that the use of a combination of acupuncture and conservative treatments more effectively treated TMD than conservative treatments alone.

The present study had some limitations. First, heterogeneity in the meta-analysis was identified. Heterogeneity, which could affect the interpretation of results, was believed to have occurred due to research design diversity or because studies included in the analysis were from many different countries, had different control groups, assessed the use of varying types of acupuncture, and utilized distinct techniques. Moreover, the TMJ indicators varied, which may have affected heterogeneity. A subgroup analysis based on the type of control used was performed to eliminate the effects of the lack of control uniformity, however, heterogeneity remained unresolved (Figs. 3–5). Second, most articles included in this meta-analysis had an unclear RoB. In particular, articles that used active controls faced blinding difficulties, since patients within active control groups tended to be aware of control therapies provided. These studies also lacked a specific description of allocation concealment and reported incomplete outcome data (Fig. 2). Numbers of studies using each type of control included in the meta-analysis were insufficient. Third, the sample size of studies included in the analysis was small. These factors reduced the quality of evidence of outcomes due to RoB and imprecision (Table 1). Finally, among all RCTs included in this study, only 7 discussed adverse events associated with acupuncture;^{[43,44,50,55,57,67,68]} 5 of these studies^{[44,50,57,67,68]} reported that there were no adverse events associated with acupuncture. However, because the absolute number of articles discussing adverse events was so small, insufficient evidence in support of the safety of acupuncture was provided. Hereafter, high-quality evidence of the therapeutic effectiveness and safety of acupuncture in TMD treatment should be presented via well-designed studies that systematically report the occurrence of adverse events.

Table 1.

Meta-analysis of outcomes and quality of evidence.

Variable	Overall effect						Studies (N)	Sample size (N)	Quality of evidence
	SMD (MD) or RR	95% CI	P	I2	P	Statistical method
Acupuncture vs Inactive control
Pain (VAS) change	0.49	0.24, 0.73	<.0001	0	.53	Fixed Inverse Variance	9	283	Low*,†
Maximum mouth opening change	4.18	2.87, 5.48	<.00001	88	<.00001	Fixed Inverse Variance	5	152	Very low*,†,‡
Effect size	7.00	1.91, 25.62	.003	NA	NA	Fixed Inverse Variance	1	30	Low*,†
Acupuncture vs Active control (Sole)
Effect rate	1.19	1.12, 1.27	<.00001	66	.002	Fixed Inverse Variance	9	746	Low*,‡
Pain (VAS)	−0.41	−0.91, 0.10	.12	40	.19	Fixed Inverse Variance	3	133	Low*,†
Maximum mouth opening	1.05	−2.36, 4.46	.55	NA	NA	Fixed Inverse Variance	1	40	Low*,†
Acupuncture with physical therapy vs Physical therapy (Add-on)
Effect rate	1.36	1.04, 1.77	.03	NA	NA	Fixed Inverse Variance	1	82	Low*,†
VAS	−1.23	−1.79, −0.67	<.0001	NA	NA	Fixed Inverse Variance	1	40	Low*,†

MD = mean difference, NA = not applicable, SMD = standardized mean difference, TCM = traditional Chinese medicine, VAS = visual analogue scale.

^*Risk of bias.

^†Imprecision.

^‡Inconsistency.

Despite these limitations, the present study had many strengths. First, a meta-analysis of previous studies based on the type of control used was performed. While articles on the effectiveness of acupuncture for TMD have been reported, no adequate quantitative analyses have yet been performed,^{[28,30,32,35,37]} nor have analyses clearly identified types of control groups used.^[28,30–37] Two recently published meta-analyses included subgroup analyses. In the analyses, various assessment indicators for TMD-related pain and function were considered. However, the control groups consisted only of inactive controls (sham acupuncture and placebo treatment).^[34,36] Conversely, our study included a meta-analysis and systematic review of various assessment indicators reported in RCTs that assessed effects of acupuncture in accordance with PRISMA reporting guidelines. Therefore, an advantage of the present study is that acupuncture effects were analyzed in a multi-faceted manner. Moreover, various acupuncture techniques used in clinical practice were included as interventions, while acupuncture combined with active treatment was compared with the use of active treatment alone to determine acupuncture effects relevant to various clinical situations. The findings of the present study have the potential to be used as evidence that acupuncture is a good alternative therapy for TMD in patients who face limitations with regard to traditional therapies including analgesics and physical therapy. The study also confirmed that acupuncture is effective when used alone or in combination with traditional clinical therapies, and showed that acupuncture may be used in addition to other treatment types. Hereafter, additional well-designed, large-scale RCTs are needed to enhance the quality of evidence that exists regarding the effectiveness of acupuncture in TMD treatment.

5. Conclusion

Herein, 32 RCTs that demonstrated the effectiveness of acupuncture for TMD were selected; among them, 22 studies that assessed the effectiveness and safety of acupuncture were included in the meta-analysis. All analyses of the effects of various assessment indicators based on the type of control group used (comparisons of inactive and active controls, and effects of add-on treatments) revealed that acupuncture significantly affected outcomes. However, it was difficult to reach a definitive conclusion regarding the effectiveness of acupuncture in TMD treatment because the studies included in the analysis were determined to contain low-quality data. Therefore, well-designed, large-scale RCTs are warranted in the future. These findings can hopefully serve as a basis for future work and have the potential to facilitate the development of improved clinical practice guidelines.

Author contributions

Conceptualization: Sook-Hyun Lee, Yoon Jae Lee.

Data curation: Yoon Jae Lee.

Formal analysis: Sook-Hyun Lee, Koh-Woon Kim, Yoon Jae Lee.

Investigation: Eun Young Park, Sook-Hyun Lee, Koh-Woon Kim, Yoon Jae Lee.

Methodology: Yoon Jae Lee.

Project administration: In-Hyuk Ha, Jae-Heung Cho.

Supervision: In-Hyuk Ha, Jae-Heung Cho.

Validation: In-Hyuk Ha, Jae-Heung Cho.

Writing – original draft: Eun Young Park.

Writing – review & editing: Eun Young Park, Yoon Jae Lee.