Effectiveness of Manual Therapy and Therapeutic Exercise for Temporomandibular Disorders: Systematic Review and Meta-Analysis

Study Selection

Inclusion criteria for this review were as follows.

Outcomes

The primary outcomes of interest for this systematic review were pain, ROM, and oral function. Oral function for this systematic review focused on limitations of daily activities of patients with TMD measured through different questionnaires. A secondary outcome of interest was pressure pain threshold (PPT).

The minimal clinically important difference for pain has been reported to range from 1.5 to 3.2 points.^30–34 The smallest detectable difference of maximal mouth opening in healthy people has been reported to be 5 mm, indicating that an important change of at least 5 mm can be considered clinically relevant.³⁵ Measurements of PPT have been shown to have good or excellent interrater and intrarater reliability.^36–38 The minimal important difference for PPTs has been reported to be ≥1.10 kg/cm²/s.^39,40 This systematic review was open to all time points: immediately posttreatment and short-term, intermediate-term, and long-term follow-up.

Data Sources and Searches

A bibliographic search of 6 electronic databases was conducted:

• MEDLINE (database root [1966]–April 7, 2015),

• EMBASE (database root [1988)]–April 7, 2015),

• Cochrane Library and Best Evidence (database root [1991]–April 7, 2015),

• ISI Web of Science (database root [1965]–April 7, 2015),

• EBM reviews–Cochrane Central Register of Controlled Trials (database root [1991]–April 7, 2015), and

• CINAHL (database root [1982]–April 7, 2015).

Key words and medical subject headings were identified with the assistance of a librarian who specialized in health science databases and experts in the orofacial pain field. No restrictions were made regarding the language of publication. A manual search of the references of selected studies was conducted as well (refer to the Appendix for an example of the search strategy).

Data Screening

Two independent investigators screened the titles of publications found in the databases and, if available, the abstract of the publication. The Early Review Organizing Software (EROS) (http://www.eros-systematic-review.org/) Web platform was used for screening the articles for inclusion. In order for papers to be included in the review, the paper had to meet all inclusion criteria of this systematic review on the rating form created in EROS software. Studies were analyzed with the available information. Authors were not contacted.

Disagreements between reviewers on inclusion were resolved by consensus. The kappa statistic was calculated using STATA software, version 12, (StataCorp LP, College Station, Texas) to determine the level of agreement between raters on trial inclusion before consensus. Criteria proposed by Byrt⁴¹ were used to interpret kappa values.

Data Extraction

The information of each study included in this review was extracted and entered into Excel or Microsoft Word (Microsoft Corp, Redmond, Washington) files. For each part of the review, data extraction was carried out independently by 2 reviewers. Data were extracted on study characteristics, including the design, type of TMD, type of interventions, main and secondary outcomes, and treatment estimates. Any disagreements on data extraction were resolved by consensus.

Quality Assessment (Risk of Bias)

Assessments of quality (risk of bias) were completed by 2 independent reviewers (any 2 members of the research team). For the assessment of RCTs, our team used a compiled set of items based on the 7 tools most commonly used to evaluate the risk of bias in complex physical therapy trials.⁴² In addition, the risk of bias tool was used with the main outcome of each study to make the assessments. We followed the guidelines established by the Cochrane Collaboration to perform assessments of risk of bias; however, we developed specific decision rules to make decisions as described elsewhere.⁴³ For the overall assessment of risk of bias, a trial was considered at low risk of bias if it was rated as low risk in all individual domains, if the rating was unclear in at least one domain and the other domains were unclear or low, or if the overall assessment of risk of bias was unclear. Finally, an overall assessment of high risk of bias was considered if at least one domain was rated as high. These criteria have been used previously by our team and other authors.^43,44

Any discrepancies in quality ratings were resolved by discussion. If consensus could not be reached, a third member of the review team with expertise in quality assessments (S.A-O.) acted as an arbitrator and made a final decision.

Data Analysis and Synthesis

Data analysis was performed based on type of intervention (ie, exercise, mobilization, and manipulation), TMD diagnosis (myogenous TMD, arthrogenous TMD, mixed TMD), and type of outcome (eg, pain intensity, range of mouth opening [ROM], oral function (oral-related quality of life]). For analysis of continuous outcome data, we used the mean difference (MD) and the standardized mean difference (SMD) with 95% confidence interval (95% CI) to pool data. Heterogeneity was evaluated statistically using the I² statistic. The MD and SMD were defined according to the Cochrane Collaboration,⁴⁵ as follows:

MD is a standard statistic that measures the absolute difference between the mean value in 2 groups in a clinical trial. It estimates the amount by which the experimental intervention changes the outcome on average compared with the control. It can be used as a summary statistic in meta-analysis when outcome measurements in all studies are made on the same scale.

The SMD is used as a summary statistic in meta-analysis when the studies assess the same outcome but measure it in a variety of ways (ie, use different psychometric scales). In this circumstance, it is necessary to standardize the results of the studies to a uniform scale before they can be combined. The standardized mean difference expresses the size of the intervention effect in each study relative to the variability observed in that study.

We decided to pool studies based on TMD diagnosis, intervention provided, and outcome. We grouped studies that had the same diagnosis (myogenous, arthrogenous, or mixed), similar intervention of interest (ie, MT, exercises), and the same underlying outcome. Thus, we created groups of studies that were similar in terms of these characteristics and pooled them. In the presence of clinical heterogeneity in the study population or intervention, the DerSimonian and Laird random-effects model of pooling was used based on the assumption of the presence of interstudy variability to provide a more conservative estimate of the true effect.^46,47

Cohen's criteria were used to interpret values of effect sizes found for our pooled estimates.⁴⁸ Cohen described 0.2, 0.5, and 0.8 as small, moderate, and large effect sizes, respectively.⁴⁸ Review Manager (RevMan) version 5.0 software (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark, 2008) was used to summarize the effects (ie, pooled MD values) and construct forest plots for all comparisons.

Role of the Funding Source

Dr Armijo-Olivo is supported by the Canadian Institutes of Health Research (CIHR) through a full-time Banting fellowship, by the Alberta Innovates Health Solution through an incentive award, by the STIHR Training Program of Knowledge Translation (KT) Canada, and by the Music and Motion Fellowship from the Faculty of Rehabilitation Medicine of the University of Alberta. The funding bodies had no input in the design, collection, analysis, or interpretation of data; writing of the manuscript; or the decision to submit the manuscript for publication.

Diagnosis

There was considerable diversity in the clinical presentations and diagnoses of participants with TMD among the included studies (eTab. 1). Fourteen of the studies examined the effectiveness of the exercise or MT interventions in muscular TMD (myogenous TMD), 14 studies examined the effectiveness in patients with articular TMD (arthrogenous TMD), and 19 studies examined the effectiveness in patients with mixed diagnoses of TMD (including both myogenous and arthrogenous TMD).^28,29 One study looked at both myogenous and arthrogenous TMD.⁵⁸ Twenty-one of the studies^57–76 used the RDC/TMD established by Dworkin and LeResche²⁷ to classify the patients as having TMD. The remaining 27 studies used their own diagnostic criteria, based on signs and symptoms of the patients.

Methodological Quality Assessment

The results of the critical appraisal of the selected studies are presented in eTable 2. Only 6 studies accomplished more than 60% of the items listed in eTable 2.^{57,66,67,76–78} Most of the studies did not accomplish items with important methodological indicators of risk of bias, such as randomization, allocation concealment, blinding, and intention to treat (ITT). For example, study flaws regarding patient selection were mainly related to description and appropriateness of the randomization procedure and concealment of allocation, with only 20 (41.6%) and 4 (8.3%) of the studies meeting these criteria, respectively. As expected, items related to blinding were not achieved by the majority of the studies. Only 3 of the studies used a double-blinded design and could blind participants. These studies used a placebo arm, which is hard to obtain in these types of interventions. In addition, only 12 (25%) of the studies used blinded assessment of outcomes, and none of the studies blinded the therapist. Thus, blinding was the area that was the hardest section to be met by the analyzed studies. When analyzing issues regarding intervention, we found that although it is expected that interventions would be well described to be reproducible, only 64.6% (n=30) of the studies described the main intervention to be tested. In addition, most studies failed to control for cointerventions. Only 6 studies met this item.

Testing participants' adherence to intervention and having adequate adherence was another issue that was not met by many studies (only 11 and 7 studies, respectively). Furthermore, adverse effects were reported on only 10 of the studies, but there was no specific description of such events when they occurred in all of the studies.

Despite the fact that the adequate handling of dropouts is considered an important method used to prevent bias in data analysis, only 17 of the analyzed studies included information regarding the reasons of withdrawals and dropouts, and only 16 studies used intention-to-treat analysis. The outcome measures were not described well in terms of validity, reliability, or responsiveness. Only a few studies reported these items (11, 17, and 3 studies, respectively). Moreover, the authors did not report intrarater or interrater reliability of the assessors who performed outcome measurements. Regarding statistical issues, it was uncertain if sample size was adequate in 30 of the studies, and only 18 studies reported an evaluation of the clinical significance of their results. Risk of bias assessments using the risk of bias tool determined that none of the studies was considered as low risk of bias. Most of them were classified as either unclear (58.4%) or high risk of bias (41.6%).

Effectiveness of Intervention by TMD Diagnoses: Posture Correction Exercises in Myogenous TMD

Two studies^59,60 evaluated the effectiveness of posture correction exercises for patients with myofascial pain. Both studies showed positive results of postural exercises for improving symptoms of muscular TMD. When pooling the data for these 2 studies, which had similar interventions, diagnoses, and outcomes, maximum pain-free mouth opening significantly increased in patients receiving postural training compared with a control group. The MD in maximum pain-free mouth opening was 5.54 mm (95% CI=2.93, 8.15) (Fig. 2), which was clinically significant in favor of postural training.³⁵ Furthermore, patients treated with postural training had significantly fewer symptoms and disturbance with daily living compared with a control group. The SMD in symptoms and disturbance of symptoms with daily life was 1.13 (95% CI=0.48, 1.78), indicating a large, clinically significant effect size for this pooled outcome.

General Jaw Exercises Alone or Combined With Neck Exercise Program in Myogenous TMD

Eight studies^{56–58,62,63,66,79,80} looked at the effect of exercises alone or combined with other therapies for myogenous TMD. The results of these studies were equivocal. Five of them did not find significant differences between a general physical therapy exercise program targeted to the jaw^56,62,79,80 or jaw and neck⁵⁷ compared with a control or other active treatments, such as biofeedback, TENS, use of the TheraBite Jaw Motion Rehabilitation System (ATOS Medical AB, Hörby, Sweden), or oral splint therapy. However, 3 studies^58,63,66 showed better outcomes, especially on pain and ROM, compared with control groups.

Data were pooled from studies that had similar outcomes and diagnoses and compared an exercise program and other forms of therapy, such as education,^62,66 or splint therapy.^56,58 Data from these studies^56,58,62,66 indicated that there was a trend to favor exercise therapy for pain-free maximum mouth opening and pain intensity compared with a control group. The MD for pain-free maximum mouth opening was 5.94 mm (95% CI=−1.0, 12.87), which is considered clinically relevant.³⁵ The SMD for pain intensity pooling 5 studies^{57,58,62,63,66} was 0.43 (95% CI=−0.02, 0.87), with a moderate effect size according to Cohen's guidelines.⁴⁸ When performing sensitivity analyses, grouping studies comparing exercise therapy and education,^62,66 a nonsignificant effect was found on pain-free maximum mouth opening (1.92 mm; 95% CI=−0.57, 4.41). However, when comparing exercises and splint therapy,^56,58 a statistically and clinically meaningful effect was found (12.31 mm; 95% CI=7.73, 16.89).

Manual Therapy Targeted to the Orofacial Region in Myogenous TMD

Four studies^64,65,77,81 looked at MT techniques, such as facial manipulation versus botulinum toxin⁸¹ or intraoral myofascial therapy versus waiting list, and self-care education and exercises for people with myogenous TMD.^64,65,77 The results of these studies support the use of MT to treat myogenous TMD, as people treated with all of these approaches had improved mouth opening and reduced jaw pain from baseline. Although the results for the intraoral myofascial therapy and exercise groups were superior to the results for the waiting-list control group, there was no statistically significant difference between them. In addition, facial manipulation had an equivalent effect as botulinum toxin. However, at 3 months after treatment, facial manipulation was slightly superior in reducing subjective pain perception, and botulinum toxin injections were slightly superior in increasing ROM. When pooling the results from 3 of these studies based on similar outcomes and diagnoses and comparing similar interventions regarding MT,^64,65,81 we found that MT significantly reduced pain at 4 to 6 weeks of treatment compared with botulinum toxin or waiting list, approaching a clinically relevant value. The MD for pain intensity was 1.35 cm (95% CI=0.91, 1.78). When pooling the studies that considered the comparison of MT versus a waiting list only, similar results were obtained (1.31 cm; 95% CI=0.86, 1.76).

Manual Therapy Mobilization of the Cervical Spine and Myogenous TMD

A recent RCT conducted by La Touche et al⁶⁷ testing a more specific approach directed to the cervical spine to treat patients with cervico-craniofacial pain of myofascial origin was performed. This preliminary study showed that mobilizations targeted to the cervical spine drastically decreased pain intensity and pain sensitivity (via PPT evaluation) in patients with cervico-craniofacial pain of myofascial origin immediately after the application of the technique compared with placebo treatment. The effect sizes found in this study for pain intensity (28.75 points; 95% CI=21.65, 35.85) and PPT (1.12 kg/cm²; 95% CI=0.96, 1.29) were considered clinically relevant.

Jaw and Neck Exercises Alone or as Part of a Conservative Regimen in Arthrogenous TMD

Eight studies^{58,61,68,82–86} that examined patients with arthrogenous TMD focused on jaw and neck exercises alone or combined with other therapies, such as medications, surgery, or self-care recommendations. Six studies^{58,61,68,84–86} focused on exercise therapy alone,^85,86 exercise therapy combined with conventional treatment,⁶¹ or the combination of jaw exercises with TheraBite⁵⁸ or myofunctional therapy.⁶⁸ The remaining 2 studies^82,83 looked at the effectiveness of surgery (arthrocentesis or arthroscopy) combined with conservative treatment including exercises for the jaw versus jaw exercises alone.

Although the results were mixed, most of the studies favored the use of exercises alone or as part of a general regimen to treat people with arthrogenous TMD, including disk displacements with or without reduction.^{58,68,84–86} However, one study⁶¹ did not find that exercises were superior to a control group involving general physical therapy treatment.

Data were pooled from studies with similar outcomes and diagnoses that compared an exercise program with other forms of therapy, such as education⁶¹ or splint therapy,⁵⁸ or with a control group.^68,86 When pooling the results of the studies investigating the effectiveness of exercise alone or in combination with other conservative therapies on pain intensity,^58,61,68,86 we found that there was no statistically significant difference in pain between exercise and control groups. Nevertheless, there was a trend to favor the exercise group compared with the control group. The SMD for pain intensity was 0.68 (95% CI=−0.04, 1.40), with a moderate effect size according to Cohen's guidelines.⁴⁸ When pooling was focused on those studies including only a control group,^68,86 similar results were found, although the SMD increased (SMD=1.11; 95% CI=−0.73, 2.94). Regarding active mouth opening, a nonsignificant effect was found between general jaw exercises and education, splint therapy, or a control group when pooling 3 studies.^58,61,86 The MD for active mouth opening was 3.13 mm (95% CI=−1.96, 8.23). A trend favoring exercises was observed based on the 95% CI values.

When pooling the studies^82,83 that looked at exercises plus arthrocentesis or arthroscopy versus conservative therapy including exercises alone on active mouth opening at 6 months, we found no differences between these approaches. The MD was −1.01 mm (95% CI=−5.43, 3.42), implying that conservative treatment plus exercises is appropriate to treat disk displacement without reduction or when patients have restricted mandibular movement. The results indicate noninvasive procedures as a first line of treatment.

Manual Therapy Plus Jaw Exercises in Arthrogenous TMD

Seven studies^{69,70,78,87–91} looked at the combined effect of MT plus jaw exercises for people with arthrogenous TMD. Three studies^69,87,88 compared MT and exercises versus splint therapy, 1 study^89,90 compared MT and exercise with self-care and advice regarding prognosis, and 2 studies^78,91 used medication as a comparison. In addition, one study⁷⁰ compared anesthetic blockage of the auriculotemporal nerve and MT and exercises in addition to blockage of the auriculotemporal nerve.

In general, we found that MT plus exercises reduced symptoms and increased ROM for patients with arthrogenous TMD, particularly for those with reduced ROM due to disk displacements without reduction (“closed lock”). Five of these studies^{69,78,87,88,91} favored the use of MT in conjunction with exercises compared with splints^69,87,88 or with medications⁹¹ or other nonconservative treatments for arthrogenous TMD, such as arthroscopy and arthroplasty.⁷⁸

When pooling the results from the studies with homogeneous interventions and similar outcomes, diagnoses, and available data,^{69,78,87,88,90} we found that pain was significantly reduced in patients receiving MT combined with exercises compared with splint therapy, self-care, or medications. The SMD for pain intensity at 4 weeks to 3 months was 0.40 (95% CI=0.13, 0.68), with a moderate effect size according to Cohen's guidelines⁴⁸ (Fig. 3).

When looking at active mouth opening,^69,87,88,90 we found that MT plus exercises significantly increased active mouth opening compared with splint therapy, self-care, or medications. The MD for active mouth opening at 4 weeks to 3 months was 3.58 mm (95% CI=1.46, 5.70).

General Jaw Exercise Program in Mixed TMD

Eleven studies^{55,71–74,92–99} looked at exercises alone or as part of a general conservative therapeutic regimen to treat patients with mixed TMD. In general, exercises for mixed TMD compared with control groups had better results for decreasing pain and improving function and pain sensitivity of the masticatory muscles.^55,71,96,97 However, compared with other forms of active treatments, such as splints, a global postural re-education program, or acupuncture,^{72–74,95,95,98,99} no significant differences between these treatments were found.

When pooling the results of studies with available data and similar interventions and outcomes,^{55,71,73,93,94} we found that exercises in the form of general jaw exercises plus conventional treatment or with the addition of an oral device⁹⁴ were not superior to other treatment modalities, such as splint therapy, global re-education posture, splint plus counseling, acupuncture, or standard conservative care, in improving pain intensity. The SMD for pain intensity was −0.06 (95% CI=−0.50, 0.38), with a very small effect size according to Cohen's guidelines.⁴⁸

When pooling results for mouth opening,^{55,71,73,92,94,95,97} nonsignificant differences were obtained between general jaw exercises and splint therapy, global re-education posture, splint plus counseling, or standard conservative care. The MD for mouth opening was −0.25 mm (95% CI=−2.08, 1.57) (Fig. 4).

Manual Therapy and Mixed TMD

Six studies^76,100–104 looked at MT alone, such as mobilization of atlantoaxial joint,^101,103 mobilization at the level of cervical spine,¹⁰² manipulation of the upper thoracic spine (D1),⁷⁶ massage to masticatory muscles,¹⁰⁴ or mobilizations at the level of TMJ joint,¹⁰⁰ for treating patients with mixed TMD. Results were mixed. The studies by Mansilla-Ferragud et al¹⁰¹ and Otano and Legal¹⁰³ showed positive results at improving mouth ROM and increasing PPT in the orofacial region when comparing mobilization of the atlantoaxial joint versus placebo. However, no statistical differences were found between MT targeted to the jaw and jaw exercises plus splint therapy,¹⁰⁰ between cervical chiropractic adjustment and cervical trigger point therapy,¹⁰² upper thoracic manipulation and placebo,⁷⁶ or masticatory muscle massage and splint¹⁰⁴ for improving symptoms of patients with mixed TMD.

When pooling data from the 2 studies that looked at similar manual techniques, TMD diagnoses, and outcomes,^101,103 although there was no significant difference between mobilization of atlantoaxial joint or control group receiving no mobilization, the MD in mouth opening between control and MT groups was 17.33 mm (95% CI=−10.39, 45.06). This difference can be considered as a clinically relevant improvement in mouth opening favoring MT treatment.

Manual Therapy Plus Exercises for Mixed TMD

Two studies^75,105,106 investigated the effect of MT combined with exercises in people with mixed TMD. Tuncer et al⁷⁵ looked at the specific effect of orofacial and cervical MT combined with stretching techniques for the masticatory and neck muscles compared with exercises for the jaw and neck alone and education (home physical therapy). Von Piekartz and Ludtke¹⁰⁶ compared the effect of orofacial physical therapy and neck exercises and MT techniques targeted to both orofacial and cervical regions plus home exercises compared with treatment targeted to the cervical spine only in people with mixed TMD. Pooling the results of the 2 studies^75,105,106 with similar interventions, outcomes, and diagnoses, we found that MT targeted to the orofacial region or in combination with cervical treatment was better than home exercises for the jaw and neck alone or treatment to cervical spine alone for improving mouth opening. The mouth opening between control and MT groups was 6.10 mm (95% CI=1.11, 11.09) favoring MT groups. This difference was clinically relevant.³⁵

Adverse Events

Adverse effects were reported in only 10 of the 48 included trials. Eight of the trials^{57,65,67,74,75,77,87,88} reported no adverse events with the treatments. Nascimento et al⁷⁰ reported some adverse events due to the anesthetic blockages procedure. In that study, 29.4% of the patients (66/224) had temporary facial nerve paralysis, 0.44% (1/224) had hematoma, and 2.23% (5/224) had positive aspirations. Niemela et al⁷³ reported that pain on TMJ palpation increased significantly in the splint group compared with the control group. No adverse events regarding exercise therapy or MT treatments were reported among the trials included.

Data Synthesis

The overall quality of evidence for most comparisons was low to moderate according to the GRADE approach.⁴⁹ The trials included in this review had unclear or high risk of bias. Thus, the evidence was generally downgraded for 3 reasons: (1) risk of bias, 2) level of heterogeneity (inconsistency), and (3) some imprecision surrounding the effect estimate. Details of GRADE assessment of the included studies are displayed in the Table. From the 14 analyses performed, most of the evidence was considered moderate (9 analyses). The rest of the evidence was considered low. Thus, we can say that the total evidence was considered low.

Main Results

Although the quality of the evidence is mostly uncertain and low, the results of our systematic review showed positive results when using postural exercises and jaw exercises to treat both myogenous and arthrogenous TMD disorders. Manual therapy alone or in combination with exercises shows promising effects. Manual therapy targeted to the cervical spine decreased pain and increased mouth ROM in patients with myogenous TMD. Exercises did not show superiority over other treatments for treating mixed TMD. A general exercise program was effective compared with arthrocentesis or arthrography for treatment of arthrogenous TMD, with conservative treatments as a first line of treatment. There remain limited RCTs of high quality that have investigated the effectiveness of MT and exercises to treat TMD.

Effect of Exercise for Treating TMD

Exercise programs are advocated for treating people with musculoskeletal disorders. Therapeutic exercises are prescribed to address TMD. Passive and active stretching of muscles are performed to increase mouth ROM and reduce pain. Postural exercises are helpful.²¹ The results of our systematic review are consistent with previous reviews,^19,23 showing positive effects when using exercises to treat myogenous and arthrogenous TMD. In particular, interventions including exercises to correct head and neck posture and active and passive oral exercises can be effective for reducing musculoskeletal pain and improving oromotor function.^59,60 However, most of these exercise programs were part of a general conservative treatment regimen including other therapies and did not provide clear information regarding dosage, frequency, or adherence, so the isolated effect of exercise to treat TMD and the optimal regimen are uncertain at this time. General aerobic exercises have been shown to improve muscle strength, flexibility, and functional capacity and could induce analgesia.¹⁰⁷ Further research is needed to investigate the usefulness of aerobic exercise and focused muscular training, especially exercises targeted to cervical muscles in people with TMD.

Effect of Manual Therapy for Treating TMD

Manual therapy has been used to restore normal ROM, reduce local ischemia, stimulate proprioception, break fibrous adhesions, stimulate synovial fluid production, and reduce pain. Based on the results of this systematic review, MT shows promising results for treatment of myogenous, arthrogenous, and mixed TMD, although the evidence is limited and low. A combination of MT for the orofacial region plus MT of the cervical spine was more effective than home exercises or treatment to cervical spine alone in people with mixed TMD. Research, to date, suggests that a mixed therapy involving MT techniques and exercises improves patient outcome. Other systematic reviews have shown similar results.²⁴

Mobilization of the cervical spine resulted in decreases in pain intensity and pain sensitivity (via PPT evaluation) in patients with myogenous TMD that exceed suggested values for minimum clinically important differences for pain and treatment of the cervical spine. Manual therapy techniques such as mobilization of the cervical spine could have an influence on orofacial pain and movement in the jaw through the connections of these 2 systems in the trigeminocervical nucleus.¹⁰⁸

Methodological Elements and Overall Quality of the Evidence Affecting Observed Effect

The overall rating of the evidence for this review was low. This finding was due mainly to the risk of bias of the analyzed studies. The methodological biases common to the included studies could have an impact on results. Selection bias could have existed, as only 20 trials reported appropriate randomization and only 4 reported concealment of allocation.

Another important bias was the lack of blinding, especially of the patients and assessors. Only 12 studies used blinded assessment of clinician-assessed outcomes such as mouth opening. However, we also were interested in pain, which is a subjective outcome and dependent on the patient's report. It is likely that lack of blinding could have affected the results of these studies. However, because of the nature of the interventions investigated, blinding would not be possible in many of them. There is empirical evidence showing that trials without appropriate randomization, concealment of allocation, and blinding tend to report an inaccurate treatment effect compared with trials that include these features.¹⁰⁹ Thus, the results of this systematic review should be interpreted with caution, especially in trials with subjective self-reported outcomes.

Other potential biases that could potentially have affected observed effects were inappropriate handling of withdrawals and dropouts (only 16 trials used ITT analysis). Effect sizes from trials that excluded participants in their analysis or that used a modified ITT protocol tended to be more beneficial than those from trials without exclusions, demonstrating that the ITT principle is important to preserve the benefits of randomization and keep unbiased estimates when the objective of the trial is to investigate effectiveness.¹¹⁰

Studies did not report interventions in sufficient detail to be reproducible. In addition, they did not control for cointerventions and did not have adequate adherence to treatment. These issues are of importance for this study, as it is unclear if the effects on selected outcomes were due to the effect of exercise, MT, or other cointerventions. In addition, it is unclear if the participants received enough dosage of treatments, as adequate adherence was accomplished by only a very small proportion of studies (15.2%). Adherence testing should be systematically studied in future studies with exercise prescriptions.

The present study used a compilation of items from all of the scales used in the reviewed physical therapy literature in addition to the risk of bias tool. Our recent analysis of health scales used to evaluate methodological quality determined that none of these scales are adequate for use alone.^44,111 Therefore, we decided to use all of the scales, using a compilation of their items, to provide a comprehensive and sensitive evaluation of the quality of individual trials. Research investigating methodological predictors for determining trial quality in physical therapy is needed.

Limitations

The findings of this review are specific to TMD (nonsurgical) and to exercise and MT. As with any systematic review, there is the potential for selection bias, yet our group used a comprehensive search strategy and included databases as well as manual search. There was a small proportion of studies in other languages that our team could not translate. However, we believe that most of the representative studies were included in the final analysis of this systematic review. In addition, it has been reported that language-restricted meta-analyses only minimally overestimate treatment effects (∼2% on average) compared with language-inclusive meta-analyses.¹¹² Therefore, language-restricted meta-analyses do not appear to lead to biased estimates of intervention effectiveness.^112,113

The heterogeneity among studies, particularly with respect to TMD diagnosis, study intervention, and chosen control or comparison intervention was a challenge. Many studies included the use of exercises or MT as part of a general treatment program, which made the evaluation of these treatments in isolation difficult. Moreover, different diagnostic criteria for TMD were used. Only 21 out of 48 studies included a diagnostic tool that had been demonstrated as being valid, reliable, and reproducible to diagnose TMD. Thus, diagnoses used for the analyzed studies might not be appropriate. Despite this lack of standardized diagnosis, the study populations in all trials appeared to be representative of patients seen in clinical practice. We encourage clinicians and researchers using the new diagnostic criteria for TMD (RDC/TMD) in future studies to allow consistent diagnoses according to the same criteria, taxonomy, and nomenclature to avoid confusion and misunderstanding.¹¹⁴

Research Implications

No high-quality evidence was found, indicating that there is great uncertainty about the effectiveness of exercise and manual MT for TMD. There is a clear need for well-designed RCTs examining exercise and MT interventions for TMD. Specifically, it is necessary that trials be performed isolating the type of exercise and manual technique that is under testing to allow understanding the effectiveness of this type of treatment. In addition, details of exercise, dosage, and frequency as well as details on manual techniques should be reported to create reproducible results. High-quality trials with larger sample sizes are needed.

Clinical Implications

Although the overall level of evidence is low, exercises and MT are safe and simple interventions that could potentially be beneficial for patients with TMD. Active and passive exercise for the jaw, postural exercises, and neck exercises appear to have favorable effects for patients with TMD. Manual therapy alone or in combination with exercises shows promising effects. Exercises did not show clear superiority over other conservative treatments for TMD.