The Effect of Upper Cervical Mobilization/Manipulation on Temporomandibular Joint Pain, Maximal Mouth Opening, and Pressure Pain Thresholds: A Systematic Review and Meta-Analysis

Highlights

• •Clinicians must consider the effect of convergence on referred pain.
• •It is unclear if manual therapy alone improves outcomes in TMJ dysfunction; further trials are needed to conclusively determine its clinical efficacy.
• •Clinicians should consider patient characteristics and clinical factors when treating TMJ dysfunction.

Abstract

Objective

To evaluate the efficacy of upper cervical joint mobilization and/or manipulation on reducing pain and improving maximal mouth opening (MMO) and pressure pain thresholds (PPTs) in adults with temporomandibular joint (TMJ) dysfunction compared with sham or other intervention.

Data Sources

MEDLINE, CINAHL, EMBASE, and Cochrane Library from inception to June 3, 2022, were searched.

Study Selection

Eight randomized controlled trials with 437 participants evaluating manual therapy (MT) vs sham and MT vs other intervention were included. Two reviewers independently extracted data and assessed risk of bias.

Data Extraction

Two independent reviewers extracted information about origin, number of study participants, eligibility criteria, type of intervention, and outcome measures.

Data Synthesis

Manual therapy was statistically significant in reducing pain compared with sham (mean difference [MD]: -1.93 points, 95% confidence interval [CI]: -3.61 to -0.24, P=.03), and other intervention (MD: -1.03 points, 95% CI: -1.73 to -0.33, P=.004), improved MMO compared with sham (MD: 2.11 mm, 95% CI: 0.26 to 3.96, P=.03), and other intervention (MD: 2.25 mm, 95% CI: 1.01 to 3.48, P<.001), but not statistically significant in improving PPT of masseter compared with sham (MD: 0.45 kg/cm², 95% CI: -0.21 to 1.11, P=.18), and other intervention (MD: 0.42 kg/cm², 95% CI: -0.19 to 1.03, P=.18), or the PPT of temporalis compared with sham (MD: 0.37 kg/cm², 95% CI: -0.03 to 0.77, P=.07), and other intervention (MD: 0.43 kg/cm², 95% CI: -0.60 to 1.45, P=.42).

Conclusion

There appears to be limited benefit of upper cervical spine MT on TMJ dysfunction, but definitive conclusions cannot be made because of heterogeneity and imprecision of treatment effects.

Temporomandibular joint (TMJ) dysfunction, or TMJ disorders, are musculoskeletal and neuromuscular conditions involving the TMJ complex,¹ including jaw muscles, temporomandibular joints, and nerves. Diagnosis of TMJ dysfunction is largely based on history and physical examination.² Signs and symptoms include abnormal mandibular movement, reduced range of motion during mouth opening and closing, tenderness of the joint line and associated musculature, pain with resisted jaw movement, clicking, crepitus, or locking of the TMJ due to disk displacement.³ Pain can arise from joint articulations, myofascial systems, central sensitization syndromes,⁴ and referred pain from the upper cervical spine and its surrounding nerves.⁵ The yearly incidence of first onset TMJ disorders increases with age, from 2.5% among those aged 18-24, to 4.5% among 35-44 years of age.⁶ Comorbid conditions include fibromyalgia, autoimmune disorders, sleep apnea, psychiatric illnesses,⁷ and cervical spine dysfunction.⁸

The etiology of TMJ dysfunction is unknown but likely multifactorial.⁹ Biomechanical dysfunction such as internal disk derangement can create excessive mechanical stress and irritation, such as clenching and bruxism, leading to abnormal tissue function.¹⁰ While TMJ dysfunction may be associated with postural abnormalities of the head, neck, and mandible, this has been disputed11, 12, 13 and may be attributed to dentition, clenching,¹⁴ localized trauma, and hormonal, occlusal, and psychosocial factors.¹⁵ Pain and craniomandibular dysfunction may also result from cervical pathology causing neurovascular compression and reduced joint mobility.¹⁶ Indeed, those with severe TMJ dysfunction experience greater cervical impairment.¹⁷

The complex biomechanical, anatomic, and neurophysiological links between the upper cervical spine and trigeminal nucleus provide a clinical basis for treatment of the cervical spine. Manual therapy (MT) stimulates a neurophysiological effect that improves pain, increases pressure pain threshold (PPT), increases pain-free grip, improves neurodynamic tension, and changes skin conductance and temperature.¹⁸ Effects can be found both locally and remotely from the treatment site. Given the association between TMJ and the upper cervical spine, MT of the upper cervical spine is commonly used to treat TMJ dysfunction. Techniques including soft tissue massage, cervical joint manipulation, and cervical joint mobilization may be beneficial on their own or when combined with other interventions, such as with postural and jaw exercises.^16,19,20

To date, no systematic review has evaluated the use of upper cervical MT alone for TMJ dysfunction. Few systematic reviews evaluate the efficacy of physical therapy interventions, and a limited number of randomized controlled trials (RCTs) evaluate any particular physical therapy intervention.^13,21,22 These reviews suggest that MT combined with other interventions improves outcomes for those with TMJ dysfunction.^5,19,23 However, limitations with those reviews and their supporting evidence makes it difficult to determine the efficacy of cervical spine joint mobilizations or manipulations. Details about specific MT techniques used were not consistently provided, and MT techniques were variable, including soft tissue release, joint mobilizations, and joint manipulations performed to the cervical spine and/or TMJ.^5,19,23,24 A recent systematic review²³ suggested that both cervical and cervico-craniomandibular MT improve pain and maximal mouth opening (MMO) in individuals with TMJ dysfunction, and adding cervico-craniomandibular MT appears to provide greater benefit than cervical MT alone. As the small number of RCTs have low to moderate quality, the benefit of cervical MT for TMJ dysfunction remains unclear. An updated review including additional RCTs may better discern the effect of upper cervical MT on TMJ dysfunction.

This systematic review and meta-analysis of RCTs explores whether upper cervical spine mobilization/manipulation is more effective than sham intervention or other interventions for reducing pain, increasing MMO, and increasing PPT of the masseter and temporalis muscles in adults with TMJ dysfunction. We defined MT as the performance of joint mobilization or manipulation only, and upper cervical as spinal segments C0/1, C1/2, or C2/3. We anticipate this review will help clarify the efficacy of upper cervical spine MT for TMJ dysfunction and inform clinicians’ decisions about its use.

Methods

Data sources and searches

An electronic search was conducted from inception to June 3, 2022, with assistance from a Medical Librarian. Databases searched were MEDLINE, CINAHL, EMBASE, and The Cochrane Library, and searches included terms related to TMJ disorder, MT, cervical vertebrae, and RCT. See appendix 1 for the full search strategy. Two independent reviewers screened titles and abstracts to identify relevant studies.

Study selection

The 2 reviewers independently reviewed full text articles according to predetermined eligibility and inclusion criteria. Any disagreement about study eligibility was resolved through discussion. Abstracts or full-text articles not written in English were translated using Google Translate. Studies were excluded if they were not RCTs, or if participants had a history of oral or jaw surgery, malignancy, inflammatory arthritis, and inflammatory conditions of the TMJ, or any metabolic, connective tissue, rheumatic, and hematological diseases. Participants in included trials were adults over the age of 18 with TMJ pain of insidious onset, and trials included articular joint mobilizations and/or manipulations to the upper cervical spine. Authors were contacted if the trial did not clarify the segmental level to which MT was applied. Comparisons of interest were MT vs sham intervention, and MT vs other intervention. Outcomes of interest included pain, maximal MMO, PPTs of the masseter, and/or temporalis muscle. Pain was measured using the numeric pain rating scale (NPRS) or the visual analog scale (VAS), MMO was measured in mm using a range of motion (ROM) scale, and PPT was measured in kg/cm² using a pressure algometer.

We used minimal clinically important difference (MCID)²⁵ values to identify clinically important treatment effects. Although the MCID of both the NPRS and the VAS specific to TMJ pain has not been identified in the literature, a change of 2.0 points on the NPRS²⁶ is deemed clinically important in chronic pain studies, with both tools demonstrating good validity²⁷ and reliability.²⁸ Maximal mouth opening using a ROM scale demonstrates good inter-rater reliability and intra-rater reliability,^29,30 and the MCID for MMO is ∼2.5 mm.³¹ The MCID for PPT has been reported to be 0.2 kg/cm² in the masticatory muscles³¹ and the reliability of pressure algometry in the masticatory structures is high in healthy volunteers³² and in patients with TMJ dysfunction.³³ Reporting aligned with Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines.³⁴

Data extraction and quality assessment

The 2 reviewers independently extracted data for pain, MMO, and PPT of the masseter and temporalis muscles, along with sample size, participant characteristics, and details of interventions. Differences between reviewers were resolved by a third reviewer. Authors were contacted to obtain data in the event of insufficient or missing data. The risk of bias in each RCT was evaluated by 2 reviewers independently using the Cochrane Risk of Bias Tool.³⁵ Items in this tool include risk of bias arising from the randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, selection of the reported result, and other bias. Each item was rated as high risk, low risk, or unclear risk. Based on the Cochrane Risk of Bias tool³⁵ algorithm to determine the overall risk of bias judgment, a trial was considered to have low risk of bias if it was judged to be at low risk of bias for all domains. A trial was deemed to have some concerns if it raised some concerns in at least 1 domain but did not have high risk of bias for any domain. A trial was judged to have high risk of bias if there was high risk of bias in at least 1 domain, or the trial was judged to have some concerns for multiple domains. Any disagreements were resolved through discussion. Evidence quality and strength of recommendations for each outcome were assessed using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system, specifying 4 levels of quality across 5 factors: within-study methodological quality, directness of evidence, heterogeneity, precision of effect estimates, and risk of publication bias.³⁶ Each outcome was graded as high, moderate, low, or very low quality.

Data synthesis and analysis

Analyses were conducted using R software (version 3.5.1; Vienna, Austria).³⁷ Interrater reliability for data extraction was quantified using correlation analyses. As determined a priori, we conducted a random-effects meta-analysis (with the Dersimonian and Laird estimator) because of anticipated variations in study design, such as with population characteristics, treatment protocols, and variability in time to follow-up. Mean difference (MD) scores were determined for pain, MMO, and PPT. We used the longest time point from each trial, and where multiple measurements were taken at a single time point, data were averaged across measurements (eg, bilateral masseter PPT measurements). We determined a priori that sensitivity analyses would be conducted when large or unexpected findings were returned in our analyses, to ensure that models with relatively few studies were not biased by extreme effect size estimates. Egger's regression asymmetry test was used to quantify funnel plot asymmetry for publication bias. In the case of an asymmetrical funnel plot and a significant Egger regression statistic, we planned a priori to apply trim-and-fill analysis to the funnel plot. If a treatment group was shared between more than 1 control group (eg, sham and other),³⁸ then the intervention group sample size was split evenly among each control group (a conservative approach to avoid outcome dependence).³⁹

Post hoc analyses were also performed for each outcome using the earliest time point from each trial to determine if early benefits or early negative effects occurred. We included trials in the post-hoc analysis if the earliest time point was less than 4 weeks after cervical MT treatment. One trial⁴⁰ conducted the earliest follow-up point for all outcomes at 24 weeks; another⁴¹ at 5 weeks for outcome of MMO. Therefore, outcomes of these trials were excluded from the post hoc analysis (see supplemental table S1, available online only at http://www.archives-pmr.org/).

Results

The electronic search yielded 184 records. Manual searching of reference lists and forward citation tracking yielded an additional 3 articles.40, 41, 42 After duplicates and clinical trial protocols were excluded, 63 potentially relevant titles and abstracts remained. One trial⁴³ published as an abstract was excluded after an unsuccessful request to the authors for complete data. After screening, 16 eligible full-text articles were identified. Of those, 8 were excluded because of trial design or the type of MT provided (ie, soft tissue massage, not applied to cervical spine). Eight trials ^38,40, 41, 42^,44, 45, 46, 47 with a total of 437 participants were included in this review. The flow of studies through the review is illustrated in figure 1. Study selection (r=0.99) and outcome data extraction (r=0.99) were highly consistent between reviewers. Of the 8 included trials, 7 were written in English^38,40, 41, 42^,44,45,47 and 1 in Spanish.⁴⁶ Google Translate was used to translate that article and extracted information was confirmed by an individual fluent in Spanish. Five trials compared MT with no intervention^38,41,44, 45, 46 and 4 trials^38,40,42,47 compared MT with another intervention. Of these, 1 contained multiple comparison groups.³⁸ See table 1 for information about each included trial. Risk of bias was variable and is presented in figure 2. Although 3 trials^38,44,47 reported adequate random sequence generation, none blinded participants or study personnel to group allocation. Four trials40, 41, 42^,45 contained other biases including small group sizes, low baseline pain levels which may have led to floor effects, use of prescribed non-steroidal anti-inflammatory medication that differed between groups, and lack of assessment of exercise compliance during follow-up, potentially confounding results. Of the included trials, 1 (12.5%) demonstrated low risk of bias,³⁸ 5 (62.5%) demonstrated unclear risk of bias,^{40,41,44,45,47} and 2 (25%) demonstrated high risk of bias.^42,46 The average age of participants ranged from 20 to 47.1 years; 5 trials^38,40,45, 46, 47 included both men and women, and 3 trials included women only.^41,42,44 The duration of TMJ pain was greater than 3 months in 3 trials^44,45,47 at least 6 months in 1 trial,⁴² between 1 and 5 years in 1 trial,⁴¹ greater than 1 month in 1 trial,⁴⁶ and not stated in 2 trials.^38,40 Pain was evaluated in 5 trials.^{40,42,44,45,47} MMO was evaluated in 6 trials,^38,40, 41, 42^,46,47 PPT of the masseter was evaluated in 4 trials,^38,42,44,45 and PPT of the temporalis was evaluated in 5 trials.^38,42,44, 45, 46

Fig 1.

Flow of trial selection based on Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines.

Table 1.

Characteristics of included studies

First Author, Year, Location	Participants	Intervention and Control/Comparison Groups	Outcome Measures	Duration of TMJ Pain
Oliveira-Campelo et al38 2010, Portugal	n=122 Sex: 31 men, 91 women Mean age: 20±3 y	Atlanto-occipital joint thrust vs suboccipital inhibition technique vs no intervention	Pressure pain threshold of masseter and temporalis, active mouth opening	Not stated
Cuccia et al,40 2010, Italy	n=50 Sex: 22 men, 28 women Age: Intervention group: 40.6±11.03 years, Control group: 38.4±15.33 y	Osteopathic manual therapy (including thrust techniques) vs conventional conservative therapy	Visual analog scale, maximal mouth opening, cervical range of motion measurement	Not stated
Bortolazzo et al,41 2015, Brazil	n=10 Sex: All women Mean age: 25.8±6.8 y	Upper cervical manipulation vs placebo	EMG of mastication muscles, range of motion of mouth opening	Mean: 25.8±6.8 y
Corum et al,42 2018, Turkey	n=60 Sex: All women Age: Cervical manipulation plus exercise group: 27.0±6.3 y, Sham manipulation plus exercise group: 26.0 ± 7.9 years, Patient education group: 28.8 ± 7.6 y	Upper cervical manipulation+neck exercises vs sham manipulation+neck exercises vs patient education only	Numeric rating scale, pressure pain thresholds of masseter and temporalis, pain-free Maximum mouth opening, Short Form 36	≥6 mo
Calixtre et al,44 2019, Brazil	n=61 Sex: All women Age: Intervention group: 26.1±5.7 y, Control group: 26.3±4.6 y	Mobilization of the upper cervical region+craniocervical flexor training vs no treatment	Visual analog scale, pressure pain thresholds of masseter and temporalis, Headache Impact Test, Mandibular Function Impairment Questionnaire	Intervention group: 4±2-9.3 y, Control group: 5±1.5-10 y
La Touche et al,45 2013, USA	n=32 Sex: 11 men, 21 women Age: Intervention group: 33.19±9.49 y, Control group: 34.56±7.84 y	Anterior-posterior upper cervical mobilization vs sham	Skin conductance, breathing rate, heart rate, skin temperature, visual analog scale, pressure pain threshold of 5 points	≥3 mo
Mansilla Ferragud Boscá Gandia,46 2008, Spain	n=52 Sex: 12 men, 40 women Age: Intervention group: 37±8 y, Control group: 36±9 y	Upper cervical manipulation vs control	Width of vertical mouth opening, pressure pain threshold of pterion	>1 mo
Reynolds et al,47 2020, USA	n=50 Sex: 7 men, 43 women Age: Intervention group: 32.2±11.3 y, Control group: 38.8±14.8 y	Upper cervical thrust+education and exercise vs sham manipulation+education and exercise	Maximal mouth opening, Numeric pain rating scale	Intervention group: 81±99.2 mo, Control group: 63.7±67 mo Mean: 72.3±84.2 months

Fig 2.

Risk of bias in the included studies. For each domain the - indicates high risk of bias, a question mark indicates some or unknown risk of bias, and a checkmark indicates low risk of bias.

Manual therapy was described as upper cervical spine (C0/1, C1/2, or C2/3) manipulation in 6 trials^38,40, 41, 42^,46,47 and upper cervical spine mobilization in 2 trials.^44,45 Treatment ranged from 1 to 10 sessions and outcomes were assessed ranging from immediately after treatment to 32 weeks post-treatment. See table 1 for a detailed description of included trials. Sham intervention involved placing the subjects in the mobilization and/or manipulation position and maintaining that position for a certain period. Other interventions included any combination of patient education, exercise, and soft tissue massage to the cervical and/or TMJ region.

Effect of upper cervical spine MT compared with sham intervention

Subjective pain ratings

Meta-analysis of 2 trials^44,45 revealed that MT significantly reduced pain intensity compared with sham intervention (MD: -1.93, 95% confidence interval [CI]: -3.61 to -0.24, P=.03, figure 3A). Included trials were heterogeneous (I²=83%; P=.01). Publication bias was difficult to determine visually (figure 3B) and could not be quantified using Egger regression because of the small number of trials.^44,45 Despite statistically significant reductions in pain, this effect did not meet the 2.0 point MCID threshold²⁶ and may not be clinically important. The true effect likely falls within the 95% CI and could be larger or smaller than the MCID.

Fig 3.

(A) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs sham intervention on subjective pain ratings. (B) Funnel plot to assess publication bias in trials comparing cervical manual therapy and sham interventions that assessed subjective pain ratings. (C) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs other intervention on subjective pain ratings. (D) Funnel plot to assess publication bias in trials comparing cervical manual therapy vs other interventions that assessed subjective pain ratings. Overall, the forest plots returned significant results, demonstrating that MT improved subjective pain ratings, but the studies were heterogenous and publication bias was difficult to assess.

Maximal mouth opening

Meta-analysis of 3 trials^38,41,46 revealed that MT significantly increased MMO compared with sham intervention (MD: 2.11 mm, 95% CI: 0.26 to 3.96, P=.03, figure 4A). The results appeared to be relatively homogenous (I²=0%; P=.55) and visual and formal analyses of the funnel plot demonstrated relative funnel plot symmetry (figure 4B; P=.19). This effect did not meet the 2.5 mm MCID for MMO²⁶ and may not be clinically important.

Fig 4.

(A) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs sham interventions on maximal mouth opening. (B) Funnel plot to assess publication bias in trials comparing cervical manual therapy and sham interventions that assessed maximal mouth opening. (C) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs other interventions on maximal mouth opening. (D) Funnel plot to assess publication bias in trials comparing cervical manual therapy vs other interventions that assessed maximal mouth opening. Overall, both forest plots showed improved MMO in MT-treated patients, and studies did not appear to be heterogenous. Publication bias was difficult to assess via funnel plots and Egger regression.

Pressure pain threshold of masseter and temporalis

Compared with sham intervention, MT did not significantly increase PPT of the masseter (MD: 0.45 kg/cm², 95% CI: -0.21 to 1.11, P=.18; figure 5A)^38,44,45 or temporalis muscle (MD: 0.37 kg/cm², 95% CI: -0.03 to 0.77, P=.07; figure 6A).^38,44, 45, 46

Fig 5.

(A) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs sham interventions on pressure pain threshold of the masseter. (B) Funnel plot to assess publication bias in trials comparing cervical manual therapy and sham interventions assessing pressure pain threshold of the masseter. (C) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs other interventions on pressure pain threshold of the masseter. (D) Funnel plot to assess publication bias in trials comparing cervical manual therapy and other interventions assessing pressure pain threshold of the masseter. Overall, the forest plots demonstrated significant heterogeneity, with no significant effect of MT on PPT thresholds of the masseter muscle, and publication bias was difficult to assess.

Fig 6.

(A) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs sham intervention on pressure pain threshold of the temporalis. (B) Funnel plot to assess publication bias in trials comparing cervical manual therapy and sham interventions assessing pressure pain threshold of the temporalis. (C) Forest plot depicting mean difference (95% CI) of the effect of upper cervical manual therapy vs other intervention on pressure pain threshold of the temporalis. (D) Funnel plot to assess publication bias in trials comparing cervical manual therapy and other interventions assessing pressure pain threshold of the temporalis. Overall, the forest plots demonstrated significant heterogeneity, with no significant effect of MT on PPT thresholds of the temporalis muscle, and publication bias was difficult to assess.

For PPT of the masseter muscle, included trials were statistically heterogenous (I²=92%; P<.001), the funnel plot was visually asymmetrical (figure 5B), and Egger regression was likely underpowered to detect this asymmetry (P=.63). For PPT of the temporalis muscle, included trials were also heterogeneous (I²=86%; P<.001), and the funnel plot was visually asymmetrical (figure 6B). The Egger regression was not significant but may have been underpowered (P=.59).

Effect of upper cervical MT on pain, MMO, and PPT compared with other intervention

Subjective pain ratings

Cervical MT significantly reduced pain (MD: -1.03, 95% CI: -1.73 to -0.33, P=.004; (figure 3C) compared with other intervention, with limited heterogeneity (I²=27%, P=.25). The funnel plot (figure 3D) was relatively symmetrical and the Egger regression statistic was non-significant (P=.25), but conclusions about publication bias are difficult to draw given the few randomized controlled trials (less than 10)⁴⁸ that assessed pain using the VAS or the NPRS.

Maximal mouth opening

Cervical MT significantly improved MMO compared with other intervention (MD: 2.25 mm, 95% CI: 1.01 to 3.48, P<.001; figure 4C). The heterogeneity analysis demonstrated that small improvements were consistent across trials, with no outlying effect size estimates (I² = 0%, P=.60). Analysis of the funnel plot and Egger regression analysis revealed no publication bias (P=.69; figure 4D).

Pressure pain threshold of masseter and temporalis

Cervical MT did not significantly improve the PPT of the masseter (MD: 0.42 kg/cm², 95% CI: -0.19 to 1.03, P=.18; figure 5C) or temporalis muscles (MD: 0.43 kg/cm², 95% CI: -0.60 to 1.45, P=.42; figure 6C) compared with other interventions. Effect size estimates were heterogeneous across trials (PPT masseter I²= 82%, P=.02; PPT temporalis I²= 90%, P=.001). The funnel plots (figures 5D and 6D) and Egger regression analysis could not be completed because of the limited degrees of freedom.

Post hoc analyses: short-term effect of upper cervical MT on pain, MMO, and PPT

The results of the post hoc analyses using the earliest time point were similar to the a priori analyses for most outcomes and subgroups. However, the analyses demonstrated that MT did not significantly improve subjective pain ratings when compared with sham (MD: -1.45, 95% CI: -3.97 to 1.06, P=.26; supplemental table S2, available online only at http://www.archives-pmr.org/), nor MMO when compared with other intervention (MD: 0.63 mm, 95% CI: -1.47 to 2.72, P=.56; supplemental table S2 available online only at http://www.archives-pmr.org/). The MCID for pain and MMO was not achieved, therefore, MT may not have clinically meaningful effects on these outcomes. See supplemental table S2 (available online only at http://www.archives-pmr.org/) for results.

Quality of evidence ratings

Using the GRADE approach, the quality of evidence was rated moderate for subjective pain and MMO, and low for PPT of the masseter and temporalis muscles. See table 2 for the GRADE Evidence Profile.

Table 2.

Grade evidence profile

Group	Number of Studies	Risk of Bias	Inconsistency	Indirectness	Imprecision	Publication Bias	Quality	Mean Difference (95% CI)
Subjective Pain Ratings
Manual therapy vs no intervention	2	Unclear bias risk, some performance bias (allocation concealment, blinding)	Substantial heterogeneity, high inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Low	-1.93 (-3.61 to -0.24)
Manual therapy vs other intervention	3	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Low inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Moderate	-1.03 (-1.73 to -0.33)
Overall effect estimate	5	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Substantial heterogeneity, moderate inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Moderate
Maximal Mouth Opening
Manual therapy vs no intervention	3	Some performance bias (allocation concealment, blinding)	Low heterogeneity, low inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	High	2.11 (0.26-3.96)
Manual therapy vs other intervention	4	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Low heterogeneity, low inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Moderate	2.25 (1.01-3.48)
Overall effect estimate	7	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Low heterogeneity, low inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Moderate
Pressure Pain Threshold-Masseter
Manual therapy vs no intervention	3	Low selection bias, some performance bias (blinding, allocation concealment)	Substantial heterogeneity, high inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Moderate	0.45 (-0.21 to 1.11)
Manual therapy vs other intervention	2	Unclear bias risk, some performance bias (allocation concealment, blinding), unclear for most studies, attrition bias	Substantial heterogeneity, high inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Low	0.42 (-0.19 to 1.03)
Overall effect estimate	5	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Substantial heterogeneity, high inconsistency	No serious indirectness	No serious imprecision	No serious publication bias	Low
Pressure Pain Threshold-Temporalis
Manual therapy vs no intervention	4	Low risk of bias, some performance bias (blinding, allocation concealment)	Substantial heterogeneity, high inconsistency	No serious indirectness	Moderate imprecision	No serious publication bias	Moderate	0.37 (-0.03 to 0.77)
Manual therapy vs other intervention	2	Unclear bias risk, some performance bias (allocation concealment, blinding) attrition bias	Substantial heterogeneity, high inconsistency	No serious indirectness	Moderate imprecision	No serious publication bias	Low	0.43 (-0.60 to 1.45)
Overall effect estimate	6	Unclear bias risk, some performance bias (allocation concealment, blinding), attrition bias	Substantial heterogeneity, high inconsistency		Moderate imprecision	No serious publication bias	Low

Pooled results are based on a few trials with small sample sizes, leading to issues such as heterogeneity and imprecision of effect estimates, affecting the GRADE categories of inconsistency and imprecision, respectively. For all outcomes, small sample sizes (<400) led to downgrading of at least 1 level for imprecision.⁴⁹

Therefore, the overall imprecision was graded as moderate. Because of the broad 95% CIs, the true effect could often fall above or below the MCID for all outcomes of interest, making it difficult to make definitive conclusions about clinical importance. The imprecise CIs led to a downgrade of 2 levels for PPT. Heterogeneity was assessed using the I² statistic.⁴⁹ Substantial heterogeneity occurred for PPT of the masseter and temporalis muscles, and for pain (compared with sham controls), leading us to grade the overall inconsistency as moderate.

Discussion

Our results suggest that mobilization or manipulation of joints of the upper cervical spine does not appear to reduce signs and symptoms of TMJ dysfunction any more than other interventions or sham treatment. Thus, there is a need to further carefully consider the role of joint mobilization/manipulation of the cervical spine in management of TMJ dysfunction. While upper cervical spine MT significantly reduced TMJ pain, this effect is unlikely to be clinically important as the MCID threshold of 2.0 points²⁶ was not achieved. Although the effect estimate did not surpass the MCID threshold for mouth opening of 2.5 mm,³¹ the broad 95% CI suggests that the true effect could be above or below the MCID. Therefore, conclusions about the clinical benefit of MT on MMO cannot be drawn. Although cervical spine MT resulted in large improvements in PPT of the masseter and temporalis muscle that surpassed the MCID of 0.2 kg/cm²,³² its effects were not statistically significant, making an estimate of clinical benefit challenging. This, coupled with the imprecise estimates of the effect, make it difficult to draw definitive conclusions about the clinical benefit of MT on PPT. Overall, our results suggest that upper cervical spine MT on its own may not provide benefit for TMJ dysfunction, but additional high-quality studies are needed.

There is currently no standardized treatment approach for TMJ dysfunction, and clinicians tend to use a multimodal approach50, 51, 52 to care including education, exercise, and MT. The link between the upper cervical spine and TMJ dysfunction has led manual therapists to evaluate the principle of regional interdependence, implying that successful outcomes can be achieved by treating areas remote from the symptomatic site.⁵³ Clinicians therefore often consider the cervical spine as a potential source of TMJ symptoms. Other systematic reviews have investigated the efficacy of various cervical MT techniques on TMJ dysfunction,^5,19,23,24 such as myofascial release to the masticatory and cervical muscles, mobilization/manipulation to the cervical and thoracic spine, and interventions combining the two.^5,19 In 1 review, cervical MT consisted of muscle and nerve tissue techniques,²³ and another included trials that considered myofascial therapy, TMJ mobilizations, soft tissue mobilization muscles, and muscle stretching exercises.²⁴ Our review is the first to specifically examine the efficacy of upper cervical articular joint mobilizations and manipulations (C0-3) on TMJ dysfunction. By defining MT as articular techniques only, we expect it may be easier to identify the efficacy of this specific treatment by reducing the likelihood that other forms of MT may have contributed to its effects.

Temporomandibular dysfunction has been associated with poor quality of sleep⁵³ and exposure to stressful life events, especially when pain is of muscular origin.⁵⁴ Psychological overlay, chronic stress, duration of pain, and patient-specific beliefs and expectations regarding care may have increased the clinical heterogeneity among trials. This may be especially true for subjective outcomes and may explain why subjective pain ratings and PPTs of the masseter and temporalis were more heterogeneous compared with that of MMO. However, even with small effect sizes and large heterogeneity, meaningful effects can sometimes be derived from a particular intervention.⁵⁵

Study limitations

Five of the 8 included trials40, 41, 42^,45,46 demonstrated selection bias, and all demonstrated similar degrees of performance bias. Our review suggests that MT provides limited clinical benefit for TMJ dysfunction, but conclusions are limited by the low to moderate quality evidence included in the review. Low precision of estimates and high heterogeneity contributed to wide CIs in the pooled results. Additional higher quality RCTs are needed in this area, including those with larger sample sizes, which should lead to more precise estimates of the effect (narrower CI).⁵⁶

Small sample sizes and limited number of trials contributed to heterogeneity,⁵⁷ but this is also influenced by unique patient characteristics, treatment techniques, and variability in measured outcomes.⁵⁸ Additional reviews evaluating the efficacy of MT and other interventions such as exercise, education, and therapeutic agents should help clinicians develop effective intervention strategies for those with TMJ dysfunction. Similarly, as many trials did not state the type of mobilization or manipulation performed such as specific joint level targeted, or criteria for successful manipulation (eg, whether cavitation was required), future RCTs should detail the segmental and directional specificity of technique to enhance replication of results.

We chose a priori to assess outcome at the longest available follow-up available to provide information about whether any potential benefit is long-lasting; however, a placebo response cannot be excluded, nor can we exclude the possibility that adverse events were missed. Therefore, a post hoc analysis using the earliest time point available was used to determine whether benefit may have occurred early on but was not long-lasting, or alternatively if there were early negative effects. Although outcomes at each time point were not directly compared, it appears that effects at the 2 time points were similar and that early transient benefits or adverse effects likely did not occur. However, clinicians must weigh the benefits and potential risks when considering upper cervical MT and recognize that adverse events such as headache and discomfort are possible. Proper screening for risk factors can ensure patient safety and minimize the likelihood of adverse events.⁵⁹

Conclusions

Although this review suggests that MT on its own may not provide clinically meaningful benefit for those experiencing TMJ dysfunction, provider preference, and expectations of MT may influence patient-reported outcomes.⁶⁰ Future studies could investigate clinical characteristics that may affect TMJ outcomes, in addition to the biopsychosocial model of disability⁶⁰ to account for the multifactorial nature of TMJ dysfunction.⁶¹ Because of limitations of this work, the low-moderate quality evidence, a small number of trials, heterogeneity of included trials, and imprecise treatment estimates, we are unable to conclusively determine whether upper cervical spine MT provides meaningful benefit for TMJ dysfunction. It appears that treatment effects are small and not likely clinically important. Additional high-quality trials are needed to corroborate these results. For now, clinicians are strongly encouraged to consider the lack of demonstrated benefit of upper cervical spine MT and potential adverse events when treating individuals experiencing TMJ dysfunction.

Appendix 1. SEARCH STRATEGY

Ovid MEDLINE(R) ALL 1946 to June 3, 2022:

Number of results: 32

Search date: June 3, 2022

• 1.exp Temporomandibular Joint/
• 2.craniomandibular disorders/ or exp temporomandibular joint disorders/
• 3.(((temporomandibular or craniomandibular) adj4 (disorder* or disease* or syndrome* or pain*)) or tmj or tmd or costen*).mp.
• 4.1 or 2 or 3
• 5.exp Cervical Vertebrae/
• 6.(Neck or cervic* or C0 or C1 or C2 or C3 or C4 or C5 or C6 or C7).mp.
• 7.5 or 6
• 8.musculoskeletal manipulations/ or manipulation, chiropractic/ or manipulation, osteopathic/
• 9.(Manual therap* or mobilisation* or mobilization* or manipulat* or thrust).mp.
• 10.8 or 9
• 11.exp Clinical trial/ or randomized.tw. or randomly.tw. or trial.tw. or groups.tw.
• 12.4 and 7 and 10 and 11

Embase 1974 to June 3, 2022:

Number of results: 47

Search date: June 3, 2022

• 1.exp temporomandibular joint/
• 2.temporomandibular joint disorder/
• 3.(((temporomandibular or craniomandibular) adj4 (disorder* or disease* or syndrome* or pain*)) or tmj or tmd or costen*).mp.
• 4.1 or 2 or 3
• 5.exp cervical vertebra/ or cervical spine/
• 6.(Neck or cervic* or C0 or C1 or C2 or C3 or C4 or C5 or C6 or C7).mp.
• 7.5 or 6
• 8.joint mobilization/
• 9.exp musculoskeletal manipulation/
• 10.manipulative medicine/
• 11.(Manual therap* or mobilisation* or mobilization* or manipulat* or thrust).mp.
• 12.8 or 9 or 10 or 11
• 13.exp clinical trial/ or randomized.tw. or randomly.tw. or trial.tw. or groups.tw.
• 14.4 and 7 and 12 and 13

Wiley Cochrane Library Trials Database:

Number of results: 68

Search date: June 3, 2022

• 1.[mh "Temporomandibular Joint"]
• 2.[mh ^"craniomandibular disorders"] or [mh "temporomandibular joint disorders"]
• 3.(((temporomandibular or craniomandibular) near/4 (disorder* or disease* or syndrome* or pain*)) or tmj or tmd or costen*):ti,ab,kw
• 4.#1 or #2 or #3
• 5.[mh "Cervical Vertebrae"]
• 6.(Neck or cervic* or C0 or C1 or C2 or C3 or C4 or C5 or C6 or C7):ti,ab,kw
• 7.#5 or #6
• 8.[mh ^"musculoskeletal manipulations"] or [mh ^"manipulation, chiropractic"] or [mh ^"manipulation, osteopathic"]
• 9.(Manual therap* or mobilisation* or mobilization* or manipulat* or thrust):ti,ab,kw
• 10.#8 or #9
• 11.#4 and #7 and #10

CINAHL Plus with Full Text (EBSCOhost interface):

Number of results: 37

Search date: June 3, 2022

S1. (MH "Temporomandibular Joint") OR (MH "Craniomandibular Disorders+") OR (MH "Masticatory Muscles+") OR ( ((temporomandibular or craniomandibular) N4 (disorder* or disease* or syndrome* or pain*)) or tmj or tmd or costen*)

S2. (MH "Cervical Vertebrae+") OR Neck or cervic* or C0 or C1 or C2 or C3 or C4 or C5 or C6 or C7

S3. (MH "Manipulation, Chiropractic") OR (MH "Manipulation, Orthopedic") OR (MH "Manipulation, Osteopathic") OR (MH "Joint Mobilization") OR (MH "Manual Therapy") OR Manual therap* or mobilisation* or mobilization* or manipulat* or thrust

S4. (MH "Clinical Trials+") or randomized or randomly or trial or groups

S5. S1 AND S2 AND S3 AND S4