Effect of protraction facemask on the temporomandibular joint: a systematic review

This systematic review protocol was registered under the PROSPERO register with the number CRD42017055343 (https://​www.​crd.​york.​ac.​uk/​PROSPERO/​).

Only full-length articles, which fulfilled following criteria according to PICOS schema, were considered for inclusion in this systematic review.

Two review authors (X.H. and X.C.) conducted the electronic searches independently, and any disagreements were solved by discussion or judged by the third reviewer (J.L.).

Five English and three Chinese databases (Cochrane Database of Systematic Reviews, the Cochrane Central Register of Controlled Trials, PubMed, Embase, MEDLINE (via Ovid), Chinese Biomedical Literature Database, China National Knowledge Infrastructure, and VIP Database) were searched to 22 March 2017, with no language restrictions. We also searched six grey literature databases (EOS abstract index, IADR abstract index, clinicaltrials.​gov, ISRCTN registry, Grey Literature Report, and Open Grey) and evaluated studies that were cited in the reference lists of the included articles to ensure the inclusion of all relevant studies. Details of the MEDLINE search is described in Additional file (Additional file 1: Table S1).

Two of the review authors (X.H. and X.C.) examined the titles and abstracts of the identified records and removed obviously irrelevant ones, independently and in duplicate. Any disagreement was solved by discussion or judged by a third author (J.L.).

Two review authors (X.H. and X.C.) examined full text reports of potentially eligible studies with the eligibility criteria, independently and in duplicate. Any disagreement was solved by discussion or judged by a third author (J.L.). If additional information was required, these two review authors would contact the corresponding author of the study and the study would be categorized as awaiting assessment.

The systematic review is a review of studies (not reports). As each study may have been reported in several articles, abstracts, or other reports, a comprehensive search for studies may identify many different reports from potentially relevant studies.

The criteria for comparing reports [35] were as follows:

Result data regarding participants information, intervention, follow-up periods, outcome measurements were extracted and recorded independently and in duplicate by two review authors (X.H. and X.C.). Any disagreement was solved by discussion or judged by a third author (J.L.).

We tested statistical heterogeneity by applying the chi square and I² tests [36]. A low P value provides evidence of heterogeneity of PFM effects. And I² describes the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error. If heterogeneity was high (I²>50%), the random-effects model was chosen for the meta-analysis; Otherwise, the fixed-effects model was adopted [37]. When there is inconsistency in the direction of PFM effect, it might be irrational to conduct a meta-analysis.

We planned to perform a meta-analysis if the data were similar enough. The measurements of PFM effect for binary data were to be expressed as relative risks along with 95% confidence intervals (CIs); as for continuous data, mean difference and 95% CIs would be used. The statistical significance of the hypothesis test was set at α = 0.05 (two tailed z tests) [37].

Two review authors (X.H. and X.C.) evaluated the risk of bias of the included studies by the Cochrane Handbook for Systematic Reviews of Interventions [38], independently and in duplicate.

For the nonrandomized studies, risk of bias was judged with the Newcastle-Ottawa Scale [39]. Using “star system”, a study usually can be awarded one star for each numbered item when it meets certain criteria. In order to give more detailed quality assessment, we regarded “to give stars” as “low risk of bias”, while regard no description as “unclear risk of bias” and the other conditions as “high risk of bias”. (Table 1).

For the randomized studies, risk of bias was judged from six separate domains:

Each study was graded with A, B, or C, according to the GRADE quality analysis criteria [40]:

Figure 1 shows the flow diagram. A total of 261 citations were identified from the electronic and manual search. Twenty studies were considered eligible, and full texts were retrieved after screening the titles and abstracts. Subsequently, five studies were excluded for specialized reasons [41‐45] (Additional file 2: Table S2). One did not set a control group [41], one used bone-anchored maxillary protraction as intervention [42], and three did not report data on TMJ [43‐45]. According to the criteria for comparing reports, we identified that two reports by Mandall (Mandall 2010 [26] and Mandall 2012 [46]) were the same study with different follow-ups, and two reports by Baccetti and Franchi (Baccetti [47] and Franchi [48]) were the same study with different measurements (cephalometric analysis and shape-coordinate Analysis, respectively). Therefore, thirteen studies [26, 46‐59] were finally included in the qualitative synthesis. Two of the included studies were in Chinese and had been translated in English.

Table 2 gives an overview of the experimental setup of the included studies. All the participants involved were patients with skeletal Class III and Angle Class III malocclusion. Three studies [47, 48, 50, 51] limited patients to European-American ancestry. All the study applied PFM with or without rapid maxillary expansion, there was an exception that the study [55] by Gallagher used PFM with slow maxillary expansion. The orthopedic force was between 300 g and 600 g, the direction of elastic traction was forwards and downwards (20°-30°), and participants were asked to wear PFM for more than 10 h. Diagnostic means involve cephalometric analysis, TPS analysis, computed tomograph (CT), cone beam computed tomograph (CBCT), X ray films, electromyography (EMGs), mandibular position indicator (MPI), and research diagnostic criteria for temporomandibular disorders (RDC/TMD).

The detail of quality assessment is outlined in Table 3. Ten studies were of prospective design, and three studies were of retrospective design. Three studies were RCTs, five studies were cohort studies, and four studies were before-after studies. In the study by Franchi and Baccetti, the allocation was not described in detail, and this study was defined as a controlled trial.

According to the GRADE quality analysis, the three RCTs [26, 46, 49, 56] were graded as high value of evidence, seven studies [47, 48, 50‐53, 55, 59] were graded as moderate value of evidence, and three studies [54, 57, 58] were graded as low value of evidence.

In the three RCTs [26, 46, 49, 56], sequence generation was random. There were no incomplete outcomes data and selective reporting in these RCTs. The study by Mandall [26, 46] described allocation concealment, the studies by Kurt [49] and EI [56] did not mentioned allocation concealment (Fig. 2). In the other 10 observational studies [47, 48, 50‐55, 57‐59], the selection of participants was representative in the community. The ascertainment of intervention was securely recorded (Fig. 3). All the before-after studies [54, 57, 58] with no control group showed no description of the derivation of the control group, and lost comparability of participants between treatment and control groups (Fig. 2).

All the included studies lacked blinding of participants and personnel, for the PFM was evident in patients (Fig. 3).

Six studies reported the short-term outcome on displacement of condyle (Table 5). Two included studies [26, 55] applied cephalometric analysis to this issue. The randomized controlled trial by Mandall et al. [26] showed that five of six items were low risk of bias (Fig. 2) and its GRADE quality was A (Table 3) which meant high quality. This study presented the prevalence of forward mandibular displacement on closure. Numbers of patients with a forward mandibular displacement were increased in the control group (52.6 to 70.3%), but decreased in the treatment group (52.9 to 21.9%). In the cohort study by Gallagher et al. [55], five of eight items showed low risk of bias (Fig. 2) and its GRADE quality was B (Table 3). This study found condyle moved inferiorly and posteriorly in treatment group (0.7 mm ± 1.7 mm), indicating that mandible displaced downward and backward during treatment, while no significant differences were found between treatment group and control group (p = 0.232).

The randomized controlled trial by EI et al. [56] showed that three of six items were low risk of bias and one item was high risk (Fig. 2), and its GRADE quality was A (Table 3). This study assessed the direction of condyle movement as well as the disharmony between centric relation (CR) and maximum intercuspation (MI) positions. The downward and forward movement of condyles was observed for most participators before treatment mainly because of occlusal interference caused by anterior crossbite. After treatment, condyles tended to move superiorly (− 0.97 mm ± 0.96 mm, p = 0.001) and posteriorly (1.13 mm ± 1.46 mm, p = 0.007), and meanwhile, the coincidence of CR and MI positions was achieved gradually.

The before-after study by Lee et al. [54] measured displacement of condyle by means of CBCT, and condyle showed displacement to the outside, backward, and upward in treatment group (p < 0.05). Meanwhile, coronoid process and ramus showed displacement to the outside (p < 0.05).

Two studies highlighted the quantitative changes of TMJ joint space. In the prospective before-after study by Gong et al. [57], four of eight items were low risk of bias and two items were high risk (Fig. 2), and its GRADE quality was C (Table 3). This study found that no significant differences were exhibited except posterior and superior joint space. The posterior joint space was decreased significantly from 2.56 mm ± 0.59 mm to 2.11 mm ± 0.67 mm (p < 0.05), and the superior joint space was also decreased from 1.69 mm ± 0.18 mm to 1.31 mm ± 0.25 mm (p < 0.05). The prospective before-after study by Yao et al. [58] showed that four of eight items were low risk of bias and three items were high risk (Fig. 2), and its GRADE quality was C (Table 3). This study observed the changes of anterior, superior, and posterior joint space after treatment. The anterior joint space was increased significantly from 1.48 mm ± 1.22 mm to 2.14 mm ± 1.30 mm (p < 0.01), and the anterior joint spaces area was increased significantly from 3.84mm² ± 1.52 mm² to 5.03mm² ± 1.37 mm² (p < 0.01). Meanwhile, the posterior joint space was decreased significantly from 2.59 mm ± 1.88 mm to 2.28 mm ± 1.56 mm (p < 0.01), and the posterior joint spaces area was decreased significantly from 6.32 mm² ± 1.85 mm² to 5.14 mm² ± 1.29 mm² (p < 0.01).

In summary, it suggests that PFM induced a posterior and superior movement of condyle in the short term. Eliminating any forward mandibular displacement on closure could be an important part of successful treatment, which reflects the success of PFM.

As for the long-term effect, only two studies contained related outcomes. The study by Mandall et al. [46], GRADE quality of which was A (Table 3), found that 50.0% of patients have a forward mandibular displacement in control group with the baseline of 52.6, and 21.9% have a forward mandibular displacement in PFM group with the baseline of 52.9%. The study by Gallagher et al. [55], GRADE quality of which was B (Table 3), found that condyle maintained the significantly inferior and posterior movement in PFM group (2.2 mm ± 4.7 mm, p < 0.05). These results were consistent with the short-term effect, affirming that the displacement effect of PFM remains stable (Table 6).

Four studies were identified for the short-term outcomes of this issue (Table 7). The study by Mandall et al. [26], whose GRADE quality was A (Table 3), measured TMJ symptoms and signs, including TMJ pain (lateral and intra-auricular), muscle tenderness (temporalis, masseter, and lateral pterygoid), crepitus, clicking, locking, and restriction of jaw movement (maximum mouth opening and lateral movement). In treatment group, lateral TMJ pain, intra-articular pain, locking, loss of movement, and temporalis/masseter spasm were not reported. 6/70 TMJs had Clicking, 2/70 TMJs had crepitus, and 1/35 patients lateral pterygoid spasm. In control group, intra-articular pain, locking, loss of movement, and temporalis/masseter spasm were not reported. 3/76 TMJs had lateral TMJ pain, 1/76 TMJs had Clicking, 9/76 TMJs had crepitus, 2/38 patients had lateral pterygoid spasm.

The randomized controlled trial by Kurt et al. [49] showed that four of six items were low risk of bias (Fig. 2), and its GRADE quality was A (Table 3). This study utilized the Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) to diagnose TMD, and found that no patients had disc displacement, 1/17 patients had myofascial pain, and 3/17 patients had arthralgia after PFM. In contrast, no patients had disc displacement, 1/13 patients had myofascial pain, 1/13 patients had arthralgia in control group.

The study by Ngan et al., [59] showed that five of eight items were low risk of bias (Fig. 2), and its GRADE quality was B (Table 3). This study covered the presence of masticatory muscle pain, which is one of the major signs of TMD. All of the pain was at level 1. As a result, superficial masseter in 4/20 TMJs, posterior temporalis in 2/20 TMJs, temporal tendon in 2/20 TMJs, and lateral pterygoid in 3/20 TMJs were identified a level 1 pain in treatment group. After removal of the appliance for 1 month, no masticatory muscle pain was detected. The study by Yao et al. [58], GRADE quality of which was C (Table 3), found that 14/19 patients were freed from TMD symptoms (i.e. clicking, TMJ and muscle pain, and restriction of jaw movement) and 5/19 patients alleviated TMD symptoms.

In summary, as TMJ signs and symptoms were very low, the downwards and backwards force caused by PFM is not a risk factor of TMD in the short term.

Only Mandall et al. [46] reported the long-term outcomes. After a 3-year follow up, in treatment group, lateral TMJ pain, intra-articular pain, locking, loss of movement, or temporalis spasm was not recorded, 11% patients had Clicking, 14.1% patients had crepitus, and 3.1% patients had masseter spasm, or lateral pterygoid spasm. In control group, loss of movement or temporalis spasm was not recorded, 1.4% patients had lateral TMJ pain or intra-articular pain, 5.6% patients had clicking or lateral pterygoid spasm, 9.7% patients had crepitus, 2.8% patients had locking, or masseter spasm. From these results, TMJ signs and symptoms are still low, only crepitus tends to be increased in both groups. It is confirmed that PFM is not detrimental to TMJ in the long-term observation (Table 8).