Psychological Outcomes on Anxiety and Depression after Interventions for Temporomandibular Disorders: A Systematic Review and Meta-Analysis

Many studies have shown mutual interaction between temporomandibular disorders (TMD) and psychological distress. However, evidence on the effectiveness of therapeutic interventions for TMD on psychological outcomes is scarce. This review aimed to summarise the best evidence on the association between interventions for TMD and psychological outcomes regarding symptoms of anxiety and depression. Electronic search was carried out in databases, including Pubmed, Web of Science, Medline, Cochrane Library, and Scopus. All eligible studies were included for narrative synthesis. Eligible randomised controlled trials (RCTs) were included for the meta-analysis. The overall effect size of interventions for TMD was analysed in standardised mean difference (SMD) in levels of anxiety and depression. Ten studies were included in the systematic review. Of these, nine were included in the narrative analysis and four were included in the meta-analysis. All included studies and the result of the narrative analysis showed a statistically significant beneficial effect of interventions for TMD on improving symptoms of anxiety and depression (p < 0.0001); however, a statistically significant overall effect was not found in the meta-analyses. Current evidence is in favour of the interventions for TMD in improving symptoms of depression and anxiety. However, the effect is statistically uncertain and warrants future studies to enable the best synthesis of the evidence.


Introduction
Temporomandibular disorders (TMD) are commonly defined as a group of orofacial disorders involving the masticatory muscles, the temporomandibular joint (TMJ), and adjacent structures with traumatic, neoplastic, and/or musculoskeletal disorders as aetiology [1,2]. Patients often present with a wide and complex range of clinical conditions, including painful conditions, such as myalgia, arthralgia, referred pain, and headache attributed to TMD, and non-painful conditions, such as disc displacement, limited opening, degenerative joint disease, and subluxation [3].
TMD affects 5-15% of adults in general, as reported in different studies, while TMDrelated symptoms have been reported to be up to 50% of adults [4]. A recent systematic review and meta-analysis has reported the overall prevalence of TMD diagnosed by the research diagnostic criteria (RDC/TMD) or diagnostic criteria (DC/TMD) to be approximately 31% for adults and elderly [5].
TMD is a common orofacial pain disease, which affects a significant percentage of the population, yet its diagnosis and management remain a challenge. There is a lack of consensus in many aspects because of its multifactorial aetiologies. Although the aetiology of TMD is complex and still not clearly understood, it is generally believed to comprise of biological, psychological, and social factors [6,7]. Therefore, it is important to also consider period of COVID-19 pandemic. Submissions which passed a basic automated check were published automatically after 30 days of waiting time, in order to allow the PROSPERO team to focus on COVID-19 related reviews. Eligibility of this protocol was not checked by the PROSPERO team before this study was commenced. ‡ The registered protocol was amended to also include studies without control groups in order to increase the variety of studies to review).

Study Selection
Population Studies reporting adult patients diagnosed with TMD using the RDC/TMD (Axis I and/or Axis II) or its revisions or the new DC/TMD instruments were included. Studies of patients diagnosed with pain disorders other than TMD were excluded.
Intervention All standard treatment options for TMD identified with the goal to improve the disease by reducing pain and/or improving jaw function were included if they were systematically delivered to the subjects according to a pre-defined algorithm or protocol and were started and completed during the perioperative period of the studies. These included conservative options, including medications (such as analgesics, non-steroidal anti-inflammatory drugs, and muscle relaxants), occlusal appliances of various designs, physiotherapy (such as muscle training and massage), changing of behaviour (soft diet and rest), minimally invasive options (such as arthroscopy, arthrocentesis and intra-articular injections, and open joint surgical options (such as disc repositioning procedures, removal of osteophytes, removal of pathologic tissue, biopsy of the TMJ and alloplastic replacement of the TMJ). Botox injection, acupuncture, extracorporeal shock wave therapy, and laser auriculotherapy, which are currently not considered standard treatment options of TMD, were excluded. Psychological interventions, such as anti-depressants, counselling, stress coping strategies, etc., were not defined as interventions for TMD in this review.
Controls Studies that have reported comparative groups of subjects receiving no treatments, placebo treatments, or interventions other than the standard treatment options for TMD mentioned above were categorised as studies with control groups. These comparative groups were analysed under the same subgroup in the meta-analysis.
Outcome Studies included had to report on psychological outcome regarding the severity of anxiety or/and depression. Assessment tools of anxiety included the State-Trait Anxiety Inventory (STAI), Beck Anxiety Inventory (BAI), Hospital Anxiety and Depression Scale (HADS), General Anxiety Disorder-7 (GAD-7) and SCL-90, while those of depression included the Beck Depression Inventory (BDI), Center for Epidemiologic Studies Depression Scale (CES-D), Geriatric Depression Scale (GDS), Patient Health Questionnaire-9 (PHQ-9), HADS and SCL-90.

Search Strategy
Electronic search was carried out in databases, including Pubmed, Web of Science, Medline, Cochrane Library, and Scopus. The literature search was constructed around search terms for "TMD", "depression", and "anxiety" (Table 1). No restrictions were considered regarding publication year or language. Titles and/or abstracts were reviewed after the elimination of duplicates to exclude seemingly irrelevant articles. Manual search was then performed through the bibliographical references of these articles. These potentially relevant articles were further screened by applying the inclusion and exclusion criteria mentioned above by two independent reviewers. A third independent reviewer (a senior researcher) was consulted on any cases of persisting disagreement. The total search of all databases was performed within March 2022. Pubmed ("Temporomandibular Joint Disorders" or "Temporomandibular joint disorder" or "TMJ Disorders" or "TMJ Disorder" or "Temporomandibular Disorders" or "Temporomandibular Disorder" or "Temporomandibular Joint Diseases" or "Temporomandibular Joint Disease" or "TMJ Diseases" or "TMJ Disease" or "Temporomandibular joint dysfunction syndrome" or "Temporomandibular joint pain" or "Temporomandibular pain" or "TMD" or "Craniomandibular Disorders" or "Craniomandibular Disorder" or "Orofacial Pain" or "Craniofacial pain") AND ("Depression" or "depressive disorders" or "depression symptoms" or "anxiety" or "mood disorders" or "psychological distress")

1285
Web of Science ("Temporomandibular Joint Disorders" or "Temporomandibular joint disorder" or "TMJ Disorders" or "TMJ Disorder" or "Temporomandibular Disorders" or "Temporomandibular Disorder" or "Temporomandibular Joint Diseases" or "Temporomandibular Joint Disease" or "TMJ Diseases" or "TMJ Disease" or "Temporomandibular joint dysfunction syndrome" or "Temporomandibular joint pain" or "Temporomandibular pain" or "TMD" or "Craniomandibular Disorders" or "Craniomandibular Disorder" or "Orofacial Pain" or "Craniofacial pain") AND ("Depression" or "depressive disorders" or "depression symptoms" or "anxiety" or "mood disorders" or "psychological distress") 1387 Medline (Temporomandibular Joint Disorders or Temporomandibular joint disorder or  TMJ Disorders or TMJ Disorder or Temporomandibular Disorders or  Temporomandibular Disorder or Temporomandibular Joint Diseases or  Temporomandibular Joint Disease or TMJ Diseases or TMJ Disease or  Temporomandibular joint dysfunction syndrome or Temporomandibular joint  pain or Temporomandibular pain or TMD or Craniomandibular Disorders or  Craniomandibular Disorder or Orofacial Pain or Craniofacial pain) and (Depression or depressive disorders or depression symptoms or anxiety or mood disorders or psychological distress) 1027 Cochrane ("Temporomandibular Joint Disorders" or "Temporomandibular joint disorder" or "TMJ Disorders" or "TMJ Disorder" or "Temporomandibular Disorders" or "Temporomandibular Disorder" or "Temporomandibular Joint Diseases" or "Temporomandibular Joint Disease" or "TMJ Diseases" or "TMJ Disease" or "Temporomandibular joint dysfunction syndrome" or "Temporomandibular joint pain" or "Temporomandibular pain" or "TMD" or "Craniomandibular Disorders" or "Craniomandibular Disorder" or "Orofacial Pain" or "Craniofacial pain") AND ("Depression" or "depressive disorders" or "depression symptoms" or "anxiety" or "mood disorders" or "psychological distress") 237 Scopus ("Temporomandibular Joint Disorders" OR "Temporomandibular joint disorder" OR "TMJ Disorders" OR "TMJ Disorder" OR "Temporomandibular Disorders" OR "Temporomandibular Disorder" OR "Temporomandibular Joint Diseases" OR "Temporomandibular Joint Disease" OR "TMJ Diseases" OR "TMJ Disease" OR "Temporomandibular joint dysfunction syndrome" OR "Temporomandibular joint pain" OR "Temporomandibular pain" OR "TMD" OR "Craniomandibular Disorders" OR "Craniomandibular Disorder" OR "Orofacial Pain" OR "Craniofacial pain") AND ("depression" OR "depressive disorders" OR "depression symptoms" OR "anxiety" OR "mood disorders" OR "psychological distress")

Data Management
The full texts of the articles included were retrieved. Detailed data were extracted from articles independently by two authors according to the data collection form, including information on the author, year of publication, country of publication, study design, size of the population at baseline, characteristics of the population (age at baseline, distribution of experimental, and control groups), duration of follow-up, diagnostic tools of TMD, types of interventions for TMD, outcome measure of TMD intervention, assessment tools of anxiety or/and depression, number of subjects included in the analysis (number of subjects in total, experimental, and control groups), change in treatment outcome of TMD, and severity of anxiety or/and depression before and after interventions.

Assessment of Risk of Bias and Quality Evaluation
Risks of bias were independently rated by two reviewers based on version 2 of the Cochrane risk-of-bias tool for randomised trials (RoB 2) for randomised controlled studies, based on five domains: bias arising from the randomization process; bias due to deviations from intended interventions; bias due to missing outcome data; bias in measurement of the outcome; and bias in selection of the reported result. A risk-of-bias judgement was reached for each domain, then an overall judgment, by assigning one of the three levels: low risk of bias; some concerns; or high risk of bias [15].
A modified Newcastle-Ottawa Quality Assessment Scale was designed to evaluate the quality of all studies included in this review, with reference to the original assessment scale for cohort studies [16]. A "star system" was employed to judge each study based on three main domains: the selection of the sample, the comparability of the groups, and the ascertainment of the outcome. A maximum of three stars for "Selection", one star for "Comparability", and three stars for "Outcome", which made up a maximum of seven stars that could be scored by each study. This modified questionnaire was designed to provide a quick and direct critical appraisal of the included studies. A study with a total score of 6-7 was categorised as good quality, 3-5 as fair quality, and 0-2 as poor quality. The detailed questionnaire is available in Appendix A.
A third independent reviewer (a senior researcher) was consulted on any discrepancies until consensus was reached.

Data Analysis
The meta-analyses were performed using the Review Manager (RevMan) 5 software (Version 5.4, The Nordic Cochrane Centre, Copenhagen) when at least two studies reporting specific outcomes were available. A fixed effects model was employed because only a small number of studies (i.e., less than five) were eligible to be included in each analysis [17,18]. All p-values were reported, and p ≤ 0.05 was considered as being statistically significant.

Meta-Analyses including Only Studies with Control Groups
The effects of interventions for TMD on depression and anxiety, compared to control interventions, were analysed.
Standardised mean difference (SMD) was used as a summary statistic in the metaanalysis since all studies assessed the same outcome, but with various measurement tools (for example, Costa et al. [19] used HADS, while Alajbeg et al. [20] used PHQ-9 in measuring the degree of depression). A SMD allowed standardization of the results of various studies to a uniform scale for analysis. It is calculated as the difference in mean outcomes between the intervention and control groups, divided by the standard deviation (SD) of the outcome among participants, with 95% confidence intervals (CIs) [15]. When the SDs were unavailable, they were estimated by calculation, using standard errors, Cls, t-values, interquartile deviations, and/or the correlation coefficient [15,21]. The correlation coefficient was obtained from calculation using reported data in Alajbeg et al.'s study [18], which was reported in considerable detail. The mean differences, when not reported, were calculated by subtracting the post-intervention measurement from the baseline measurement.
Measurements taking the closest to the beginning and the end of the interventions were chosen for calculation when more than one baseline and/or post-intervention measurement was reported.
A positive SMD was defined to represent the beneficial effects of interventions for TMD compared to the control intervention for all outcomes (e.g., improvement in the levels of pain, depression, and/or anxiety). A combined SMD was computed in RevMan when there were more than one intervention group (for example, in Melo et al.'s study [22], there were three intervention groups: occlusal splint, manual therapy, and combined therapy) using the mean difference and SD of each group [15]. Improvement was defined as reduction in the levels of pain, depression, and/or anxiety in all statistical analyses in this review.
The overall effect size was evaluated by interpreting the SMDs using the Cohen's categories, where SMD = 0.2 to 0.5 represents a small effect, SMD = 0.5 to 0.8 a moderate effect, and SMD > 0.8 a large effect [23].
The certainty of the evidence of each outcome was evaluated by the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach by two independent reviewers. Five GRADE considerations were used for assessment, including risk of bias, consistency of effect, imprecision, indirectness, and publication bias [15,24].

Assessment of Heterogeneity
The statistical heterogeneity was assessed by a chi-squared (χ 2 ) test and inconsistency (I 2 ) statistics. A rough guide to interpret I 2 was as follows: 0 to 40%: might not be important; 30 to 60%: moderate; 50 to 90%: substantial; and 70 to 100%: considerable. Considering the low power of the χ 2 test when only a few studies were included in an analysis, a p-value of ≤0.10 was used to indicate significant heterogeneity.

Narrative Analysis including All Studies
Narrative syntheses of the mean difference between the outcomes before and after interventions in all studies (including those without control groups) were conducted by obtaining the mean change and standard error (SE) in each intervention group. When the SEs were unavailable, they were estimated using the SDs and the sample size of the groups [15]. The findings were interpreted with caution because any placebo effect or effects due to background inclusion were not excluded in these analyses. Neither judgement of the overall effect size nor the certainty of evidence was derived to eliminate possible misinterpretations.

Literature Search
5592 records were retrieved through the electronic search, and 2408 records were screened after the elimination of duplicates. After the review of titles and/or abstracts, 2386 irrelevant records were excluded because their diagnoses for TMD were not by DC/TMD or RDC/TMD or their variations, and/or there were no interventions for TMD carried out. Out of the 22 full texts reviewed, 12 of them were excluded after being assessed for eligibility because either psychological outcomes were not reported [25][26][27][28][29][30][31], or no standard interventions for TMD were delivered [32][33][34][35][36]. No additional records were retrieved after manual search through the reference lists of identified articles. Among the 10 studies (8 RCTs and 2 non-randomised clinical trials) included for qualitative review, 1 RCT [37] was excluded from any quantitative analyses because of insufficient statistical details. A total of 9 studies with 713 patients were included in the narrative analysis. Three RCTs were further excluded from the meta-analysis because either all subjects received interventions for TMD, including the control group (i.e., conservative treatments for TMD) [38], or the assigned interventions were not considered to be standard treatment options [39,40]. Finally, 4 RCTs with 203 patients were eligible and included in the metaanalysis ( Figure 1).
Intervention characteristics All included studies delivered conservative treatments for TMD, with or without control groups. Most of them used occlusal splint as the major intervention, with adjunct diet and lifestyle modification. A thin (ranged from 1.5-2.5mm), full-coverage upper hard acrylic splint, with even occlusal contact and a canine/anterior guidance occlusal scheme, to be worn only during sleep, was the most common protocol [19,20,40]. One study required patients to wear splints for upper or lower arches at all times, except during meals [43]. One study required patients to wear a splint during the day and/or night [22]. One study did not specify the design of splints [39]. Four studies used massage, a warm pack, and/or cryotherapy at masticatory muscles as interventions [22,37,38,42]. One study used anti-inflammatory medications as the only standardised intervention for TMD [41]. The duration of treatment ranged from 1-6 months (median: 2 months). One study did not specify the duration of treatment [41].
Among the eight RCTs included, only five studies fulfilled the definition of control group in this review. One RCT compared the occlusal splint to the placebo splint with the same wearing schedule [20]. The other four RCTs compared interventions for TMD to other non-standardised treatments, including counselling [19,22], self-care protocol [37], and aerobic exercise [42].

Quality of Studies
The quality of the 10 studies included is summarised and presented in Table 3. Three studies [20,22,42] were judged as "good" quality, six studies [19,[37][38][39][40]43] were judged as "fair quality", and one study [41] was judged as "poor quality". Most of the studies that were judged as "poor" or "fair quality" were due to the lack of representativeness of the sample, small sample size, inadequate follow-up period, or lack of description to data lost. Total score of 6-7: good quality; 3-5: fair quality; and 0-2: poor quality. *, **, *** Represents the score awarded in each section.

Risk of Bias in Studies Included in the Meta-Analysis
Only one RCT [20] was judged to have low risk of bias, one RCT [22] was judged to have some concern of bias, while the other two RCTs [19,42] were judged to have high risk of bias. The summary and description of the risk of bias assessment is presented in Figure 2. All studies had a low risk of bias in the measurement of the outcome, as they used common and standardised screening tools for the assessment of anxiety and/or depression with adequate description. The risk of bias in the randomization process was somewhat high because one RCT [22] did not report on adequate allocation concealment, and one RCT [42] did not allocate participants in a randomised manner, but according to participants' preferences. The risk of bias in missing outcome data was high in one RCT [19] because of a high dropout rate of 32%, in which the number of dropped-out participants doubled in the control group compared to the test group, which was likely to induce bias in the result. The risk of bias in the selection of the reported result was generally of some concern or high because the numerical results reported in most of the studies were likely to be selected, such as the mean difference between the test and control groups were not always reported.

Narrative Analysis
Nine out of ten studies provided sufficient data regarding anxiety and/or depression

Narrative Analysis
Nine out of ten studies provided sufficient data regarding anxiety and/or depression to evaluate the overall effects of interventions over time, without controlling for the placebo effect for narrative analyses. Therefore, the results shall be interpreted with caution.
Anxiety Four studies provided sufficient data regarding anxiety for the narrative analysis [19,20,22,42]. The combined data of the 129 participants who received interventions for TMD showed a statistically significant improvement in the symptoms of anxiety (SMD = 2.15; 95% CL 1.66 to 2.65; p < 0.00001). Very low and statistically insignificant heterogeneity was observed between studies (Heterogeneity: I 2 = 0%; χ 2 = 2.94; p = 0.40) (Figure 3). There was an overall significant mean improvement in symptoms of anxiety after interventions for TMD [19,20,22,42].

Figure 4.
Forest plot of subgroup analysis of effects of intervention for TMD on symptoms of depression according to different interventions. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. There was an overall significant mean improvement in symptoms of depression after interventions for TMD, as well as significant subgroup differences between different interventions [19,20,[38][39][40][41]43]. Table 4. Summary of the narrative analysis of effects of interventions for TMD on severity of anxiety and depression and sensitivity analyses.

Meta-Analysis
Four RCTs out of ten studies provided sufficient data regarding anxiety and/or depression for the meta-analysis to evaluate the overall effects of intervention over time, with the control of placebo effects. The summary of the results of the overall effects of intervention on anxiety and depression compared with the control group and the sensitivity analysis are presented in Table 5.

Anxiety
All four RCTs reported data regarding symptoms of anxiety, as evaluated using the GAD-7 questionnaire [20,42] or HADS [19,22]. The level of anxiety was assessed by two screening tools, HADS and BAI, in Melo's RCT [22]. Data evaluated using HADS was extracted for this meta-analysis to minimise the heterogeneity between different screening tools. Analysis of these 4 studies (139 participants in the intervention arm and 64 participants in the control arm) showed no significant difference between the 2 groups (SMD = 0.29; 95% CL −0.02 to 0.60; p = 0.06) ( Figure 5).
Diagnostics 2023, 13, x FOR PEER REVIEW 18 of 24 Figure 5. Forest plot of effects of intervention for TMD on symptoms of anxiety after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of anxiety favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20,22,42].

Depression
Only two RCTs reported sufficient data regarding symptoms of depression, as evaluated using the Patient Health Questionnaire-9 [20] or the HADS [19]. Analysis of these 2 studies (43 participants in the intervention arm and 32 participants in the control arm), showed no significant differences between the 2 groups (SMD = 0.40; 95% CL −0.06 to 0.87; p = 0.09) ( Figure 6). Figure 6. Forest plot of effects of interventions for TMD on symptoms of depression after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of depression favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20].

Figure 5.
Forest plot of effects of intervention for TMD on symptoms of anxiety after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of anxiety favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20,22,42].

Depression
Only two RCTs reported sufficient data regarding symptoms of depression, as evaluated using the Patient Health Questionnaire-9 [20] or the HADS [19]. Analysis of these 2 studies (43 participants in the intervention arm and 32 participants in the control arm), showed no significant differences between the 2 groups (SMD = 0.40; 95% CL −0.06 to 0.87; p = 0.09) ( Figure 6).
Sensitivity analysis A sensitivity analysis was performed by repeating the meta-analysis regarding the effect of interventions on symptoms of anxiety after removing two studies [19,42] with a high risk of bias. The difference between the intervention and control groups remained insignificant (SMD = 0.11; 95% CL −0.3 to 0.51; p = 0.06). Sensitivity analyses were performed by repeating the meta-analyses, according to the assessment tools used for anxiety. The mean differences for each tool were individually analysed. However, there were still no significant differences between the intervention and control groups observed. Since all the studies delivered occlusal splints in their intervention arms, except Moleirinho-Alves et al. [42], which used massage and warm pack or cryotherapy as intervention, the analysis was repeated after removing its influence. Similarly, no significant differences between the intervention and control groups were observed ( Table 5).
Quality of evidence The level of certainty of the evidence was judged in the GRADE approach. Despite the low heterogeneity between studies in the analyses regarding both anxiety and depression, there were considerable risks of bias due to the generally small sample size in all studies, lack of blinding in both participants and clinicians in most studies, and high attrition rate in some studies. Therefore, it was deemed appropriate to downgrade the certainty of the evidence by two levels, from high to low, due to the imprecision of the results and the study limitations. Figure 5. Forest plot of effects of intervention for TMD on symptoms of anxiety after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of anxiety favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20,22,42].

Depression
Only two RCTs reported sufficient data regarding symptoms of depression, as evaluated using the Patient Health Questionnaire-9 [20] or the HADS [19]. Analysis of these 2 studies (43 participants in the intervention arm and 32 participants in the control arm), showed no significant differences between the 2 groups (SMD = 0.40; 95% CL −0.06 to 0.87; p = 0.09) ( Figure 6). Figure 6. Forest plot of effects of interventions for TMD on symptoms of depression after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of depression favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20].
Sensitivity analysis A sensitivity analysis was performed by repeating the meta-analysis regarding the effect of interventions on symptoms of anxiety after removing two studies [19,42] with a high risk of bias. The difference between the intervention and control groups remained insignificant (SMD = 0.11; 95% CL −0.3 to 0.51; p = 0.06). Sensitivity analyses were performed by repeating the meta-analyses, according to the assessment tools used for anxiety. The mean differences for each tool were individually analysed. However, there were still no significant differences between the intervention and control groups observed. Since all the studies delivered occlusal splints in their intervention arms, except Moleirinho-Alves et al. [42], which used massage and warm pack or cryotherapy as intervention, the analysis was repeated after removing its influence. Similarly, no significant differences between the intervention and control groups were observed (Table 5). Table 5. Summary of effects of interventions for TMD on severity of anxiety and depression and sensitivity analysis.

SMD (95% Cl)
p Value Heterogeneity I2; χ2; P Figure 6. Forest plot of effects of interventions for TMD on symptoms of depression after controlling for placebo effect. Box size reflects study size. The diamond at the bottom reflects the overall pooled effect with a 95% confident interval. Positive SMD reflects effect on improving symptoms of depression favouring interventions for TMD over control. No statistically significant differences were observed between the two groups [19,20].

Summary of the Findings
This systematic review and meta-analysis explored the best available evidence on the effectiveness of interventions for TMD on psychological outcome regarding symptoms of anxiety and depression in patients diagnosed with TMD. A total of 10 studies fulfilled the inclusion criteria and underwent qualitative analysis, while 9 studies provided sufficient data for the narrative analysis, and 4 RCTs for the meta-analysis. The results in all the studies generally suggested significant improvement in anxiety and depression after interventions for TMD, which is further demonstrated in our narrative analysis by an overall statistically significant reduction in the level of anxiety and depression. An obvious tendency of overall effects on improving symptoms in both depression and anxiety favouring interventions for TMD over control was observed in the meta-analyses; however, the effectiveness was found not statistically significant regarding a 95% confident interval. Furthermore, the subgroup analysis for the treatment effect on the improvement in depression regarding different interventions showed statically significant group differences, which in turn suggested that different interventions significantly modified the effect on the improvement in symptoms of depression. Heterogeneity was observed within subgroups, which suggested possible background factors that contributed to the varied results. In the sensitivity analysis, no heterogeneity was observed within studies using the same psychological assessment tools, suggesting that the use of various psychological assessment tools might be the reason for the heterogeneity.

Role of Interventions for TMD in Improving Anxiety and Depression
The statistically significant effect observed in the narrative analysis suggests a beneficial effect of interventions for TMD on reducing levels of depression and anxiety, regardless of the types of interventions given. The mechanism of this beneficial effect was suggested to be associated with the relationship between pain and TMD. Previous studies have indicated the mutual interaction between pain and psychological distresses [14,44] Successful therapeutic treatments in patients with TMD are suggested to have a positive effect in improving symptoms of anxiety and depression by pain management strategies [45].

Statistical Significance Not Found in Treatment Effect When Compared to Control Group
The overall treatment effect tended to favour interventions for TMD over the control in improving symptoms of depression and anxiety. However, it was not found to be statistically significant. This finding was likely because only a limited number of eligible studies were included in this meta-analysis [46]. Furthermore, most studies [19,22] provided treatments such as counselling to patients in the control group; only one study [20] used placebo splint in the control setup. These non-standardised interventions have likely resulted in a positive effect on the psychological outcomes, which have in turn weakened the effects of the standardised interventions shown in the statistics.

Implication for Clinical Practice
This review suggested a supportive role of interventions for TMD in improving anxiety and depression. It is demonstrated in the Turner et al. [38] and Costa et al. [19] studies that the combination of treatments for TMD and psychological interventions, such as cognitive-behavioural therapy and counselling, resulted in the best outcome. They believed the involvement of a psychological approach allowed relaxation and better paincoping strategies which worked hand-in-hand with the standardised interventions in the management of TMD. Previous studies also supported the implication of psychosocial interventions for chronic orofacial pain [47]. On the other hand, it is also important for psychologists to be aware of any signs of TMD in their patients. A timely referral to oral surgeons might help in the management of psychological distress of their patients. A multidisciplinary approach is suggested to best manage this multifactorial illness.

Implication for Future Research
Future RCTs should ensure the high quality of the methodology and reporting, including larger sample sizes, allocation concealment, control groups with no treatments or placebo treatments, and intention-to-treat analyses. Meta-analyses could be repeated when there are more eligible studies available to improve generalization and obtain an accurate overall treatment effect. Future RCTs could be conducted to compare the effectiveness between standardised interventions for TMD; psychological interventions; and combinations of both and no treatments, on both pain control and psychological outcomes. This requires contributions of expertise from both oral surgery and psychology.

Strengths and Limitations of This Review
There were several limitations in this review. First, only a small number of studies could be included in this meta-analysis. The pooled sample size was relatively small to identify significant relationships within the dataset.
Secondly, high heterogeneity existed in the various assessment scales of anxiety and depression applied in different studies. Multiple cut-off points were used among studies that used the same assessment tools. The duration of intervention varied, and measurements of outcome parameters were obtained at different time-points across studies. These have made direct comparison of the study outcomes difficult. The summary statistics required for meta-analysis were unavailable in most studies, and much statistical estimation was performed, which might induce inaccuracy in the analysis. Furthermore, the low methodological quality of the available RCTs might also include bias. Since all the assessment tools of anxiety and depression relied on questionnaires completed by patients, blinding of outcome measurements became impossible. Some studies did not conduct intention-to-treat, but rather per-protocol analyses when there were missing data.
In addition, the studies that fulfilled the inclusion criteria, and thus were included in this systematic review, consisted only of a limited array of the currently available treatment options, such as occlusal splint and anti-inflammatory medications. Studies pertaining to other common interventions for TMD, such as intra-articular injection and arthrocentesis, which also fulfil the inclusion criteria of this systematic review, were not found. It is, therefore, not possible to relate the findings of the current systematic review and metaanalysis to those other common interventions for TMD.
Lastly, the patients included in the studies were mostly psychological healthy individuals with symptoms, but not diagnosed with anxiety and depression. The difference before and after interventions might, therefore, be too small to be reflected in the statistics.
Nevertheless, to the best of our knowledge, this is the first systematic review and metaanalysis to evaluate the effectiveness of interventions for TMD in reducing psychological distress. A comprehensive search of available literature was conducted, with an established review methodology applied, to minimise possible bias. Although only a handful of studies could be included in the meta-analysis, we attempted to summarise the best available evidence and identify the current research gap in this topic. This systematic review and meta-analysis serves as an exploratory review, providing a plausible estimate that could be tested in the future in subsequent reviews of the role of interventions for TMD in correcting psychological stress.

Conclusions
This systematic review and meta-analysis have suggested the interventions for TMD may be beneficial in improving symptoms of depression and anxiety, based on the current available evidence. However, the effect is statistically uncertain and warrants future studies to enable the best synthesis of the evidence. Multidisciplinary management, with the input of both the surgeons and the psychologists, is recommended in treating patients presented with TMD and symptoms of psychological distress.

Conflicts of Interest:
The authors declare no conflict of interest.