This multicenter, prospective, double-blind, randomized, sham-controlled, superiority trial was designed to evaluate the efficacy and safety of a DTx intervention in patients diagnosed with TMD in combination with conventional treatment. The trial compared clinical outcomes between a group using the DTx smartphone app and a control group using a sham app over a 6-week period, with assessments across 4 scheduled visits.

The trial was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice guidelines and followed the CONSORT (Consolidated Standards of Reporting Trials) 2025 guidelines. This trial was reviewed and approved by the institutional review boards of both participating centers (Hallym IRB 2024-04-001-006 and HDT IRB 2024-04-001-004) as well as by the South Korean Ministry of Food and Drug Safety (1700). The study was registered on the Clinical Research Information Service (KCT0009493; registration date April 30, 2024).

All participants provided written informed consent before enrollment. Participant data were pseudonymized using study ID numbers and stored in a secure, password-protected electronic data capture system. Only authorized study investigators had access to the dataset, and no personally identifiable information was included in the analysis.

Participants received financial compensation (including transportation expenses) of approximately ₩500,000 (approximately US $350) upon completion of the trial. No deviations from the registered protocol occurred during the study.

The clinical evaluation period was set to 6 weeks based on prior studies, which generally assessed TMD outcomes over shorter durations [15]. Supporting evidence from trials involving cognitive behavioral therapy showed significant reductions in pain scores on a visual analog scale (VAS) over a period of 5.9 to 6 weeks [16]. Assuming a mean VAS improvement of 3.26 mm in the intervention group and 2.06 mm in the sham group, with an SD of 1.579 mm, a 2-tailed α of .05, and 90% power, the required sample size was 38 participants per group (calculated using G*Power [version 3.1, Heinrich Heine University]). Allowing for a 25% dropout rate, the final enrollment target was 102 participants (n=51, 50% per group).

Participants were recruited from outpatient clinics at the departments of oral and maxillofacial surgery at 2 university hospitals in South Korea (Hallym University Sacred Heart Hospital and Hallym University Dongtan Sacred Heart Hospital). Both institutions serve large regional populations and maintain specialized oral and maxillofacial surgery units that provide comprehensive diagnostic and therapeutic services for TMDs. Recruitment was conducted through direct referral during clinical visits, as well as posters and flyers displayed at the clinics.

Participants were eligible if they were adults aged between 19 and 65 years, had a diagnosis of TMD based on the diagnostic criteria for TMDs (DC/TMD) [17], had available clinical and radiographic findings, and provided written informed consent. Participants were excluded if they had structural joint diseases (eg, ankylosis or tumors), traumatic injuries (eg, mandibular fracture), neuromuscular or severe metabolic disorders, neuropsychiatric illnesses, or recent use of sleep medications or illicit substances; were undergoing orthodontic treatment; were pregnant or planning pregnancy; or had participated in another clinical trial within the prior 4 weeks.

Participants were randomly assigned (1:1) to the DTx or sham app group using stratified block randomization by site, implemented through the PLAN procedure in SAS (version 9.4 or higher; SAS Institute Inc). An independent randomization statistician created and safeguarded the allocation list, which was integrated into an interactive web-based response system for automated assignment upon eligibility confirmation.

To maintain blinding, the randomization process was separated from outcome assessment and data analysis. The sham app mimicked the appearance and functionality of the DTx app but lacked therapeutic content. Both participants and investigators were blinded to grouping. Nonblinded clinical research coordinators monitored adherence and adverse events without engaging in outcome assessments. Randomization codes were disclosed only after database freezing to secure data integrity.

All data were securely stored and managed using a validated electronic data capture system with encrypted access. All statistical analyses were effectively conducted in a per-protocol population, defined as participants who completed the week 6 assessment and demonstrated at least 66.7% adherence to the intervention, as minimal engagement with the app would not constitute meaningful exposure to the therapeutic content. For all outcomes, normality was assessed before analysis. Within-group differences from baseline to week 6 were evaluated using paired 2-tailed t tests. Between-group differences in change scores were analyzed using independent t tests. When the assumption of normality was not met, nonparametric Wilcoxon rank sum tests were used. A 2-tailed P value <.05 was considered statistically significant. Missing data were minimal and limited to participants who did not complete the week 6 follow-up. These participants were excluded from the primary per-protocol analysis. An intention-to-treat (ITT) analysis was also conducted, with missing data handled by last-observation-carried-forward imputation. All statistical analyses were performed using SAS (version 9.4).

From June 2024 to June 2025, participants were recruited from 2 institutions. A total of 126 participants were assessed for eligibility, and 102 (81%) were subsequently enrolled and randomly assigned (Hallym University Sacred Heart Hospital—DTx group: n=34, 33%; sham group: n=33, 32%; Hallym University Dongtan Sacred Heart Hospital—DTx group: n=18, 18%; sham group: n=17, 17%). Of the 102 participants, 50 (49%) were assigned to the Clickless DTx group and 52 (51%) to the sham control group, with all participants receiving their allocated intervention. For the primary outcome analysis, we included 44 (88%) of the 50 participants in the DTx group (n=4, 8% were excluded because of protocol noncompliance [defined as intervention adherence below 66.7%], and n=2, 4% were lost to follow-up). In the sham control group, 3 (6%) of the 52 participants were lost to follow-up, leaving 49 (94%) participants for inclusion in the analysis. The participant flow through each stage of the trial is illustrated in Figure 2. No additional item-level missing data were observed among participants who completed the study assessments. Accordingly, the final analysis included 93 participants in the per-protocol population (DTx group: n=44, 47%; sham group: n=49, 53%). This approach ensured that the outcome comparisons reflected participants with meaningful exposure to the intervention. All participants completed the trial without experiencing any serious harm or adverse events. Demographic characteristics were comparable between groups. Full baseline data are summarized in Table 1. The characteristics of participants included in the per-protocol analysis are provided in Multimedia Appendix 1.

The results for the 5 outcome variables are presented in Tables 2 and 3 and illustrated in Figure 3.

At baseline, the mean VAS scores were 50.57 (SD 22.13) in the DTx group and 47.55 (SD 17.41) in the sham group (P=.36). At 6 weeks, significant decreases in VAS scores were observed in both groups compared with baseline (DTx group: mean −33.64, SD 25.34; P<.001; sham group: mean −9.86, SD 24.96; P=.008). The mean VAS score at 6 weeks was significantly lower in the DTx group than in the sham group, with a between-group difference at 6 weeks of −20.76 (95% CI −29.67 to −11.86; P<.001). Additionally, the change in VAS scores from baseline to week 6 was significantly greater in the DTx group than in the sham group, with a between-group difference in change scores of −23.78 (95% CI −34.15 to −13.41; P<.001). These results are illustrated in Figures 3A and 3B.

For MMO, mean baseline measurements were 40.51 (SD 6.94) mm in the DTx group and 39.52 (SD 7.94) mm in the sham group (P=.53). At 6 weeks, mean MMO increased to 46.97 (SD 6.30) mm in the DTx group and 41.09 (SD 8.44) mm in the sham group. The increase was statistically significant only in the DTx group (P<.001). The mean MMO at 6 weeks was significantly greater in the DTx group than in the sham group, with a between-group difference of 5.87 mm (95% CI 2.78-8.97 mm; P<.001). The between-group difference in change from baseline to 6 weeks was 4.36 mm (95% CI 1.92-6.81 mm; P<.001). The MMO results are illustrated in Figures 3C and 3D.

The mean JFLS-20 global scores were 2.58 (SD 1.69) in the DTx group and 2.52 (SD 1.24) in the sham group (P=.74) at baseline. At 6 weeks, both groups showed significant reductions in mean scores: from baseline to 0.86 (SD 1.01) in the DTx group (within-group change: mean −1.72, SD 1.54; P<.001) and to 1.82 (SD 1.47) in the sham group (within-group change: mean −0.70, SD 1.44; P=.001). The reduction was significantly greater in the DTx group, with a between-group difference in change scores of −1.02 (95% CI −1.63 to −0.42; P=.005). The JFLS-20 global score results are illustrated in Figures 3E and 3F.

At baseline, the mean OBC scores were 26.68 (SD 9.13) in the DTx group and 22.12 (SD 9.27) in the sham group (P=.01). At 6 weeks, the mean scores decreased to 18.84 (SD 11.95) in the DTx group and 20.12 (SD 11.41) in the sham group; the decrease was statistically significant only in the DTx group (P<.001). The mean within-group change was −7.84 (SD 12.56) for the DTx group (P<.001) and −2.00 (SD 7.48) for the sham group (P=.07), indicating a greater reduction in the DTx group. The between-group difference in change scores was −5.84 (95% CI −10.18 to −1.50; P=.009). The OBC score results are illustrated in Figures 3G and 3H.

At baseline, the mean PHQ-4 scores were 3.68 (SD 2.30) in the DTx group and 2.78 (SD 2.00) in the sham group (P=.04). At 6 weeks, the mean scores decreased to 2.91 (SD 2.44) in the DTx group and 2.37 (SD 2.43) in the sham group; the reduction was statistically significant only in the DTx group (P=.007). The mean within-group change was −0.77 (SD 1.84) in the DTx group (P=.007) and −0.41 (SD 2.15) in the sham group (P=.20). At 6 weeks, the between-group difference in change scores was −0.36 (95% CI −0.57 to 1.11; P=.42). The PHQ-4 score results are illustrated in Figures 3I and 3J.

The results of the ITT analysis were comparable to those of the per-protocol analysis, with minor variations. Detailed results of the ITT analysis can be found in Multimedia Appendix 3.

This multicenter, double-blind, randomized, sham-controlled trial demonstrated that a DTx intervention, when combined with TAU, led to a significant reduction in pain at 6 weeks compared with sham. Additional benefits were observed in jaw mobility, perceived functional limitation, and oral behaviors, although no significant between-group difference was found in psychological distress. These findings suggest that a DTx may serve as a promising adjunctive intervention to conventional care for TMD, particularly for addressing both physical (axis I) and behavioral (axis II) components of the disorder.

Regarding pain intensity, the DTx group showed a significantly greater reduction in VAS at week 6 compared with the sham group. Moreover, the pain-relieving effect of the DTx appeared earlier and with greater magnitude compared with the sham group. These findings suggest that while conventional TAU alone may result in symptomatic relief over a 6-week period, the addition of a DTx leads to faster and more substantial clinical benefits. This pattern is consistent with previous randomized controlled trials involving DTx for chronic conditions such as hypertension and diabetes, where adjunctive digital interventions yielded superior outcomes [18-21]. Furthermore, our findings are consistent with previous observations that diseases heavily reliant on patient self-management exhibit greater differential treatment effects with adjunctive DTx use [12,22].

In terms of MMO, the DTx group exhibited a larger improvement, likely attributable to the structured exercise modules included in the app. This pattern corresponded with pain reduction and functional improvement, as indicated by the JFLS-20 scores, reflecting consistency among key clinical measures. Notably, pain relief was accompanied by improved functional capacity and subjective perception of functional limitations, demonstrating a coherent pattern across symptomatic and functional outcomes [23]. These results indicate that patients not only experienced pain relief but also subjectively recognized improvements in their condition. The alignment between objective clinical outcomes and patient perception reinforces the therapeutic value of DTx in enhancing both physical function and patient-reported experiences.

OBC outcomes indicated that parafunctional behaviors declined in both groups, with a greater mean reduction in the DTx group than in the sham group, although the between-group difference in change scores did not reach statistical significance. This finding suggests that DTx may effectively address modifiable behavioral risk factors, particularly through structured app-based components. In conventional care, strategies to guide or monitor behavioral changes between clinic visits are limited [8,9,24]. DTx facilitate continuous behavioral support through real-time tracking and feedback, enhancing motivation and leading to more consistent patient engagement [25]. This mechanism is analogous to the role of continuous glucose monitors in diabetes, which are known to improve behavioral adherence and outcomes. Such capabilities are typically not addressed in routine clinical care [26,27].

PHQ-4 scores decreased slightly in both groups, without significant between-group differences. Although the DTx app did not specifically target mental health, improvements in symptom burden may have contributed to secondary psychological benefits. While the effects on mental health were limited in terms of between-group differences, a statistically significant improvement was observed in the DTx group at week 6, suggesting the potential utility of DTx components in addressing psychological burden. Given the complex nature of psychological symptoms, longer and more targeted interventions may be necessary; however, our findings provide preliminary evidence of the feasibility of digital approaches in supporting mental health in TMD [28]. Additionally, for patients presenting with high psychological scores, DTx may offer a practical framework to support tailored interventions by providing structured, evidence-based approaches to mental health management. This principle also applies to behavioral outcomes such as those measured by the OBC, where individuals with high baseline scores may benefit from structured guidance and consistent behavioral feedback that conventional care often lacks [29].

Mechanistically, the DTx platform likely enhanced patient engagement through guided content, self-monitoring, and feedback. These features support self-regulation and may promote sustainable behavioral modification [29]. Prior studies have demonstrated that structured behavioral interventions in TMD, including habit reversal training and self-care education, can lead to meaningful clinical improvements [11,30]. The observed efficacy of the DTx platform in this trial is consistent with these findings, suggesting that digital formats can facilitate and scale such evidence-based strategies. Importantly, these benefits were achieved without additional clinical burden, highlighting the feasibility of DTx integration into routine care.

Methodologically, to our knowledge, this study is among the first to apply a double-blind, sham-controlled design to a DTx trial for orofacial pain. The use of a structurally identical sham app enabled rigorous control of the placebo effect [11]. Additional strengths include centralized randomization, strong adherence, and low rates of missing data (5/102, 5% participants), with only 9 participants overall excluded from the primary outcome analysis (n=5, 56% lost to follow-up and n=4, 44% excluded due to predefined protocol noncompliance).

However, this study has several limitations. First, the relatively short follow-up period limits the ability to assess the durability and sustainability of treatment effects, particularly in the context of chronic conditions such as TMD. Second, although overall adherence was acceptable, individual variation in digital literacy, device access, and environmental factors may have influenced user engagement and the consistency of intervention delivery [31]. Third, outcomes were based on patient-reported measures. However, reliance on patient-reported measures aligns with the prevailing methodology in TMD research, where objective clinical end points remain limited. Pain intensity (measured by the VAS) and MMO are widely used metrics in global TMD trials [15]. Moreover, the DC/TMD Axis II protocol endorsed by the National Institutes of Health prioritizes self-report instruments for psychosocial and behavioral assessment, highlighting their accepted clinical relevance despite the subjectivity [17,32]. Fourth, with the exception of minor baseline differences in OBC and PHQ-4 scores, randomization achieved comparable distributions across groups, as no statistically significant differences were observed in other key variables, including VAS pain, MMO, and demographic characteristics. These minor imbalances were not clinically meaningful and are unlikely to have influenced the study outcomes. The observed differences in OBC and PHQ-4 scores were small relative to their full scale ranges and below established clinical cutoffs. Given the modest sample size, additional adjusted analyses, such as analysis of covariance or mixed models, would have been limited in statistical power. Participants were recruited from 2 institutions, and although group allocation was reasonably balanced, the sample size may still have permitted some random imbalances. Finally, although conventional treatment was maintained across study sites, heterogeneity in clinical practice patterns or patient interactions could not be entirely controlled. This reflects a common challenge in real-world clinical trials and underscores the need to balance ecological validity with protocol standardization [33]. Nevertheless, such pragmatic variability is inevitable in studies intended for widespread clinical application. For a DTx model to be broadly applicable in TMD care, analyses such as those performed in this study are essential at the initial stage. Further research should validate these findings in more narrowly defined patient subgroups and clinical scenarios to refine and tailor DTx implementation.

Clinically, DTx represent a practical and scalable adjunct to conventional care, especially for managing behavioral and psychosocial components of TMD that are often underaddressed in routine clinical settings. The ability of DTx to deliver structured interventions remotely can help overcome limitations in provider availability and patient access, while also promoting consistent engagement outside of clinic visits [34]. These features are particularly relevant in chronic conditions such as TMD, where continuous self-management plays a critical role [35].

Future investigations should include extended follow-up periods to evaluate the long-term clinical outcomes and sustainability of DTx therapeutic effects. In addition, the incorporation of AI-driven analytics into DTx platforms holds promise for optimizing patient engagement through real-time feedback, symptom-based personalization, and adaptive content delivery. Such enhancements may improve both adherence and clinical efficacy across heterogeneous patient populations.