Temporomandibular disorders (TMDs) are common chronic orofacial conditions characterized by pain and functional limitations. Digital therapeutics (DTx) have emerged as a nonpharmacologic treatment modality for TMD and have shown efficacy in prior studies. However, the mechanisms by which DTx produces clinical benefit—whether through direct therapeutic content, induced behavioral modification, or simply higher user engagement—are not well characterized in TMD populations.
This secondary analysis reports a post hoc investigation into the behavioral and clinical mechanisms that may underlie response to a DTx intervention for TMD. The analysis synthesizes multiple data sources to probe mediation and moderation of treatment effects.
The stated objective was to investigate behavioral mechanisms, responder profiles, and moderators of clinical response to a DTx intervention for TMD. The analysis aimed to integrate self-reported outcomes, server-derived engagement metrics, and clinician-rated measures to determine whether behavioral modification or the intensity of DTx engagement drives therapeutic benefit.
This work is a post hoc secondary analysis of a multicenter, double-blind, sham-controlled randomized superiority trial conducted at two tertiary care centers in South Korea. The per-protocol cohort used for these analyses included 93 participants: 44 allocated to the active DTx arm and 49 to the sham arm.
The investigators combined three complementary data streams for this secondary analysis:
The abstract indicates that five complementary analyses were applied to the dataset, beginning to list a “responder logistic regressio” approach, suggesting the use of responder-based logistic regression among other regression or profile methods. The full list and specification of these five analytic approaches are not available in the provided text.
According to the abstract, the authors planned multiple complementary analytic strategies to identify mechanisms and responder profiles. The approaches referenced include responder-focused logistic regression and other methods intended to model mediation and moderation. The integration of server-derived engagement data with clinical and self-report measures is a strength for disentangling whether observed benefits reflect true behavioral change or correlate with mere adherence or intensity of use.
Because this is a secondary, post hoc analysis, the methods likely explored associations and potential mediators rather than testing a single prespecified mediation model; however, precise modeling choices, covariates, thresholds for responder status, and prespecified moderators are not reported in the available source text.
The abstract text available here is truncated and does not contain the detailed numerical results, effect estimates, statistical significance values, or explicit mediation findings. The only concrete reported data are the trial design elements (multicenter, double-blind, sham-controlled randomized superiority trial), the per-protocol cohort size (n=93), and allocation numbers (DTx n=44; sham n=49).
Therefore, the following were not reported in the provided source text and cannot be inferred:
Those details are required to draw conclusions about mediator status for behavioral change or engagement and must be obtained from the full article.
The question addressed by this secondary analysis—whether DTx benefits patients primarily by producing behavioral modification or because engaged users are intrinsically more likely to improve—has practical implications. If behavioral change mediates benefit, DTx can be optimized to target and reinforce specific behaviors. If engagement intensity rather than content drives outcomes, efforts might focus on adherence strategies or blended care models. However, because the provided abstract text omits results, actionable clinical recommendations cannot be derived from the available source.
Key limitations of interpreting this source excerpt include substantial truncation of the abstract and absence of reported results. The source does not provide the full set of methods, analytic specifications, or any numerical outcomes. As such, this summary reports only trial design, the analytic intent, and cohort counts. Any detailed findings, statistical inferences, or clinical conclusions were not reported in the provided text and cannot be invented.
For clinicians or researchers seeking to apply these findings, consult the full published article (JMIR Mhealth Uhealth; DOI: 10.2196/104264; PMID: 42691400) for complete methods, full analyses, and results before changing practice or designing follow-up studies.