Study design and setting

This comparative cross-sectional study conducted in New Delhi, India and reported in accordance with the STROBE guidelines (Supporting file 1). The city of New Delhi was chosen because it ranks among the most noise polluted capitals around the globe. Daytime road traffic noise levels along commercial and arterial corridors have been reported to exceed 70–85 dBA [17]. Participants were enrolled from one high traffic location and one low traffic location within New Delhi, India. Briefly the high traffic location was defined by sustained traffic flow and measured noise levels ≥65 dBA. The low traffic location consisted of residential neighbourhoods with near constant low-volume traffic flow and ambient noise <65 dBA. This 65 dBA cut off corresponds to the threshold established by the WHO Environmental Noise Guidelines (2018), above which road traffic noise is associated with adverse health outcomes [18].

A sampling frame of households in each area was obtained from door to door household listing conducted by the study team. Households were selected using systematic sampling with a sampling interval of every 5^th household, starting from a randomly selected household in each area, until the target sample size was reached. Within each selected household, one eligible adult was chosen using the kish grid method. Groups were additionally frequency matched on age and sex at the time of recruitment. Application of exclusion criteria to both groups reduced confounding by exclusion of individuals with major comorbidities affecting the TMJ. Objective noise measurement at participants residences (rather than delf reported exposure) reduce information bias. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Research Ethics Committee at King Khalid University, Saudi Arabia [KKU-11-2026-5]. The study protocol was developed, and all participants gave their written informed consent for inclusion before they participated in the study.

Participants

Three hundred adult participants were recruited and enrolled into the study (n = 150 per group). Participants were required to be between the ages of 18–60 years old and reside at the subject address for at least two consecutive years. Participants were excluded from participation if they had previous surgery or trauma to the TMJ or craniofacial structures, neurologic disorders, were taking centrally acting drugs known to alter muscle tone, or had severe systemic disease. Groups were matched on frequency by age and sex. Educational status was self-reported and classified into two categories: “primary/secondary” (no formal schooling through completion of secondary/high school) and “tertiary” (any diploma, undergraduate, or postgraduate qualification beyond secondary school). An a priori power calculation demonstrated that 138 participants per group (276 total) were required to detect an OR of 2.0 with 80% power at α = 0.05 assuming 40% baseline prevalence of TMD. This minimum was inflated to 150 participants per group (300 total), a buffer of approximately 9%, to safeguard statistical power against minor data quality exclusions during analysis. A separate inflation for anticipated non response was not incorporated into this calculation, as prior community engagement in the target neighbourhoods indicated a very low expected refusal rate among eligible residents; this was consistent with the 100% consent rate ultimately observed among eligible participants. The larger initial approach of 412 adults (206 per area) instead reflects the abticipated proportion of individuals expected to be ineligible during door to door screening (approximately 27%, based on the ≥ 2 year residency and age criteria), rather than anticipated non participation.

Four hundred twelve adults were approached at the target neighbourhoods (206 per area). Of these, 112 were excluded for not meeting eligibility criteria (<2 years residence [n = 68]; age outside range of 18--60 years [n = 21]; previous TMJ surgery/trauma [n = 15]; other [n = 8]), leaving 300 eligible adults, all of whom consented to participate (response rate among eligible individuals = 100%). No eligible individual refused participation. Three hundred participants completed signed informed consent (150 per group) and were enrolled. Zero participants were lost to follow-up, and all 300 completed the study outcome assessments.

Noise exposure measurement

Noise exposure was assessed at the participants’ primary residence. Ambient noise levels were recorded using a calibrated Type-2 Integrating Sound Level Meter compliant with IEC 61672−1. Measurements were made at six time points throughout day (06: 00–08:00, 09:00–11:00, 12:00–14:00, 15:00–17:00, 18:00–20:00, 21:00–22:00) while positioned 1.5 metres from the structure façade at ear height. The energy-averaged LAeq,24h was calculated and recorded for each participant per ISO 1996–2:2017 standard. Participants were assigned to either high or low noise exposure groups based on their LAeq,24h measurements. High noise exposure was defined as LAeq,24h > 65 dBA and low noise exposure was defined as LAeq,24h ≤ 65 dBA using cut points established by WHO Environmental Noise Guidelines (2018) [19].

Clinical examination and TMD diagnosis

All clinical examinations were performed by two calibrated examiners (intra-examiner κ = 0.84, 95% CI: 0.70–0.98; inter-examiner κ = 0.79, 95% CI: 0.65–0.93). Diagnosis of TMD was made using the DC/TMD Axis I criteria [20]. Briefly, clinical evaluation consisted of standardised assessment of mandibular range of motion, joint auscultation and palpation for clicking and crepitus, bilateral extraoral and intraoral muscle palpation at standardized sites. Based on clinical findings, participants were determined to be TMD present or absent according to DC/TMD Axis I diagnostic criteria. It is noted that DC/TMD Axis I criteria incorporated both signs (e.g., crepitus clicking, limited range of motion) and symptoms (e.g., pain on palpation, pain on jaw movements) both were assessed as part of the standardised clinical examination. An additional ordinal severity grading of TMD was completed using the Helkimo Clinical Dysfunction Index (Di 0–III). DC/TMD Axis II psychosocial screening included Patient Health Questionnaire-4 (PHQ-4) a 4-item screener (2 anxiety items, 2 depression items) scored 0–3 per item, yielding a total range of 0–12; a score ≥6 was defined as indicating clinically relevant anxiety/depression, consistent with established distress thresholds. The Graded Chronic Pain Scale (GCPS) was also administered, classifying participants into five hierarchical grades (0 = no pain; I–II = low/high pain intensity with low disability; III–IV = moderate-to-severe pain-related disability) based on pain intensity and disability subscale scores. These psychosocial instruments were included as covariates in regression models.

Classification of occlusal status (Angle’s classification) was completed with malocclusion being defined as any angle classification other than class I ideal occlusion. Pain intensity in the TMJ was measured on a 100 mm Visual Analogue Scale (VAS) with scores categorized as none (o mm), mild (1–30 mm), moderate (31–60 mm), or severe (61–100 mm). Maximum unassisted mouth opening (MMO) was measured in millimetres using a ruler calibrated to the nearest millimetre, serving as an objective quantitative measure of mandibular opening. Restricted mouth opening was defined as MMO < 40 mm, according to DC/TMD criteria.

Assessment of TMJ dysfunction severity

As noted above, an additional ordinal severity grading of TMD was completed using the Helkimo Clinical Dysfunction Index (Di 0–III), categorizing participants as having none (Di 0), mild (Di I), moderate (Di II), or severe (Di III) dysfunction. This rating system was based on clinical assessment of mandibular range of motion, TMJ function, muscle and joint pain upon palpation, as well as pain on jaw movement. The Helkimo Index was used for purposes of severity stratification and group comparison only; study participants were diagnosed with and classified according to DC/TMD Axis I criteria as described above [20]. The decision to utilize both diagnostic tools was made in order to capture both standardized diagnostic information that meets contemporary requirements for standardization in research, as well as an ordinal severity grading that could be compared to the larger epidemiological TMD literature.

Bruxism assessment

Bruxism was evaluated according to the three-level diagnostic criteria by Lobbezoo et al. [21]: self-reported grinding/snapping at night and clenching during the day in combination with clinical examination grading of tooth wear facets using BEWE index, masseter hypertrophy, linea alba and scalloped tongue edges. Individuals who presented both self-reported bruxism and clinical signs were classified as definite bruxism cases and those presenting only one criterion were classified as probable bruxism cases. Definite bruxism and probable bruxism groups were combined into bruxism group for further analyses.

Assessment of psychological stress and sleep quality

Subjective stress was assessed using the 10-item Perceived Stress Scale (PSS-10), in which each item is scored 0–4 (never to very often) and summed to a total range of 0–40, with higher scores indicating greater perceived stress; scores ≥20 were operationally defined as high stress for this study, consistent with prior noise-exposure literature. Subjective sleep quality was measured using the 19-item self-rated Pittsburgh Sleep Quality Index (PSQI), which yields seven component scores (0–3 each: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication, daytime dysfunction) summed to a global score of 0–21; scores >5 were defined as poor sleep quality [22]. Both scales were administered by trained research coordinator.

Statistical analysis

Statistical analysis was performed using a licensed version of IBM SPSS Statistics v26.0. Normality was checked with the Shapiro–Wilk test and variance homogeneity with Levene’s test (all p > 0.05). Categorical variables were compared using chi-square, continuous variables using Student’s t-test, and the ordinal VAS scale using the Mann–Whitney U test with effect size r. A multivariable binary logistic regression (TMD as outcome) included noise exposure, bruxism, malocclusion, sex, PSS-10, PSQI, and PHQ-4, chosen a priori for clinical relevance. Multicollinearity was checked via VIF (all VIF < 2.0, tolerance > 0.5). Model fit was assessed with the Hosmer–Lemeshow test and Nagelkerke R²Significance was set at α < 0.05. logistic regression and mediation analysis were conducted to address distinct, complementary questions: logistic regression identified independent predictors of TMD and quantified their adjusted association after accounting for confounders, whereas medication analysis explored the extent to which the noise TMD association may operate through hypothesized psychosocial and behavioural pathways (bruxism and perceived stress). Mediation analysis used a bootstrapped (5000 resamples) product of coefficients approach; given the simultaneous cross-sectional measurement of exposure, mediators, and outcomes, results are interpreted as evidence of statistical medication rather than confirmed causal pathways. Odds ratios were used as the primary measure of association because the multivariable analysis was conducted using binary logistic regression, the standard approach for adjusting simultaneously for multiple categorical and continuous covariates in this study design. However, given the high prevalence of TMD in both groups (61.3% and 25.3%), the odds ratio likely overestimates the true prevalence ratio; this should be considered when interpreting the magnitude, though not the direction, of the observed associations.

Biological mechanism

Theoretically, these results can be explained by a neuroendocrine stress response pathway that has been well established in animal models and supported by human studies. Noise-induced activation of the HPA axis is associated with an increase in circulating cortisol and catecholamines, potentially associated with heightened muscle tonus of the masticatory muscles [12]. Chronically elevated muscle tonus may be associated with bruxism, which further loads the TMJ and is associated with tooth grinding through accelerated occlusal wear [13,14]. Mediation analysis showed bruxism and perceived stress jointly mediated 41.4% of the total association between noise exposure and TMD (Natural Indirect Effect = 41.4%, 95% CI: 29.1–55.6%), with bruxism as the dominant mediator (24%). As exposure, mediators and outcome were all measured simultaneously in this cross sectional design, temporal precedence among these variable cannot be established; these findings are therefore compatible with, but not proof of, a mediating pathway. Of note, the majority of the risk was retained directly by the noise-exposure pathway (58.6%; abstinence mediated effect), indicating noise-related influences on TMD arise from pathophysiological mechanisms beyond psychologic arousal (i.e., direct acoustic-vibratory stimulation of periarticular structures in the TMJ [14], or possibly systemic inflammation independent of HPA axis activation [12]). These findings were further substantiated by the extremely high prevalence of bruxism observed in the high traffic group compared to published population norms of 8–13% for sleep bruxism and 22–30% for awake bruxism [16,21]. In sum, chronic noise was an upstream risk indicator for bruxism.

Psychosocial pathways

Adjusted odds of TMD were greater in respondents who reported high perceived stress (aOR = 1.80), poor sleep quality (aOR = 1.60), or anxiety/depression (aOR = 1.70), lending support to prior studies by Pavlou et al. [16] and Arregi et al. [14] which demonstrated that road traffic noise causes sustained sympathetic nervous system activation and sleep disturbance regardless of subjective stress perception. We found significantly higher scores on both the PSS-10 and PSQI in participants living on high traffic roads compared to their low traffic counterparts, indicating stress and sleep disturbance mediate the relationship between noise exposure and TMD risk rather than serving as proxy measures for noise.

Bidirectionality and confounding

Reverse causation cannot be ruled out with this cross-sectional study design. Pain sensitisation as a consequence of TMD may increase perceived annoyance from noise and sleep disturbance due to TMD may increase noise sensitivity creating a vicious cycle. Traffic-heavy areas may also be subject to co-exposures correlated with noise such as air pollution, socioeconomic deprivation, occupational noise exposures, all of which could play a role in TMD risk. The higher prevalence of malocclusion observed in the high traffic group likely reflects pre-existing skeletal or occlusal characteristics rather than a causal effect of noise exposure, and should not be interpreted as evidence that environmental noise drives malocclusion development. Confounding by these and other unmeasured factors is possible despite adjustment for known confounders in the multivariable model [7].

Sex differences

Female sex remained significantly associated with TMD in the full adjusted model (aOR = 1.50), as has been shown previously and is likely due to hormonal effects on ligament looseness and sensitisation of pain pathways. The unadjusted analysis was not significant (p = 0.11), however appeared after controlling for noise exposure and bruxism in the multivariable analysis. This highlights that sex-based differences in disease risk may be dependent on other environmental and behavioural factors.

Clinical and public health implications

Noise should be considered as a potential risk factor for TMD in dental clinical settings, especially in high traffic areas. Dentists should be aware of bruxism as a symptom and screen their patients accordingly. Public health initiatives aimed at noise reduction should consider oral health as a co-objective when planning noise mitigation strategies such as rerouting traffic, sound proofing residences and implementing green space buffers. Noise reduction could have many benefits for cardiovascular health, sleep disturbance, and as demonstrated here, TMD.

Limitations

This study has several limitations: the cross-sectional design precludes causal inference; non-probability sampling from a single neighbourhood per group in one city limits external validity; residual confounding (air pollution, occupation, income, safety, healthcare access, population density) could not be excluded; the short-duration noise assessment does not fully confirm long-term acoustic stability; bruxism relied on self-report rather than polysomnography/electromyography; occlusal assessment was limited to Angle’s classification; and findings are limited to one city and a five-day recruitment window. Additionally, given the high prevalence of the primary outcome (TMD) in both exposure groups, the odds ratios reported likely overestimate the corresponding prevalence ratios; a Poisson regression with robust variance or a log-binomial model would have yielded a prevalence ratio better suited to a cross-sectional design with a common outcome, and future studies should consider this approach.