Lifestyle behaviors as Effect Modifiers in the Association Between Oral Health and Carotid Artery Calcification: A Cross-Sectional Study Authors: Yi-Ling Chen, Mei-Chi Chang, Kuo-Liong Chien, Shu-Hui Chang, Chi-Hung Chen, Yi-Lun Chung, Yi-Ling Tsai, Wan-Chuen Liao, Jiiang-Huei Jeng Published: 2026 ABSTRACT AbstractIntroduction and aimsThe relationship between carotid artery calcification (CAC) detected on panoramic radiographs and oral health remains unclear due to potential confounding factors. This study aimed to evaluate this association after adjusting for these confounders and explore potential effect modification by lifestyle behaviours.MethodsThis cross-sectional study assessed the association between CAC on panoramic radiographs and oral health outcomes, including decayed, missing, and filled teeth (DMFT) index, periodontal disease, and apical periodontitis. Directed acyclic graphs were used to determine the minimal sufficient adjustment set, including sociodemographic factors, medical conditions (hypertension, hyperlipidaemia, and diabetes mellitus), and lifestyle habits (alcohol consumption, betel quid chewing, and smoking). A complete dataset was obtained after multiple imputation of missing values. Associations were examined using regression models, and effect modification by lifestyle behaviours was assessed using interaction terms, with stratified analyses performed for significant interactions.ResultsThe prevalence of CAC (670 patients) was 21.49%. CAC patients had a DMFT index 2.026 units higher, 0.290 more teeth with apical periodontitis, 1.873-fold higher odds of periodontal disease, and 1.554-fold higher odds of concurrent periodontal and apical diseases. After adjustment for confounders, the association between CAC and oral health outcomes disappeared. Age was associated with higher DMFT index (0.279 units), more teeth with apical periodontitis (0.017 units), and higher odds of periodontal and apical disease (3.5%-7.8% per year) (P < .001). Males were associated with periodontal disease. A significant interaction between CAC and alcohol consumption was observed, with CAC significantly associated with periodontal disease among drinkers (odds ratio = 9.051).ConclusionAfter adjustment for confounders, associations between CAC and oral health outcomes disappeared. Alcohol consumption modified the association between CAC and periodontal disease, with an association observed among drinkers.Clinical relevanceWhile radiographic detection of CAC facilitates cardiovascular risk identification and referral, alcohol consumption modifies the association between CAC and periodontal disease.Key words: Carotid artery calcification, Medical condition, Lifestyle behaviour, Oral health, Periodontitis, Apical periodontitis Introduction Carotid artery calcifications (CACs), located at the carotid bifurcation, are associated with an increased risk of cerebrovascular events,1,2 and account for about 50% of strokes due to atherosclerosis.3 Detected incidentally on panoramic radiographs, they appear as radiopaque vertical-linear, nodular, or irregular shapes near the C3 to C4 vertebrae,2 with a prevalence of 2.9% to 30.7%.4, 5, 6 CACs are present in 84% of patients with ≥50% carotid stenosis,7 and are significantly correlated with hypertension,8,9 hyperlipidaemia,8,9 diabetes mellitus (DM),10 and poor oral health.9 An age- and sex-matched investigation reported that patients with CACs exhibited higher decayed, missing, and filled teeth (DMFT) index and suboptimal oral health.9 After adjusting for systemic risk factors, CACs were associated with greater bone loss, periodontal inflammation, and advanced apical periodontitis.11 This association may involve periodontal pathogens invading endothelial cells and promoting atherothrombosis.12 Additionally, apical periodontitis or endodontic pathosis may independently contribute to atherosclerosis and cardiovascular disease.13,14 However, results from numerous investigations into CAC and periodontal disease have yet to reach conclusive outcomes. Some other studies have found no significant association between CACs and periodontal status.15,16 These inconsistencies may reflect the influence of multiple confounders in real-world data. Given the complex interplay between systemic conditions and dental status, lifestyle factors are also important contributors to oral health outcomes.17,18 Interestingly, one study showed that the link between CAC and periodontitis became nonsignificant after adjusting for smoking.18 This underscores the importance of accounting for underlying lifestyle factors, as they may confound the results or interact with the observed effect of CAC on oral health if not properly considered. Although studies have demonstrated the detection of CACs on panoramic radiographs,19,20 and their association with cardiovascular events,21 few have thoroughly examined the influence of sociodemographic, medical, and lifestyle factors on common oral health outcomes such as the DMFT index, periodontal status, and apical condition in patients with or without CACs. Moreover, potential confounders remain unclear, and effect modification has been underexplored, highlighting a critical gap in the literature. This article addresses these limitations by using directed acyclic graphs (DAGs) to identify the minimal sufficient adjustment set and applying interaction terms for analysis. The purposes of this study were to investigate the association between CAC and oral health outcomes (DMFT index, periodontal condition, and apical status), identify potential confounding variables including sociodemographic factors (age and sex), medical conditions (hypertension, hyperlipidaemia, and DM), and lifestyle habits (alcohol consumption, betel quid [BQ] chewing, and smoking), and explore potential effect modification by lifestyle behaviours. The null hypotheses of this study were as follows: (1) there was no significant difference in oral health outcomes between patients with and without CAC; (2) the association between CAC and oral health outcomes was not influenced by sociodemographic, medical, or lifestyle factors; and (3) there was no significant interaction between CAC and lifestyle behaviours on oral health outcomes. Methods Study design This cross-sectional study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (Appendix 1) and was approved by the Institutional Review Board (IRB) of the authors’ affiliated university hospital (IRB: 201206039-RIC). Anonymized medical records and panoramic radiographs with coded identifiers ensured confidentiality. Setting and study participants Patients who visited the Dental Department of the affiliated university hospital from 1 July to 2 August 2021 were consecutively recruited using a convenience sampling approach to reduce selection bias. Patients aged 20 years or older with available panoramic radiographs were included. Exclusion criteria were age under 20, unavailable panoramic images, with oral malignancies, undergoing orthognathic surgery or orthodontic treatment, complete edentulism, presence of metal bone plate implants, osteonecrosis, dental arch fractures, and panoramic radiographs with artefacts interfering with image assessment. Panoramic radiograph acquisition Routine panoramic radiographs from dental examinations were used to minimize radiation exposure. Images were captured using the Veraviewepocs 2D system (Morita Corp.) and processed with i-Dixel software, following manufacturer’s protocol (65 kV, 5 mA, and 1.3× magnification) and standard patient positioning. Variables Sociodemographic characteristics Patient information recorded included age, sex, and education level, categorized as lower (≤12 years of education) or higher (>12 years of education). Medical conditions CAC CACs were assessed on panoramic radiographs as nodular, linear, irregular, or heterogeneous radiopaque formations near the C3 to C4 vertebrae, according to established criteria.22 Two specialists independently evaluated CACs, yielding an interobserver kappa coefficient of 0.851 and an agreement rate of 94.29%, indicating strong reliability. Hypertension assessment Patients were classified as having hypertension if they met any of the 2024 European Society of Cardiology guideline criteria: office blood pressure ≥140/90 mmHg, home blood pressure ≥135/85 mmHg, or daytime ambulatory blood pressure ≥135/85 mmHg.23 Hyperlipidaemia assessment Patients were classified as having hyperlipidaemia if their triglyceride levels were ≥150 mg/dL, total cholesterol ≥200 mg/dL, low-density lipoprotein cholesterol ≥130 mg/dL, or high-density lipoprotein cholesterol <40 mg/dL, in accordance with the National Cholesterol Education Program guidelines of the United States.24 DM assessment According to patients’ medical records, individuals were diagnosed with DM if they met any of the following criteria: a fasting plasma glucose level of ≥126 mg/dL, a 2-hour plasma glucose level of ≥200 mg/dL during an oral glucose tolerance test, a glycated haemoglobin level of ≥6.5%, or the presence of classic hyperglycaemic symptoms or hyperglycaemic crisis accompanied by a random plasma glucose level of ≥200 mg/dL.25 Lifestyle behaviours Lifestyle habits, including alcohol consumption, BQ chewing, and smoking, were obtained from patient records. Weekly alcohol use defined consumers. Any past or current BQ chewing or smoking categorized participants as users, compared with those who had no history of these habits. Dental assessment DMFT index The number of decayed teeth, missing teeth (MT), and filled teeth (FT) was determined through review of patient records, clinical assessments performed by dental specialists, and analysis of corresponding panoramic radiographs. The DMFT index was subsequently derived following the criteria established by the World Health Organization.26 Periodontal assessment Full-mouth periodontal condition was evaluated using panoramic radiographs, with bone loss defined as exceeding one-third of the root length, according to a previously published method.27 For this study, periodontal disease was specifically represented by bone loss involving ≥30% of the teeth, whereas individuals not meeting this criterion were categorized as being without generalized periodontal destruction. Two specialists independently performed periodontal evaluations, demonstrating substantial reliability with an interobserver kappa coefficient of 0.771 and 88.57% agreement. Periapical assessment Periapical condition was assessed on panoramic radiographs using the periapical index.28 Teeth with periapical index scores of 3 or higher were classified as having apical periodontitis, with the highest root score recorded for multirooted teeth. Both endodontically treated and untreated teeth meeting these criteria were included. The total number of affected teeth was calculated for each patient. Two specialists independently assessed periapical status, achieving an interobserver kappa coefficient of 0.711% and 85.71% agreement, indicating substantial concordance. Statistical analysis Handling of missing data To address missing data, multiple imputation was performed. The missing data were assumed to be missing at random, and five imputations were generated using the Markov chain Monte. Carlo method.29 All covariates, including sociodemographic factors, medical conditions, lifestyle behaviours, and dental status, were included in the imputation model. A complete dataset was subsequently obtained for statistical analyses. Descriptive statistics Patient characteristics, including sociodemographic, medical, lifestyle, and dental parameters, were summarized. Continuous variables were reported as means ± standard deviations, and categorical variables as frequencies and proportions. Comparisons of patient characteristics stratified by CAC status Logistic regression analyses were used to evaluate (1) associations between CAC and sociodemographic, medical, and lifestyle variables, and (2) associations between periodontal disease and apical periodontitis and medical conditions and lifestyle behaviours. Odds ratios (ORs) and 95% confidence intervals (CIs) were reported. Multicollinearity was assessed using variance inflation factor (VIF < 10 threshold). With CAC as the exposure and oral health (DMFT, periodontal disease, and apical periodontitis) as outcomes, associations were analysed using univariable and multivariable linear regression for continuous variables and logistic regression for categorical outcomes. To minimize confounding bias,30 multivariable models were adjusted for minimal sufficient sets identified by a DAG. This DAG was based on prior investigations illustrating relationships between CAC (exposure) and oral health outcomes. Associations were analysed using two models: Model 1, crude; and Model 2, adjusted based on DAGs. Interaction analysis The effect modification of lifestyle behaviours on the association between CAC and oral health outcomes (periodontal disease, apical periodontitis, and concurrent periodontal and apical diseases) was investigated using interaction terms between CAC and alcohol consumption, BQ chewing, and smoking in the logistic regression model. Stratified analyses were performed for variables with significant interaction terms to further evaluate effect modification. DAGitty software (Theoretical Biology & Bioinformatics Group, Utrecht University, http://dagitty.net) was used to construct the DAG. Analyses were conducted using SPSS software (IBM SPSS Statistics for Windows, Version 25.0: IBM Corp.). Statistical significance was defined as P < .05. Sample size calculation Using G*Power (version 3.1.9.7), a two-tailed test with α = 0.05 and 80% power estimated a minimum required sample size of 260, based on a reported prevalence of alveolar bone loss of 69.74% in patients without CAC and 84.31% in those with CAC.11 More patients were initially enrolled to account for potential exclusions. Results Participant characteristics Of the 972 patients initially enrolled, 670 were included for analysis after exclusions (Appendix 2). Missing values were observed for educational level (n = 346, 51.64%), medical conditions (n = 107, 15.97%), and lifestyle behaviours (n = 298, 44.48%). A total of 353 cases (52.69%), containing 1509 missing values (12.51% of the total dataset), were imputed. The mean ± standard deviation age was 54.72 ± 17.88 years. The prevalence of CAC was 21.49% (Figure 1). The majority of participants were female (53.58%) and did not have CAC (78.51%), hypertension (74.48%), hyperlipidaemia (91.34%), or DM (87.91%). Most reported no alcohol use (82.99%), BQ chewing (77.76%), or smoking (72.99%). The mean ± standard deviation values for decayed teeth, MT, FT, and DMFT index, as well as the number of teeth with apical periodontitis, were 1.17 ± 2.16, 4.80 ± 5.96, 3.09 ± 3.60, 9.06 ± 8.01, and 1.04 ± 1.57, respectively. Periodontal disease and apical periodontitis were absent in 58.66% and 56.12% of patients, respectively, and 74.33% had neither condition concurrently (Table 1). Panoramic radiographs of patients with and without carotid artery calcification. (A) Patient without carotid artery calcification. (B) Patient with carotid artery calcification. (C) Patient with carotid artery calcification. Arrows indicate the carotid artery calcifications. Table 1Patient characteristics by sociodemographic factors, medical conditions, lifestyle behaviours, and dental status after multiple imputation.Table 1 dummy alt textTotal (n = 670)VariablesMean ± standard deviationCase number (%)Sociodemographic characteristicsAge (y)54.72 ± 17.88SexMale311 (46.42%)Female359 (53.58%)EducationLower educational level*355 (52.99%)Higher educational level†315 (47.01%)Medical conditionCarotid artery calcificationYes144 (21.49%)No526 (78.51%)HypertensionYes171 (25.52%)No499 (74.48%)HyperlipidaemiaYes58 (8.66%)No612 (91.34%)Diabetes mellitusYes81 (12.09%)No589 (87.91%)Lifestyle behavioursAlcohol consumptionYes114 (17.01%)No556 (82.99%)Betel quid chewingYes149 (22.24%)No521 (77.76%)SmokingYes‡181 (27.01%)No489 (72.99%)Dental assessmentDMFTDecayed teeth1.17 ± 2.16Missing teeth4.80 ± 5.96Filled teeth3.09 ± 3.60DMFT index9.06 ± 8.01Periodontal diseaseYes277 (41.34%)No393 (58.66%)Apical periodontitisYes (periapical index ≥3)294 (43.88%)No (periapical index <3)376 (56.12%)Number of teeth with apical periodontitis1.04 ± 1.57Concurrent periodontal and apical disease§Yes172 (25.67%)No498 (74.33%)⁎Lower educational level: attended primary school, elementary school, junior high school, or senior high school (≤12 years of education).†Higher educational level: attended higher education or with postgraduate degrees (>12 years of education).‡Smoker and ex-smoker. DMFT: decayed, missing, and filled teeth.§Concurrent periodontal and apical disease were defined as the presence of periodontal disease and apical periodontitis in the same patient. Patient characteristics by sociodemographic factors, medical conditions, lifestyle behaviours, and dental status after multiple imputation. Lower educational level: attended primary school, elementary school, junior high school, or senior high school (≤12 years of education). Higher educational level: attended higher education or with postgraduate degrees (>12 years of education). Smoker and ex-smoker. DMFT: decayed, missing, and filled teeth. Concurrent periodontal and apical disease were defined as the presence of periodontal disease and apical periodontitis in the same patient. Associations of medical conditions and lifestyle behaviours with periodontal and apical disease Hypertension (OR = 4.130, P < .001), hyperlipidaemia (OR = 2.189, P = .012), and DM (OR = 2.928, P < .001) were associated with periodontal disease. Hypertension (OR = 2.711, P < .001) and DM (OR = 1.981, P = .011) also showed associations with concurrent periodontal and apical disease, while only hypertension (OR = 1.617, P = .009) was significantly associated with apical periodontitis (Appendix 3). Comparison of sociodemographic, medical, and lifestyle characteristics by CAC status Age was significantly associated with higher odds of CAC (OR = 1.034, P < .001), while male sex was associated with lower odds of CAC (OR = 0.444, P < .001). Among medical conditions, hypertension (OR = 1.973, P = .001), hyperlipidaemia (OR = 2.533, P = .003), and DM (OR = 2.365, P = .004) were significantly associated with increased odds of CAC. Smoking was associated with lower odds of CAC (OR = 0.497, P = .008) (Table 2). Table 2Univariate logistic regression analysis of the associations between sociodemographic, medical, and lifestyle variables and carotid artery calcification.Table 2 dummy alt textCarotid artery calcificationVariablesOR (95% CI)P valueSociodemographic characteristicsAge (y)1.034 (1.022, 1.046)<.001*Male sex0.444 (0.300, 0.657)<.001*Medical conditionHypertension1.973 (1.310, 2.972).001*Hyperlipidaemia2.533 (1.380, 4.650).003*Diabetes mellitus2.365 (1.322, 4.231).004*Lifestyle behavioursAlcohol consumption0.597 (0.297, 1.201).144Betel quid chewing0.819 (0.285, 2.351).686Smoking0.497 (0.297, 0.832).008*Logistic regression analyses.CI, confidence interval; OR, odds ratio.⁎Significant difference. Univariate logistic regression analysis of the associations between sociodemographic, medical, and lifestyle variables and carotid artery calcification. Logistic regression analyses. CI, confidence interval; OR, odds ratio. Significant difference. Confounder selection by DAG In the DAG used to identify potential confounders, CAC was defined as the exposure, and the DMFT index, periodontal disease, and apical periodontitis were defined as the outcomes (Appendixes 4 and 5). The minimal sufficient adjustment set for estimating the total effect of CAC on these outcomes included age, sex, hypertension, hyperlipidaemia, DM, alcohol consumption, BQ chewing, and smoking (Figure 2). These variables were subsequently included in all multivariable regression models. Directed acyclic graphs illustrating the relationships between carotid artery calcification and the DMFT index, periodontal disease, and apical periodontitis. DMFT, decayed, missing, and filled teeth. Linear regression analyses of the associations between CAC and the DMFT index and number of teeth with apical periodontitis VIF values (1.000-1.224) indicated minimal multicollinearity (Appendix 6). Crude linear regression (Model 1) showed that CAC was significantly associated with a higher DMFT index (2.026 units; P = .007) and a 0.290 increase in teeth with apical periodontitis (P = .049). After DAG adjustment (Model 2), CAC showed no significant association with DMFT (P = .322) or apical periodontitis (P = .269). However, age was significantly associated with both outcomes (P < .001), with each additional year predicting an increase of 0.279 DMFT units and 0.017 teeth with apical periodontitis (Table 3). Table 3Linear regression analyses of the association between carotid artery calcification and oral health outcomes (the DMFT index and the number of teeth with apical periodontitis) across crude and adjusted models.Table 3 dummy alt textModel 1Model 2UnivariableMultivariableCrudeAdjusted for potential confounders identified by DAGDMFT indexThe number of teeth with apical periodontitisDMFT indexThe number of teeth with apical periodontitisVariablesβ (95% CI)P valueβ (95% CI)P valueβ (95% CI)P valueβ (95% CI)P valueExposure variableCarotid artery calcification2.026 (0.556, 3.497).007*0.290 (0.001, 0.579).049*−0.620 (−1.849, 0.608).3220.168 (−0.130, 0.467).269ConfoundersAge (per 1-y increase)0.279 (0.248, 0.309)<.001*0.017 (0.009, 0.024)<.001*Men (vs women)−0.456 (−1.724, 0.811).4780.088 (−0.251, 0.427).605Hypertension0.866 (−0.505, 2.237).2150.098 (−0.227, 0.422).554Hyperlipidaemia−1.123 (−3.122, 0.876).268−0.230 (−0.680, 0.220).317Diabetes mellitus−0.657 (−2.403, 1.090).4590.158 (−0.241, 0.556).438Alcohol consumption0.675 (−1.535, 2.885).5320.171 (−0.396, 0.737).534Betel quid chewing0.705 (−1.392, 2.803).499−0.320 (−0.780, 0.141).171Smoking0.267 (−1.398, 1.932).7490.186 (−0.241, 0.614).383Linear regression analyses.CI, confidence interval; DAG, directed acyclic graph; DMFT, decayed, missing, and filled teeth.⁎Significant difference. Linear regression analyses of the association between carotid artery calcification and oral health outcomes (the DMFT index and the number of teeth with apical periodontitis) across crude and adjusted models. Linear regression analyses. CI, confidence interval; DAG, directed acyclic graph; DMFT, decayed, missing, and filled teeth. Significant difference. Logistic regression analyses of the associations between CAC and the risk of periodontal and apical diseases In Model 1, CAC was significantly associated with higher odds of periodontal disease (OR = 1.873, P = .001) and concurrent periodontal and apical disease (OR = 1.554, P = .032). In Model 2, each year of age increased the odds of periodontal disease, apical periodontitis, and concurrent conditions by 7.8%, 3.5%, and 6.8%, respectively (all P < .001). Males also exhibited higher odds of periodontal disease (OR = 1.904, P = .037) (Table 4). Table 4Logistic regression analyses of the association between carotid artery calcification and the risk of periodontal and apical diseases across crude and adjusted models.Table 4 dummy alt textModel 1Model 2UnivariableMultivariableCrudeAdjusted for potential confounders identified by DAGPeriodontal diseaseApical periodontitisConcurrent periodontal and apical diseasePeriodontal diseaseApical periodontitisConcurrent periodontal and apical diseaseVariablesOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueOR (95% CI)P valueExposure variableCarotid artery1.873.001*1.369.0961.554.032*1.286.2811.118.5911.097.700calcification(1.292, 2.717)(0.946, 1.982)(1.040, 2.324)(0.814, 2.031)(0.744, 1.679)(0.684, 1.761)ConfoundersAge (per 1-y increase)1.078<.001*1.035<.001*1.068<.001*(1.063, 1.095)(1.024, 1.046)(1.052, 1.085)Men (vs women)1.904.037*1.350.1781.439.189(1.042, 3.481)(0.870, 2.096)(0.832, 2.487)Hypertension1.390.1690.920.7091.159.528(0.870, 2.220)(0.595, 1.424)(0.732, 1.835)Hyperlipidaemia1.147.6940.742.3380.631.207(0.580, 2.266)(0.402, 1.368)(0.309, 1.290)Diabetes mellitus1.121.7111.010.9701.005.988(0.612, 2.053)(0.588, 1.735)(0.567, 1.780)Alcohol consumption1.331.4551.204.6671.528.355(0.616, 2.874)(0.478, 3.036)(0.594, 3.926)Betel quid chewing1.383.3880.743.4140.672.297(0.654, 2.923)(0.357, 1.548)(0.317, 1.426)Smoking1.069.8650.968.9091.117.784(0.472, 2.421)(0.544, 1.723)(0.483, 2.583)Logistic regression analyses.CI, confidence interval; DAG, directed acyclic graph; OR, odds ratio.⁎Significant difference. Logistic regression analyses of the association between carotid artery calcification and the risk of periodontal and apical diseases across crude and adjusted models. Logistic regression analyses. CI, confidence interval; DAG, directed acyclic graph; OR, odds ratio. Significant difference. Interaction analysis Increasing age was significantly associated with a higher risk of periodontal disease, apical periodontitis, and concurrent periodontal and apical disease across all models (OR = 1.035-1.079, all P < .001). Male sex was significantly associated with an increased risk of periodontal disease (OR = 1.831-1.880, all P < .05). Among the interaction terms, only the interaction between CAC and alcohol consumption was significantly associated with periodontal disease (P = .042). No other significant interaction effects were observed (Table 5). Stratified analysis showed that among drinkers, CAC was significantly associated with periodontal disease (OR = 9.051; 95% CI = 1.354, 60.510; P = .023), suggesting that alcohol consumption may act as an effect modifier in this relationship (Appendix 7). Table 5Logistic regression analyses of the association between carotid artery calcification and the risk of periodontal and apical diseases with lifestyle behaviour interaction terms.Table 5 dummy alt textPeriodontal diseaseApical periodontitisConcurrent periodontal and apical diseaseVariablesOR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)OR (95% CI)P valueP valueP valueP valueP valueP valueP valueP valueP valueAge (per 1-y increase)1.0791.0791.0791.0351.0351.0351.0681.0691.068(1.063, 1.096)(1.063, 1.095)(1.063, 1.095)(1.024, 1.046)(1.024, 1.046)(1.024, 1.046)(1.052, 1.085)(1.052, 1.085)(1.052, 1.085)<.001†<.001†<.001†<.001†<.001†<.001†<.001†<.001†<.001†Men (vs women)1.8801.8441.8311.3571.3441.3551.4371.4001.419(1.022, 3.460)(1.027, 3.308)(1.004, 3.338)(.875, 2.105)(.858, 2.105)(.868, 2.116)(.831, 2.484)(.794, 2.469)(.812, 2.478).043†.041†.049†.171.194.179.191.240.215Carotid artery calcification1.0301.0781.0241.2091.0871.1401.092.9501.026(.613, 1.730)(.616, 1.886)(.613, 1.710)(.772, 1.894)(.675, 1.750)(.693, 1.875)(.640, 1.863)(.518, 1.743)(.589, 1.788).912.792.929.406.732.604.746.867.928Hypertension1.4041.4111.405.923.922.9191.1621.1721.165(.878, 2.244)(.886, 2.249)(.879, 2.246)(.598, 1.425)(.595, 1.429)(.593, 1.426)(.734, 1.840)(.736, 1.865)(.735, 1.848).156.147.155.717.715.707.522.503.516Hyperlipidaemia1.1531.1331.142.738.740.740.629.618.626(.575, 2.315)(.574, 2.234)(.584, 2.236)(.399, 1.364)(.401, 1.366)(.400, 1.367)(.309, 1.283)(.298, 1.281)(.303, 1.293).687.719.697.332.336.336.202.195.205Diabetes mellitus1.1561.1521.096.9991.0161.0181.0031.0231.004(.625, 2.137)(.626, 2.120)(.590, 2.035)(.582, 1.714)(.589, 1.752)(.592, 1.751)(.563, 1.785)(.573, 1.827)(.566, 1.782).644.649.771.996.955.947.993.937.989Alcohol consumption1.0171.3601.3271.3361.2051.2031.5241.5431.527(.430, 2.404)(.625, 2.961)(.613, 2.876)(.466, 3.828)(.478, 3.040)(.476, 3.038)(.488, 4.762)(.604, 3.942)(.593, 3.932).967.425.461.552.666.669.435.342.356Betel quid chewing1.4341.1421.399.740.720.742.675.571.670(.638, 3.225)(.509, 2.561)(.628, 3.117)(.350, 1.563)(.324, 1.599)(.353, 1.559)(.317, 1.440)(.249, 1.309)(.316, 1.423).372.743.400.414.405.416.306.184.294Smoking1.0651.089.907.973.970.9851.1221.1331.065(.453, 2.507)(.483, 2.455)(.403, 2.039)(.548, 1.728)(.543, 1.734)(.552, 1.758)(.485, 2.593)(.485, 2.648)(.469, 2.421).877.826.803.923.917.959.776.759.874Carotid artery calcification × alcohol consumption*6.860.497.953(1.069, 44.006)(.091, 2.711)(.169, 5.382).042†.402.955Carotid artery calcification × betel quid chewing*2.8521.1881.857(.395, 2.577)(.318, 4.444)(.415, 8.303).294.798.417Carotid artery calcification × smoking*3.510.8771.318(.893, 13.798)(.230, 3.336)(.373, 4.659).072.841.665Logistic regression analyses. Models were adjusted for age, sex, carotid artery calcification, hypertension, hyperlipidaemia, diabetes mellitus, alcohol consumption, betel quid chewing, and smoking.CI, confidence interval; OR, odds ratio.⁎Interaction term.†Significant difference. Logistic regression analyses of the association between carotid artery calcification and the risk of periodontal and apical diseases with lifestyle behaviour interaction terms. Logistic regression analyses. Models were adjusted for age, sex, carotid artery calcification, hypertension, hyperlipidaemia, diabetes mellitus, alcohol consumption, betel quid chewing, and smoking. CI, confidence interval; OR, odds ratio. Interaction term. Significant difference. Discussion All three hypotheses of this study were rejected. Initially, CAC was correlated with adverse oral health outcomes; however, these associations disappeared after adjustment. These findings indicate that the observed crude relationships were largely confounded by sociodemographic, medical, and lifestyle factors. Notably, a significant interaction between CAC and alcohol consumption was observed, with CAC significantly associated with periodontal disease among drinkers. Overall, these results suggest that oral health is shaped by a multifactorial interplay of sociodemographic, medical, and lifestyle factors rather than by the presence of CAC alone. Consistent with previous studies, CAC prevalence was 21.49%,11 with higher frequencies observed in older adults, females, and patients with hypertension, hyperlipidaemia, or DM.9,19,31,32 Hypertension induces inflammation and reduces vascular elasticity,33 while hyperlipidaemia promotes the formation of oxidized phospholipids in cardiovascular tissues.34 Additionally, DM accelerates atherosclerosis through hyperglycaemia-driven endothelial dysfunction and elevated inflammatory medicators.35 These factors contribute to the progression of CAC. Interestingly, smoking was inversely related to the odds of CAC in this study. While smoking is associated with CAC, cessation may mitigate this process.36 In this study, individuals were classified as smokers if they had any history of tobacco use, either current or past. This broader definition of smoking may influence the observed associations and explain the discrepancies. In the crude analysis of our study, patients with CAC presented a higher DMFT index, a higher number of teeth with apical periodontitis, a greater risk of periodontal disease, and a higher prevalence of concurrent periodontal and apical disease, which is consistent with prior investigations. In a Parogene cohort study in Finland, which included 492 patients, CAC was associated with severe apical periodontitis, alveolar bone loss, and advanced periodontal inflammation, suggesting a link between CAC and oral infections.11 The relationship between oral health and CACs has been explored. Inflammation in periodontal tissues may contribute to the progression of atherosclerosis, and transient bacteraemia from periodontal infections can introduce pathogens into pre-existing atherosclerotic plaques.37 Similarly, endodontic infections and apical periodontitis facilitate atherosclerosis by releasing bacteria and virulence factors into the bloodstream, thereby triggering systemic inflammation.11 The association between the DMFT index and CACs is influenced by the number of MT, which reflects advanced periodontitis, while treated or FT show little relation.37 A study found 46% of individuals with 0 to 9 MT had carotid artery plaque, rising to 60% in those with 10 or more MT.38 Surprisingly, after adjusting for the minimal sufficient adjustment sets identified by a DAG, the previously significant associations between CAC and oral health outcomes were no longer significant, suggesting that these associations may be attributable to confounding factors. In particular, age remained significantly associated with the DMFT index, the number of teeth with apical periodontitis, periodontal disease, and concurrent periodontal and apical diseases in the adjusted models, indicating that age may influence both vascular and oral health. This is consistent with previous studies showing that advanced age is a well-documented risk factor for both CAC and cumulative oral diseases.16,39, 40, 41 Our study also supports earlier findings that periodontal disease is more prevalent in men after controlling for covariates.39 Aligned with existing evidence,42, 43, 44 hypertension, hyperlipidaemia, and DM were significantly associated with periodontal disease. Hypertension and periodontitis are linked by common inflammatory mediators (C-reactive protein, tumour necrosis factor-α, and interleukin-6), with smoking also contributing to both conditions.42 DM promotes periodontal disease through immune impairment, oxidative stress, and altered oral microbiota.43 Hyperlipidaemia interacts with these conditions by inducing insulin resistance and elevating proinflammatory cytokines.44 These findings underscore the complex relationships between systemic health and oral disease. Several investigations have shown that lifestyle habits such as alcohol consumption, BQ chewing, and smoking are associated with an increased risk of periodontal disease.45, 46, 47 These findings highlight the role of lifestyle factors in oral health. In addition, alcohol consumption was identified as a potential effect modifier of the association between CAC and periodontal disease in this study. A meta-analysis showed that alcohol use elevates the risk of periodontitis, possibly through impaired neutrophil function and enhanced bacterial proliferation.45 Interestingly, a significant association between CAC and periodontal disease was observed among drinkers in the present analysis; however, the wide CI warrants cautious interpretation, and future studies with larger alcohol-consuming samples are needed to validate this effect. This is the first study to use a DAG-based model to determine the minimal sufficient adjustment set for confounding without overadjustment and to apply interaction analysis to investigate the potential effect-modifying role of lifestyle behaviours in the relationship between CAC and oral health outcomes. Most previous studies have focused solely on the independent association between medical conditions and CAC, with lifestyle behaviours and effect modification remaining largely unexplored. In addition, stratified analysis of significant interaction terms further enhanced the interpretation of the findings. Another strength is the use of multiple imputation to address missing data, enabling a complete analytic sample. The study also met the minimum required sample size as calculated. In this study, confounding and effect modification were addressed using distinct analytical approaches. Confounders, defined as variables associated with both exposure and outcome, were identified using a DAG and adjusted for in multivariable regression models. Accordingly, CAC was initially associated with adverse oral health outcomes, but these associations disappeared after adjustment, suggesting confounding by sociodemographic, medical, and lifestyle factors. Effect modifiers, defined as variables that alter the magnitude of the exposure-outcome association, were assessed using interaction terms, with stratified analyses performed when significant interactions were identified. Interaction terms were used to evaluate effect modification between CAC and lifestyle behaviours (alcohol consumption, BQ chewing, and smoking), and a significant interaction was observed between CAC and alcohol consumption, with CAC associated with periodontal disease among drinkers. However, we acknowledge several limitations in this study. Residual confounding cannot be excluded, given that variables such as socioeconomic status, access to dental care, oral hygiene practices, dietary habits, medication use, and inflammatory biomarkers were not available due to the limitations of the retrospective study design. Although panoramic radiographs have lower resolution than periapical imaging, they are routinely performed during initial dental evaluations, are covered by the National Health Insurance system, and provide sufficient diagnostic information for this study. Alcohol consumption, BQ chewing, and smoking were classified only in broad categories due to the retrospective design, and the lack of detailed exposure data (eg, frequency, quantity, duration, type, and intensity) limits the precision of the analysis and precludes assessment of dose-response relationships. Derived from a single ethnic and geographic population in a university hospital setting using convenience sampling, the findings may be subject to selection bias and limited generalizability. This study’s findings, particularly the effect modification of lifestyle behaviours on the association between CAC and oral health outcomes, highlight the importance of considering patients’ habits such as alcohol consumption, BQ chewing, and smoking. Future studies could further stratify these behaviours by frequency, type, and duration to better understand their impact on oral health outcomes in patients with CAC. Additionally, analyses of microbiological profiles and biomarkers of systemic inflammation may provide further insights into the underlying mechanisms. A recent study identified that the neutrophil percentage-to-albumin ratio is significantly associated with poor periodontal health.48 Multicentre investigations would also be valuable to validate these findings and improve their generalizability across different populations. Conclusion In conclusion, after adjustment for confounders, the relationships between CAC and oral health outcomes were not statistically significant. However, alcohol consumption significantly modified the association between CAC and periodontal disease, particularly among drinkers, suggesting that lifestyle behaviours may influence oral health assessment in patients with CAC. Author contributions Conceptualization, methodology, software, validation, data curation, project administration, funding acquisition, writing – revised manuscript: Jeng, M.C. Chang, Liao. Methodology, investigation, validation, data curation, writing – original draft: Y.L. Chen, Chung, C.H. Chen, Tsai. Supervision, project administration, funding acquisition: S.H. Chang, K.L. Chien. Data availability All data generated or analysed during this study are presented within this article. Additional inquiries can be addressed to the corresponding author. Ethics statement This study was approved by the Institutional Review Board (IRB) of the authors’ affiliated university hospital (IRB number: 201206039-RIC). Funding This study was supported grants from National Science and Technology Council (NSTC), Taiwan (NSTC114-2314-B-002-098, NSTC113-2314-B-255-001-MY3, MOST111-2314-B002-109-MY3, MOST111-2314-B002-107-MY3), Chang Gung Memorial Hospital (CMRPF1N0061, NMRPF3L0043, NMRPF3L0033, CMRPF1K0072, CMRPF1M0081), National Taiwan University Hospital (NTUH114-EDN0003, NTUH114-CGN0010), and Kaohsiung Medical University (KMU-DK(A)-110002, KMUH111-1T09, KMU-TB114004, KMU-TB11404-2). Conflict of interest All authors declare there are no conflicts of interest for this submission. Acknowledgements The authors would like to thank Ms Chen Yu-Ting for her technical assistance. Footnotes Supplementary material associated with this article can be found in the online version at doi:10.1016/j.identj.2026.111102. Contributor Information Wan-Chuen Liao, Email: d08450001@ntu.edu.tw. Jiiang-Huei Jeng, Email: jhjeng@kmu.edu.tw, jhjeng@ntu.edu.tw. Supplementary materials REFERENCES Associated Data Supplementary Materials Data Availability Statement All data generated or analysed during this study are presented within this article. Additional inquiries can be addressed to the corresponding author.