Dental Caries, Race and Incident Ischemic Stroke, Coronary Heart Disease and Death

Abstract

Background:

Dental caries are a highly prevalent disease worldwide. In the US, untreated dental caries are present in more than 1 in 5 adults. The objective of this study was to determine the relationship between dental caries and incident ischemic stroke, coronary heart disease (CHD) events, and death.

Methods:

The dental cohort (n=6,351) of the Atherosclerosis Risk In Communities study was followed for incident ischemic stroke, CHD event, and all-cause mortality. Of all the participants at visit 4 (n=11,656), those unable to go through dental examination, or with prevalent ischemic stroke and CHD events, were excluded. The full-mouth dental examination was conducted at visit 4 (1996–1998), assessing dental caries. Dose response of decayed, missing, and filled surfaces due to caries was assessed and related to outcome. Outcomes were assessed through the end of 2019. Additionally, the effect of regular dental care utilization on dental caries was evaluated.

Results:

Participants with ≥1 dental caries had an increased risk of stroke [adjusted Hazard Ratio (HR) 1.40 95% Confidence Interval (CI) 1.10–1.79] and death (adjusted HR 1.13 95% CI 1.01–1.26), but not for CHD events (adjusted HR 1.13 95% CI 0.93–1.37). The association of dental caries and ischemic incident stroke was significantly higher in the African American population when compared to the white subgroup (interaction term p=0.0001). Increasing decayed, missing, and filled surfaces were significantly associated with stroke (adjusted HR 1.006, 95% CI 1.001–1.011) and death (adjusted HR 1.003, 95% CI 1.001–1.005), but not CHD (adjusted HR 1.002, 95% CI 1.000–1.005). Regular dental care utilization lowered (adjusted Odds Ratio (OR), 0.19, 95% CI 0.16–0.22, p<0.001) the chance of caries.

Conclusions:

Among the cohort, dental caries were independently associated with risk of ischemic stroke and death, with the effect higher in African American participants. Regular dental care utilization was associated with a lower chance of caries, emphasizing its relevance in prevention of these events.

Graphical Abstract

INTRODUCTION

Tooth loss is a marker of oral health that represents the end stage of oral diseases, such as dental caries and periodontal disease. It is associated with incident cardiovascular disease including ischemic stroke, myocardial infarction, and all-cause mortality¹. Dental caries are a highly prevalent disease worldwide, ranking as the most common health condition according to the Global Burden of Disease in 2019, with 2 billion of the 3.5 billion people who suffer from oral diseases having dental caries of permanent teeth². A recent update in the US population reports more than 1 in 5 adults have untreated dental caries³. Although both dental caries and periodontal disease can lead to tooth loss, they are distinct in their pathophysiology as well as their causative organisms. Periodontal disease (PD) induces a chronic low-grade inflammation that damages the dental support tissues, whereas, dental caries lead to oral bacteria penetrating the tooth structure, including dentin and enamel, causing chronic infection and inflammation of the dental pulp^4,5. Inflammatory mediators from both conditions are released from the inflamed oral tissues, inducing a chronic systemic low-grade inflammatory response in the vascular system and subsequent atherosclerosis⁶. Several types of periodontal disease pathogens have been detected in atheromatous plaques obtained from the carotid and coronary arteries⁷. The principal dental caries pathogen, Streptococcus mutans, has been detected in heart valve and atheromatous plaque specimens⁸. This evidence indicates that individual oral conditions may be a risk factor for coronary heart disease (CHD) events.

Periodontal disease has been shown to be associated independently with incident cardiovascular disease including ischemic stroke, myocardial infarction, and all-cause mortality⁹. However, there is limited data on the association between dental caries and an incident CHD event¹⁰. We hypothesized that dental caries are predictive of ischemic stroke, CHD events, and death in the Atherosclerosis Risk In Communities (ARIC) study. Based on the same dental cohort of the ARIC (DARIC) dataset we reported an association between periodontal disease and incident atrial fibrillation (AF)¹¹, which in turn increased the risk of systemic ischemic events by approximately three times¹². Similarly, periodontitis or dental plaque has been associated with incident AF¹³. We investigated the association between dental caries and incident AF and whether AF mediates associations between caries and ischemic events.

It has been reported that regular dental care use was associated with 23% lower rates of ischemic stroke compared with episodic dental care users¹⁴. Additionally, we propose to test whether regular dental care use will be associated with lower levels of dental caries when compared to episodic dental care use.

MATERIALS AND METHODS

Study Population:

The authors declare that all supporting data are available within the article and the ARIC Investigators are open to providing the data utilized in this article to a researcher, allowing them to replicate the findings, contingent upon the execution of a data use agreement that guarantees the safeguarding of the confidentiality of ARIC participants’ data. Starting in 1987, the ARIC Study began its cohort to examine the etiology of atherosclerosis and its associated outcomes¹⁵. Participants for this cohort were enrolled from four states in the US, including participants from North Carolina, Minnesota, Mississippi, and Maryland. The baseline dental examination was conducted at visit 4 (1996–1998) and outcomes were assessed at follow-up through the end of 2019. Vascular risk factors for this analysis were assessed during visit 4 (1996–1998). Hospital medical records were obtained for any hospitalizations occurring over the median follow-up period of 21 years. The study team also contacted the ARIC study participants every year to evaluate for stroke, CHD events, or all-cause mortality¹⁵. All participants provided written informed consent. The four examination field centers and the coordinating center were granted human subject approvals.

At visit 4, 11,656 individuals were assessed. Of those participants, 6,744 dental subjects were enrolled in DARIC [1,317 refused oral examination, 1,895 had medical contraindications to a dental exam at the time (contraindications include a dentist or physician stating individual needed to take antibiotics before every dental visit, congenital heart disease, rheumatic heart disease, heart murmur, endocarditis, mitral valve prolapse, pacemaker, heart, kidney or other organ transplant, artificial heart valve, stent, shunt or artificial joint, kidney dialysis, major surgery, radiation or chemotherapy within the last two months, taking prednisone or an immunosuppressive medication), and 1,619 were edentulous (total tooth loss, comprising the major reasons for exclusion)]¹⁶. Of these, 6,351 study participants without prior CHD or ischemic stroke had complete covariate information and underwent full mouth periodontal assessment of six sites per tooth, as well as a dental caries exam. Individuals not of the African American or white race were excluded (due to the limited sample size of other races)¹⁶. The individuals included in the study were compared to those excluded from the study.

Assessment of Dental Caries and Periodontal Disease:

Participants underwent full mouth caries assessments at visit 4 using modified Radike criteria¹⁷. The modifications consisted of using a tongue blade instead of an explorer, which made the examination more visual. The number of coronal and root carious lesions was recorded separately. For each tooth, a numerical status was recorded and denoted if a tooth was sound, decayed or filled, missing, restored with a crown, a decayed root fragment, a sound root fragment, or an implant. If the tooth status was decayed, filled, or restored with a crown, then it would be coded as dental surface caries and/or dental root caries. Root caries were recorded if there was a discrete, well-defined, and discolored cavitation on the root surface and the tongue blade could scrape away some of the enamel. Tooth fractures, erosion, and abrasions were considered distinct and were not recorded as caries. Participants with lesions, but not frank caries, were further investigated to detect pit, fissure, and smooth surface lesions, as well as root and coronal caries. The presence of coronal caries was used as the primary exposure of interest. To further explore the dose-response of dental caries to our outcomes, decayed, missing, and filled surfaces (DMFS) due to caries index were assessed¹⁸.

The presence or absence of PD was also assessed as a covariate during visit 4 within the DARIC study. The periodontal profile class (PPC) was classified on a scale from PPC-A (periodontal health) or PPC-B through PPC-G (Severe PD¹⁹.

Ischemic Stroke, CHD Event, and Death:

In the ARIC study, cases were identified through annual contact of participants and searching discharge lists from local hospitals. Deaths were ascertained through the review of hospital discharge records, death certificates, and informant interviews or physician questionnaires for out-of-hospital deaths. Incident CHD events included fatal CHD, definite or probable myocardial infarction, and silent CHD (as determined by EKGs). Definite or probable hospitalized embolic or thrombotic strokes were determined from diagnosis codes (ICD-9 codes 430–437), hospital records, and neuroimaging reports. Eligible ischemic stroke cases were classified by a computer algorithm and a physician reviewer, and differences were adjudicated based on the event criteria adapted from the National Survey of Stroke²⁰. Follow-up information was available through 2019. All potential CHD and stroke events were classified as definite, probable, or suspected according to standardized criteria^{21, 22}. Incident CHD events, ischemic stroke, and death that occurred after dental assessment on visit 4 were included in the analysis of outcomes.

Assessment of incident AF:

AF was assessed by study EKGs, hospital discharge diagnosis codes, and death certificates¹¹. Standard 10-second, 12-lead EKGs were obtained at baseline and at each of the subsequent follow-up examinations. Tracings were transmitted electronically to the ARIC EKG Reading Center (Epidemiological Cardiology Research Center, Wake Forest School of Medicine, Winston Salem, North Carolina), where they underwent automated reading and coding. Tracings with AF were reviewed by a cardiologist. AF was identified from hospitalizations or death certificates using International Classification of Diseases, Ninth Revisions, and Clinical Modification (ICD-9-CM) codes 427.31 or 427.32. The methodology has high sensitivity and specificity¹¹. For analysis of data related to incident AF, those with AF detected prior to visit 4 (N=90, 1.4%) were excluded.

Assessment of Dental Care Utilization:

Dental care utilization was classified using participant-reported responses to the administered dental history questionnaire. Subjects were classified as regular users (those who sought routine dental care) or episodic users (those who sought dental care only when symptomatic).

Statistical Analysis:

The cohort consisted of 6,351 participants assessed for dental caries at visit 4. Demographics and covariates, including age, sex, race, body mass index, hypertension, diabetes, education, smoking status, alcohol use, and cholesterol, were assessed using a t-test for continuous variables and a Χ² test for categorical variables. Triglycerides and serum CRP are not normally distributed and were assessed using a non-parametric Mann-Whitney U test. Gray’s test was run to determine differences in the time-to-event distribution for incident ischemic stroke, CHD event, and death between the dental caries and no dental caries groups. The proportional hazards assumption was tested using interaction terms between time measures and covariates in the model. Cox proportional hazards models were then used to assess crude, followed by adjusted, hazard ratios (HRs), and 95% confidence intervals of incident ischemic stroke, CHD events, and death, from the initial dental assessment. Following the crude hazards ratio analyses, stratified analyses and tests for interaction were run using individual covariates in the crude model. The interaction term and stratified analysis for ischemic stroke, CHD events, and all-cause mortality are depicted in Figures 1A, 1B, and 1C respectively. The associations between dental caries and ischemic stroke, CHD events, and death were adjusted for covariates that did not meet the criteria for significant interaction. Those with significant interaction on stratified analysis are reported as effect modifiers, such as race, and were not included in the model as confounders. Since all-cause mortality would be a competing risk for both CHD event and ischemic stroke, we used the Fine and Gray method to estimate multivariable-adjusted subdistribution relative hazards of ischemic stroke and CHD event respectively, while adjusting for the same covariates as listed above²³. Our data may have unmeasured confounders (for example dietary factors associated with caries). As such, we assessed the sensitivity of the caries results in fitted survival models to the unmeasured confounders²⁴.

Figure 1A:

Subgroup analysis, test for interaction dental surface caries and incident ischemic stroke

Figure 1B:

Subgroup analysis, test for interaction dental surface caries and incident CHD event

Figure 1C:

Subgroup analysis, test for interaction dental surface caries and death from all cause

By using structural equation modeling (SEM) of the Baron and Kenny approach, mediation analysis was performed to determine the role of incident AF in mediating the effect of caries on incident ischemic stroke. Additionally, the proportion mediated was computed from the difference in the natural direct and natural indirect effect²⁵. Finally, we assessed if regular dental care use reduced the chances of developing dental caries on univariate (Odds Ratio analysis and X² test) and multivariable logistic regression analysis. All the data analyses for this study were conducted utilizing SAS version 9.4 (SAS Institute).

RESULTS

At visit 4, individuals included in the DARIC study (n=6,351) were compared to those excluded (n=5,305) from the study (Table 1). Of the 6,351 participants assessed in DARIC at visit 4 without prior stroke or CHD events, 1,171 were found to have dental caries and 5,180 were noted without caries (Table 2). The group of subjects with dental caries was significantly younger, had a higher proportion of males compared to females, was more likely to be African American than white, and had a higher mean BMI and waist-to-hip ratio when compared to those without dental caries. They were also more likely to have hypertension, diabetes, periodontal disease, and more likely to be current smokers. On the opposite end of the scale, they were also less likely to have intermediate or advanced education and less likely to be current alcohol users. Additionally, the dental caries group had a higher LDL cholesterol, hs-CRP, and a lower median triglyceride level compared to those without dental caries.

Table 1.

Characteristics of the Study Participants, according to inclusion/exclusion in DARIC at visit 4

	DARIC
	Excluded (n = 5305)	Included (n=6351)	p value
Age (yr) †	63.4±5.7	62.3±5.6	<0.001
Sex (%)
Female	57.2	54.7	0.009
Male	42.8	45.3
Race (%)
African-American	28.6	18.2	<0.001
White	71.4	81.8
Body Mass Index†	29.1±5.9	28.6±5.3	<0.001
Waist-to-hip ratio†	0.95±0.07	0.94±0.07	<0.001
Hypertension (%)	45.1	33.2	<0.001
Diabetes (%)	21.3	13.2	<0.001
Education (%)
Basic	27.4	12.7	<0.001
Intermediate	40.6	43.1
Advanced	32.0	44.2
Smoking (%)
Never	39.8	47.5	<0.001
Former	41.9	40.5
Current	18.3	12.0
Alcohol (%)
Never	23.1	19.0	<0.001
Former	34.6	26.2
Current	42.3	54.7
Cholesterol (mg/dl) †	200.3±38.4	201.2±35.9	0.21
LDL†	122.4±35.1	122.0±33.0	0.58
HDL†	49.1±16.2	50.7±16.8	<0.001
Triglycerides (mg/dl)‡	122 (90–176)	123 (87–172)	0.200

^†mean±standard deviation

^‡median (inter-quartile range)

Table 2.

Characteristics of the Study Participants, according to Presence of Dental Caries

	Dental Caries
	Absent (n = 5,180)	Present (n=1,171)	p value
Age (yr) †	62.5±5.6	61.8±5.6	<0.001
Sex (%)
Female	56.0	49.7	<0.001
Male	44.0	50.3
Race (%)
African American	10.1	46.9	<0.001
White	89.9	53.1
Body Mass Index†	28.2±5.7	30.0±6.0	<0.001
Waist-to-hip ratio†	0.94±0.07	0.96±0.07	<0.001
Hypertension (%)	31.1	42.6	<0.001
Diabetes (%)	12.0	18.7	<0.001
Education (%)
Basic	9.4	27.3	<0.001
Intermediate	43.9	39.8
Advanced	46.8	32.9
Smoking (%)
Never	47.4	47.9	<0.001
Former	41.5	35.8
Current	11.1	16.3
Alcohol (%)
Never	16.9	29.9	<0.001
Former	24.8	32.3
Current	58.4	37.8
Cholesterol (mg/dl) †	201.3±35.2	201.1±39.0	0.88
LDL†	121.5±32.4	124.3±35.5	0.02
HDL†	50.1±16.8	50.0±16.5	0.12
Triglycerides (mg/dl)‡	125 (89–175)	111 (81–161)	<0.001
High sensitivity CRP (mg/dl)‡	2.6 (0.9–6.8)	3.2 (1.2–7.6)	0.001
Periodontal disease	67.3	94.0	<0.001

^†mean±standard deviation

^‡median (inter-quartile range)

During a median of a 21-year follow-up period, the group with dental caries was associated with higher cumulative incidence of ischemic stroke (Figure 2A, Gray’s test p<0.0001), higher cumulative incidence of CHD events (Figure 2B, Gray’s test p=0.0187), and higher cumulative incidence of all-cause mortality (Figure 2C, Gray’s test p<0.0001), when compared to the group without dental caries. Of note, the cumulative incidence of ischemic stroke in the dental caries group was higher than the no caries group after an initial 3–5 years period after visit 4. However, dental caries still met the proportional hazards assumption using the time interaction criteria.

Figure 2A:

Cumulative incident function of incident ischemic stroke with or without evidence of dental caries.

Figure 2B:

Cumulative incident function of CHD event with or without evidence of dental caries.

Figure 2C:

Cumulative incident function of death with or without evidence of dental caries.

During the same 21-year follow-up period, incident ischemic stroke was noted in 113 (10 %) out of the 1,171 participants with dental caries, and 328 (6%) out of 5,180 without dental caries (Table 3). Univariable Cox regression analysis showed a crude HR of 1.70 (95% CI 1.37–2.10). In addition, an incident ischemic CHD event was noted in 163 (14 %) out of 1,171 with dental caries and 634 (12%) out of 5,180 participants without dental caries. Univariable Cox regression analysis showed a crude HR of 1.23 (95% CI 1.03–1.46). During a median 21-year follow-up period, incident all-cause mortality was noted in 534 (46 %) out of 1,171 with dental caries and 2,051 (40%) out of 5,180 without dental caries. Univariable Cox regression analysis showed a crude HR of 1.24 (95% CI 1.12–1.36). In model 1, HR was adjusted for the covariates age, gender, hypertension, diabetes, BMI, LDL cholesterol, smoking status, three-level educational status, and presence of periodontal disease. Dental caries were significantly associated with ischemic stroke (adjusted HR 1.40 95% CI 1.10–1.79), all-cause mortality (adjusted HR 1.13 95% CI 1.01–1.26), but not CHD event (adjusted HR 1.13 95% CI 0.93–1.37). Since all-cause mortality would be a competing risk for both CHD events and ischemic stroke, in model 2, we report a Fine and Gray Subdistribution HR that included death as an additional event type, adjusting for the same covariates as listed above. Dental caries were significantly associated with ischemic stroke (adjusted HR 1.19 95% CI 1.06–1.33), as well as CHD event (adjusted HR 1.16 95% CI 1.02–1.32). Full model information are included in Tables S1A–C.

Table 3.

Hazard ratios (HR) and 95% confidence interval (95% Cis) for the association of presence of dental caries, incident ischemic stroke, CHD event (CHD death and myocardial infarction) and death (all-cause mortality)

	Dental Caries		Unadjusted			Model 1 Cox Proportional Hazards			Model 2 Fine and Gray, Adjusted Subdistribution
	Absent (n=5,180)	Present (n=1,171)	Crude HR	95% CI	p values	Adjusted HR*	95% CI	p values	Adjusted HR†	95% CI	p values
Stroke	328	113	1.70	1.37–2.10	<0.001	1.40	1.10–1.79	0.01	1.19	1.06–1.33	0.003
CHD event	634	163	1.23	1.03–1.46	0.02	1.13	0.93–1.37	0.22	1.16	1.03–1.32	0.02
Death	2051	534	1.24	1.12–1.36	<0.001	1.13	1.01–1.26	0.04		-

^*Adjusted for age, gender, hypertension, diabetes, BMI, LDL cholesterol, smoking status, three-level educational status and presence of periodontal disease.

^†Fine and Gray, Adjusted Subdistribution HR that included death as a competing event, also adjusted for age, gender, hypertension, diabetes, BMI, LDL cholesterol, smoking status, three-level educational status and presence of periodontal disease.

The association of dental caries and ischemic incident stroke was significantly higher in the African Americans population (HR 2.89, 95% CI 1.82–4.59) when compared to the white subgroup (HR 0.96, 95% CI, 0.67–1.38) (Figure 1A). On the introduction of the interaction terms of dental caries and race group into the Cox regression model, the interaction term was statistically significant (p=0.0001), suggestive of effect modification. The interaction remained significant after adding the remaining covariates as confounders in the model (p=0.0001). The interaction for other covariates remained non-significant. While the association between race and CHD event were similar across both races (Figure 1B), the African American subgroup had a stronger association with all-cause mortality (HR 1.54, 95% CI 1.39–1.89) when compared to the white subgroup (HR 1.20, 95% CI, 1.05–1.37) with the interaction term achieving statistical significance (p=0.03, Figure 1C). The association of dental caries and incident all-cause mortality was significantly higher in the no diabetes group (HR 1.32, 95% CI 1.18–1.47) when compared to the diabetes subgroup (HR 1.02 95% CI 0.83–1.25). On the introduction of the interaction terms of dental caries and diabetes into the Cox regression model, the interaction term was statistically significant (p=0.03). The interactions for all other covariates were statistically non-significant.

The severity of dental caries was gauged by the presence of DMFS (median 68, IQR 50–87). Increasing DMFS was significantly associated with ischemic stroke (Crude HR 1.008, 95% CI 1.004–1012), CHD events (Crude HR 1.006, 95% CI 1.004–1.008), and all-cause mortality (Crude HR 1.007, 95% CI 1.005–1.008), suggestive of a dose response of DMFS to outcome (Figure S1). The association remained significant for ischemic stroke (Adjusted HR 1.006, 95% CI 1.001–1.011) and all-cause mortality (Adjusted HR 1.003, 95% CI 1.001–1.005) after adjusting for the covariates age, gender, hypertension, diabetes, BMI, LDL cholesterol, smoking status, three-level educational status, and periodontal disease. The sensitivity of caries results in fitted survival models to the unmeasured confounder is shown in Table S2. To fully mediate the association of caries, an unmeasured confounder must have an HR of at least 3.0 (assuming a difference of at least 30% of prevalence of the confounder between the caries and no caries group) to result in a non-significant association between caries and ischemic risk.

Using SEM (Baron and Kenny approach), we found that incident AF is not a likely mediator of dental caries-ischemic stroke association (Figure S2). Dental caries was not associated with incident AF (HR 1.03 95% CI 0.88–1.19). In the mediated model the effect of dental caries on ischemic stroke (HR 1.75 95% CI 1.40–2.18) was similar to the effect in the unmediated model (HR 1.74 95% CI 1.40–2.15). The proportion mediated was small and not significant (0.03, p=0.43).

DISCUSSION

The prevalence of dental caries noted in our study is approximately 1 in 5 reported in the US adult population³. We report an independent association between dental caries and both ischemic stroke and all-cause mortality. The association remained significant upon the use of an adjusted sub-distribution model accounting for death as a competing event. Additionally, we report associations between DMFS and outcomes, with significant independent associations noted between DMFS and ischemic stroke, as well as all-cause mortality. Case-control studies that measured the number of decayed, missing, and filled teeth (DMFT) as a surrogate of dental caries show mixed results, with some showing an association and a possible dose response with ischemic stroke^{26, 27}, whereas others failed to find a significant association^28–31. A recent meta-analysis of pooled data showed a significantly higher mean DMFT score among patients with ischemic stroke³². Following the meta-analysis, an analysis of data from the Third National Health and Nutrition Examination Survey (NHANES III) showed that the number of missing teeth was independently associated with self-reported history of stroke³³. The association with dental caries was statistically non-significant. Recent reports from the large Korean nationwide Cohort study found a dose-dependent association between tooth loss (a marker of both PD and caries) and incident myocardial infarction, heart failure, ischemic stroke, and all-cause death³⁴. After a median follow-up of 9.5 years, the risk of major CHD events was higher when a subject had a higher number of dental caries, periodontal disease, or more tooth loss³⁵. This study did not investigate the individual categories of CHD events separately. The study also reported that one or more tooth brushing sessions a day were associated with a 9% significantly lower risk of CHD events. Once a year or more dental visits for cleaning were also shown to reduce CHD risk by 14%. In line with these findings, we have reported in the past that regular dental care was associated with 23% lower rates of ischemic stroke compared with regular dental care users after adjustment for covariates¹¹. We now report that regular dental care users have a significantly lower risk of dental caries when compared with episodic dental care users. The finding that regular dental care by lowering the risk of dental caries may aid in stroke prevention, emphasizes the importance of oral health in maintaining cerebrovascular health.

We observed a racial disparity in the prevalence of caries. Prior studies have noted disparities in the prevalence of untreated decayed teeth between African American (42%) and White (22%) populations³⁵. This disparity is partially explained by sociocultural differences between racial groups. African American individuals are less likely to have access to and utilize oral health care means³⁶. Other factors contributing to the disparities in oral care include differences in oral health education, socioeconomic status, and transmission of cariogenic bacteria. Caries may be a contributing factor to the higher risk of ischemic stroke noted among young African American individuals³⁷. Stroke continues to disproportionately affect African Americans³⁸. Stroke racial disparities are not entirely explained by traditional stroke risk factors. The Reasons for Geographic and Racial Differences in Stroke (REGARDS) study found that tooth loss due to PD was more common among African American participants than white participants, and tooth loss was associated with higher stroke risk and stroke risk factors in the stroke belt³⁹. Given these two associations, REGARDS investigators proposed that PD may be contributing to the racial disparity in stroke. Additional results from this study show that low socioeconomic status is associated with greater tooth loss and is seen more frequently in the African American population as opposed to their white counterparts³⁹.

In addition to the racial disparity in the prevalence of caries, we noted significant interaction in both caries-stroke, as well as caries-death associations. The reason for these interactions in the adult population needs further investigation. Diabetes had a negative interaction with all-cause mortality, as there was a stronger association between caries and death noted among those without diabetes (HR 1.32, 95% CI 1.18–1.47) compared to the diabetes subgroup (HR 1.02 95% CI 0.83–1.25). This may have been driven by the proportion of non-CHD mortality in the no diabetes group, included within all-cause mortality.

As we did not find a mediator role of incident AF in the association between dental caries and ischemic stroke, it remains unknown if other mechanisms may be at play. Streptococcus mutans, the primary pathogen causing dental caries, was detected at higher frequencies and quantities than the periodontal pathogens in both heart valve tissues obtained during heart valve surgery and atheromatous plaque samples obtained from aortic samples during CHD surgeries⁸. Due to this finding, it is possible to speculate that Streptococcus mutans in dental caries may be responsible for ischemic stroke via cardioembolism (infective endocarditis) and/or atherothrombosis and risk factors (for example diabetes) of cerebral vasculature. This may also explain the 3–5 years delayed increase in the cumulative incidence of ischemic stroke in the dental caries group noted in Figure 1A.

Few limitations were observed in our study. The individuals included in the study were younger, had lower BMI, less prevalent hypertension, diabetes, and higher levels of education. This may be attributed to older persons with overall “bad oral health” who had been excluded due to tooth loss. Other limitations of the study include a single dental assessment and the use of education level only as a surrogate for socioeconomic status. Limitations also include residual and unmeasured confounding that might be present despite trying to account for potential confounders in the statistical analysis. To address the issue of unmeasured confounders, we conducted a sensitivity analysis to test its effect on the association between caries and ischemic stroke. To fully mediate the association of caries, the unmeasured confounder must have an HR of at least 3.0, an unlikely scenario. Despite this, the strength of our study includes its large community-based cohort, long follow-up duration, adjudication of outcome events, and assessment of multiple confounders including periodontal disease. Since periodontal disease is associated with dental caries, it may be difficult to disentangle the independent effect of caries specifically on cardiovascular outcomes.

We believe this may be one of the first studies to evaluate the association between dental caries and incident ischemic stroke within a large US community-based cohort. Regular dental use is associated with lower levels of dental caries, suggesting that it may be a means of lowering this risk for incident ischemic stroke.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

AF

Atrial Fibrillation

ARIC

Atherosclerosis Risk In Communities

CHD

Coronary Heart Disease

CI

Confidence Interval

DARIC

Dental cohort of ARIC

DMFS

Decayed, Missing, and Filled Surfaces

DMFT

Decayed, Missing, and Filled Teeth

HR

Hazards Ratio

OR

Odds Ratio

PD

Periodontal Disease

PPC

Periodontal Profile Class

SEM

Structural Equation Modeling