Dementia and the Risk of Periodontitis: A Population-Based Cohort Study

Abstract

Dementia and Alzheimer’s disease (AD) are proposed to be comorbid with periodontitis (PD). It is unclear whether PD is associated with dementia and AD independent of confounding factors. We aimed at identifying the relationship between the longitudinal risk of developing PD in a cohort of patients with dementia and AD who did not show any signs of PD at baseline. In this retrospective cohort study, 8,640 patients with dementia without prior PD were recruited, and 8,640 individuals without dementia history were selected as propensity score–matched controls. A Cox proportional hazard model was developed to estimate the risk of developing PD over 10 y. Cumulative probability was derived to assess the time-dependent effect of dementia on PD. Of the 8,640 patients, a sensitivity test was conducted on 606 patients with AD-associated dementia and 606 non-AD propensity score–matched controls to identify the impact of AD-associated dementia on the risk for PD. Subgroup analyses on age stratification were included. Overall 2,670 patients with dementia developed PD. The relative risk of PD in these patients was significantly higher than in the nondementia group (1.825, 95% CI = 1.715 to 1.942). Cox proportional hazard models showed that patients with dementia were more likely to have PD than individuals without dementia (adjusted hazard ratio = 1.915, 95% CI = 1.766 to 2.077, P < 0.0001, log-rank test P < 0.0001). The risk of PD in patients with dementia was age dependent (P values for all ages <0.0001); younger patients with dementia were more likely to develop PD. The findings persisted for patients with AD: the relative risk (1.531, 95% CI = 1.209 to 1.939) and adjusted hazard ratio (1.667, 95% CI = 1.244 to 2.232; log-rank test P = 0.0004) of PD in patients with AD were significantly higher than the non-AD cohort. Our findings demonstrated that dementia and AD were associated with a higher risk of PD dependent of age and independent of systemic confounding factors.

Introduction

Dementia is a complex neurodegenerative disease with a prevalence of 5% to 7% among populations aged ≥60 y. Several forms of dementia, such as vascular dementia, dementia with Lewy bodies, and frontotemporal dementia, are characterized by various risk factors and pathogenesis (Prince et al. 2013). As the most common form of dementia, Alzheimer’s disease (AD) serves as a leading cause of primary degenerative dementia, with odds of 1:3 among patients >85 y old (Winblad et al. 2016). The pathologic hallmarks of the AD are the formation of extracellular amyloid β peptide plaques and intraneuronal neurofibrillary tangles of hyperphosphorylated tau protein, followed by loss of synapses and neuronal degeneration (Bloom 2014). Risk factors for AD and dementia include cardiovascular diseases (CVDs), hyperlipidemia (HLD), malnutrition, and low physical activity (Livingston et al. 2020), which are also predisposing factors of periodontitis (PD; Slots 2017; Demmer et al. 2020; Nadim et al. 2020). Moreover, neurodegenerative sequelae and complications of dementia have been proposed to lead to a lack of motor skills and reduced manual dexterity required for oral care (Rozas et al. 2017), which may place patients with dementia or AD at high risk of PD.

PD is a chronic inflammatory disease initiated by dysbiotic oral biofilm affecting the supporting tissues of the teeth, including gingiva, periodontal ligament, cementum, and alveolar bone. Clinical presentation of PD involves progressive destruction of the periodontium, followed by pocket formation, gingival recession, alveolar bone loss, tooth mobility, and ultimately tooth loss (Slots 2017). The interaction between PD and inflammatory diseases has been proposed (Ma, Chiang, Chen, and Wang 2021; Ma, Chiang, Lopez, and Wang 2021; Ma, Lai, Veeravalli, Chiu, Van Dyke, and Wei 2021). Following PD, tooth loss is the second strongly associated oral condition with AD and/or dementia (Li et al. 2000). Thus, there is biological plausibility of a potential link between PD and neurodegenerative diseases (Noble et al. 2009; Hashioka et al. 2018). Cross-sectional human studies and longitudinal epidemiologic data support this link (Sparks Stein et al. 2012). Manifestations observed in patients with PD and dementia include poor nutrition, decreased masticatory function (Tucker et al. 2005; Kim et al. 2007), hippocampal morphologic impairments, and hippocampus-dependent spatial memory deficits (Chen et al. 2015). However, it is unknown whether PD risk is associated with dementia or specifically AD-associated dementia, characterized by neuroinflammatory pathogenesis (Bonham et al. 2018). Due to the absence of large-scale cohort studies elucidating these associations, we focused on how periodontal health was affected and whether the risk of developing PD was increased in patients with dementia and specifically AD-associated dementia in a large population with available longitudinal data.

Materials and Methods

Study Design and Data Sources

The present study was a retrospective longitudinal cohort study based on data collected from the Longitudinal Health Insurance Dataset (LHID) in Taiwan. The database includes all claimed diagnoses and treatment from outpatient visits, emergency, and hospitalization medical records. From the LHID, eligible patients diagnosed with dementia from 2000 to 2013 were enrolled for the dementia cohort. A subset with AD was included for the sensitivity test that focused on AD-associated dementia. Propensity score–matched participants without dementia and AD were selected as controls. After exclusion of participants with PD prior to dementia onset, all enrolled patients with dementia and controls did not have any history of PD. The sampled database was reidentified to remove patient identifications. This study followed the principles of the Declaration of Helsinki (General Assembly of the World Medical Association 2014) and the National Statements of Ethical Conduct Elements, from which the values of research merit and integrity, justice, beneficence and respect were highlighted (National Bioethics Advisory Commission 2001). The study followed the STROBE guidelines to structure the cohort study (Strengthening the Reporting of Observational Studies in Epidemiology; von Elm et al. 2007) and was approved by the Institutional Review Board of Chung Shan Medical University Hospital (CS15134).

Diagnostic Criteria for Dementia and Subgroup Analyses for the Dementia Cohort

The clinical diagnosis of dementia was made following criteria from the Diagnostic and Statistical Manual of Mental Disorders (fifth edition) for major neurocognitive disorder or dementia (American Psychiatric Association 2013), which required evidence of significant cognitive decline in 1 or more of the following domains: complex attention, executive function, learning and memory, language, perceptual-motor function, and social cognition. In particular, cognitive deficits of the 6 domains must be sufficient to cause loss of independent functioning and must not be attributable to other mental disorders. To identify whether the risk of PD among patients with dementia was age dependent, a subgroup analysis stratified by age was carried out specifying dementia onset at age 51 to 60 y, 61 to 70 y, 71 to 80 y, and >81 y. The diagnostic criteria of AD are outlined in the Appendix.

Outcome Measurement

The diagnosis of PD, the primary outcome of this study, was based on periodontal examinations and panoramic radiographs evaluating 1) probing depth ≥5 mm in at least 4 teeth with each ≥1 site, 2) clinical attachment level (CAL) ≥5 mm on the same site, and 3) observed bleeding on probing, with the diagnosis of PD having to meet all 3 criteria (Caton et al. 2018). The periodontal evaluation and corresponding diagnosis were made by dentists from medical centers, community hospitals, and private dental clinics. The diagnosis of chronic PD was confirmed in at least 2 outpatient visits in 2 y and peer reviewed by other dentists and the board of medical records to ensure its reliability and repeatability. The study was followed up until the occurrence of PD by December 2013 or withdrawal from the national insurance system, whichever occurred first.

To ensure having sufficient and appropriate data to test the study questions, we excluded the following patients: 1) those with missing demographic data (n = 72), 2) those who were edentulous (n = 83), 3) those diagnosed with PD within 2 y before dementia or AD onset (n = 6,916), 4) those who never visited the dental clinic within 2 y before dementia or AD onset (n = 22,524), and 5) those whose diagnoses were made before 2000 or after 2012 (n = 1,374).

Covariates and Comorbidities Adjusted in Propensity Score Matching

Propensity score matching is a probability estimated through logistic regression. We compared the outcome between those who did and did not have dementia by balancing the heterogeneity of baseline characteristics and comorbidities: age, sex, income, urbanization, diabetes mellitus, stroke, hypertension, traumatic brain injury, chronic obstructive pulmonary disease (COPD), CVD, and HLD. HLD was defined as any of the following: total cholesterol >240 mg/dL, triglyceride >150 mg/dL, low-density lipoprotein cholesterol >130 mg/dL, or total cholesterol/high-density lipoprotein cholesterol ratio >5.0 (Catapano et al. 2016). Those comorbidities were considered if they took place within 1 y before the onset of dementia. Propensity score matching was performed by adjusting age, sex, income, urbanization, diabetes mellitus, stroke, hypertension, traumatic brain injury, COPD, CVD, HLD, and the index year of dementia onset.

Effect of Dementia and AD on PD in Subgroups

To identify subgroups of dementia/AD that were more susceptible to PD, patients with dementia/AD were stratified by sex and age of dementia/AD onset and whether they had concomitant diabetes mellitus, stroke, hypertension, traumatic brain injury, COPD, CVD, or HLD. Among these comorbidities, the risk of PD in probable vascular dementia was inferred as the risk of PD in patients with dementia/AD and concomitant stroke. In this case, vascular dementia defined as stroke-associated dementia/AD events, following the diagnostic criteria of the National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherche et l’Enseignement en Neurosciences (Roman et al. 1993).

Statistical Analysis

The comparison of the dementia and nondementia groups was made with the absolute standardized difference. When the absolute standardized difference was <0.1, the characteristics of both groups were defined as being similar. Kaplan-Meier analysis was used to calculate the cumulative incidence of chronic PD, and the log-rank test was used to test the significant difference between the groups.

A Cox proportional hazard model was used to estimate the hazard ratio (HR) of chronic PD between the dementia and nondementia groups. In a multiple Cox proportional hazard regression model, the time variable for the dementia or AD cohort was the period from the diagnosis of dementia or AD to the diagnosis of chronic PD. In the dementia or AD cohort, the independent variable was the diagnosis of dementia or AD with covariates such as sex, age, comorbidities at baseline, and socioeconomic status, while the dependent variable was the diagnosis of chronic PD. The statistical software was SPSS 18.0 (IBM).

Results

Demographics of the Study Patients

A total of 42,269 patients who were newly diagnosed with dementia were identified from the LHID. After propensity score matching, 8,640 patients with dementia were selected with a matched age distribution for the nondementia group for the final cohort (Table 1). Among them, 606 patients with AD (Appendix Fig.) were selected and matched by age with 606 patients from a group without AD (P = 0.9616). There was no statistically significant difference between the dementia and nondementia groups or the AD and the non-AD groups (all the absolute standardized differences < 0.1).

Table 1.

Baseline Characteristics for the Dementia and Alzheimer’s Disease Cohorts After Propensity Score Matching.

	Dementia, n (%)			Alzheimer’s Disease, n (%)
	Control (n = 8,640)	Case (n = 8,640)	ASD	Control (n = 606)	Case (n = 606)	ASD
Sex			0.0071			0.0232
Female	4,434 (51.32)	4,465 (51.68)		340 (56.11)	333 (54.95)
Male	4,206 (48.68)	4,175 (48.32)		266 (43.89)	273 (45.05)
Age, y			0.0651			0.0000
≤60	1,913 (22.14)	2,065 (23.90)		49 (8.09)	46 (7.59)
61 to 70	1,591 (18.41)	1,629 (18.85)		99 (16.34)	95 (15.68)
71 to 80	3,018 (34.93)	2,930 (33.91)		247 (40.76)	247 (40.76)
≥81	2,118 (24.51)	2,016 (23.33)		211 (34.82)	218 (35.97)
Comorbidities
Diabetes mellitus	1,840 (21.30)	1,800 (20.83)	0.0113	133 (21.95)	137 (22.61)	0.0159
Stroke	624 (7.22)	620 (7.18)	0.0017	52 (8.58)	53 (8.75)	0.0059
Hypertension	4,730 (54.75)	4,584 (53.06)	0.0339	339 (55.94)	350 (57.76)	0.0367
TBI	254 (2.94)	260 (3.01)	0.0040	23 (3.80)	27 (4.46)	0.0332
COPD	1,378 (15.95)	1,366 (15.81)	0.0038	94 (15.51)	102 (16.83)	0.0359
CVD	1,978 (22.89)	1,989 (23.02)	0.0030	132 (21.78)	138 (22.77)	0.0238
Hyperlipidemia	1,957 (22.65)	1,896 (21.94)	0.0169	128 (21.21)	136 (22.44)	0.0320

Values are presented as No. (%) unless noted otherwise.

ASD, absolute standardized difference; COPD, chronic obstructive pulmonary disease; CVD, cardiovascular diseases; TBI, traumatic brain injury.

Risk of Developing PD in Patients with Dementia

Among the 8,640 patients who had dementia, 2,670 developed PD over 10 y. The relative risk of developing PD in patients who had dementia was significantly higher than in the nondementia group (1.825, 95% CI = 1.715 to 1.942; Table 2). The Cox proportional hazard model revealed that after adjusting for the confounding variables, the risk for developing PD in patients with dementia continued to be significantly higher than in the nondementia group (adjusted HR = 1.915, 95% CI = 1.766 to 2.077, P < 0.0001; Table 3). The cumulative probability of PD in the Kaplan-Meier curves of the dementia group was significantly higher than that of the nondementia group over a 120-mo follow-up, with a log-rank test for the comparison of cumulative incidence curves demonstrating a statistical significance (P < 0.0001; Fig. A).

Table 2.

Incidence and Hazard Ratio of PD for the Dementia and Alzheimer’s Disease Cohorts After Propensity Score Matching.

	Dementia		Alzheimer’s Disease
	Control (n = 8,640)	Case (n = 8,640)	Control (n = 606)	Case (n = 606)
New PD case, n (%)	1,582 (18.31)	2,670 (30.90)	123 (20.30)	160 (26.40)
Incidence rate (95% CI)	3.80 (3.62 to 4.00)	6.99 (6.73 to 7.27)	3.76 (3.15 to 4.49)	6.09 (5.21 to 7.11)
Crude relative risk (95% CI)	Reference	1.825 (1.715 to 1.942)	Reference	1.531 (1.209 to 1.939)
Adjusted HR (95% CI)	Reference	1.915 (1.766 to 2.077) a	Reference	1.667 (1.244 to 2.232) b

HR, hazard ratio; PD, periodontitis.

^aP < 0.0001.

^bP = 0.0006.

Table 3.

Multiple Cox Proportional Hazard Regression for Estimation of PD for the Dementia and Alzheimer’s Disease Cohorts by Presenting Comorbidity.

	Dementia, Adjusted HR (95% CI)		Alzheimer’s Disease, Adjusted HR (95% CI)
Comorbidity a	Effect on PD b	P Value	Effect on PD b	P Value
Diabetes mellitus	0.952 (0.883 to 1.025)	0.1905	0.981 (0.751 to 1.282)	0.8905
Hypertension	0.999 (0.938 to 1.064)	0.9684	1.074 (0.859 to 1.342)	0.5321
TBI	0.917 (0.818 to 1.028)	0.1378	1.065 (0.688 to 1.649)	0.7772
COPD	0.996 (0.918 to 1.080)	0.9222	1.176 (0.889 to 1.556)	0.2558
CVD	1.099 (1.024 to 1.180)	0.0091	1.082 (0.840 to 1.394)	0.5421
Hyperlipidemia	1.066 (0.994 to 1.142)	0.0722	1.081 (0.836 to 1.397)	0.5533

ASD, absolute standardized differences; COPD, chronic obstructive pulmonary disease; CVD, cardiovascular diseases; HR, hazard ratio; PD, periodontitis; TBI, traumatic brain injury.

^aReference: nondementia and non-AD.

^bWhen matched on sex and age before propensity score matching.

Figure.

Kaplan-Meier survival curves of cumulative probability for developing periodontitis (PD) in the dementia cohort and the Alzheimer’s disease (AD) cohort after propensity score matching (PSM). (A) PD risks in the dementia cohort. The survival curve suggests that patients with dementia have higher risks of developing PD after PSM. Subgroup analysis based on age stratification demonstrates PD risks in patients with dementia aged (B) <61 y, (C) 61 to 70 y, (D) 71 to 80 y, and (E) >80 y, which indicates that the risk of developing PD after dementia onset is age dependent. The younger these patients were, the more likely that they would develop PD after dementia onset. (F) PD risks in the AD cohort, in which patients with AD have higher risks of developing PD after PSM.

A subgroup analysis based on age stratification suggested that among all age groups (≤60, 61 to 70, 71 to 80, and >80 y), patients with dementia consistently presented with a higher risk of developing PD (P < 0.0001; Fig. A–E). Kaplan-Meier curves of subgroup analysis suggested that the risk of developing PD after dementia onset was age dependent: the younger the patients with dementia were, the more likely they would develop PD after dementia onset, suggesting a cumulative effect of dementia on the risk of PD. Specifically, the effect of dementia on risk of PD was the highest in participants aged <60 y (adjusted HR = 2.126, 95% CI = 1.913 to 2.361), followed by 61 to 70 y (adjusted HR = 1.721, 95% CI = 1.511 to 1.96), 71 to 80 y (adjusted HR = 1.74, 95% CI = 1.546 to 1.957), and >80 y (adjusted HR = 1.835, 95% CI = 1.537 to 2.19; Table 4).

Table 4.

Subgroup Analysis for Estimation of Incidence Rates and Adjusted HRs on Periodontitis for the Dementia and Alzheimer’s Disease Cohorts Stratified by Sex, Age, and Comorbidity.

	Dementia, Incidence Rate (95% CI)			Alzheimer’s Disease, Incidence Rate (95% CI)
	Control	Case	Adjusted HR (95% CI)	Control	Case	Adjusted HR (95% CI)
Sex
Female	3.70 (3.45 to 3.97)	7.14 (6.78 to 7.52)	1.946 (1.784 to 2.122)	3.82 (3.04 to 4.79)	6.17 (5.04 to 7.56)	1.53 (1.123 to 2.084)
Male	3.92 (3.65 to 4.20)	6.83 (6.46 to 7.22)	1.765 (1.612 to 1.932)	3.68 (2.78 to 4.87)	5.97 (4.69 to 7.60)	1.631 (1.123 to 2.368)
Age, y
≤60	4.50 (4.14 to 4.89)	9.28 (8.73 to 9.87)	2.126 (1.913 to 2.361)	3.50 (1.94 to 6.32)	9.23 (5.95 to 14.3)	2.657 (1.225 to 5.766)
61 to 70	4.25 (3.85 to 4.70)	7.42 (6.83 to 8.05)	1.721 (1.511 to 1.96)	4.40 (3.02 to 6.42)	7.16 (5.12 to 10.0)	1.61 (0.95 to 2.727)
71 to 80	3.31 (3.02 to 3.64)	5.76 (5.36 to 6.19)	1.74 (1.546 to 1.957)	3.65 (2.77 to 4.82)	6.35 (5.01 to 8.06)	1.692 (1.169 to 2.448)
≥81	2.91 (2.52 to 3.34)	5.10 (4.59 to 5.67)	1.835 (1.537 to 2.19)	3.59 (2.58 to 5.00)	4.38 (3.19 to 6.02)	1.129 (0.711 to 1.793)
Comorbidity
DM	3.48 (3.09 to 3.92)	6.24 (5.68 to 6.85)	1.784 (1.533 to 2.076)	4.97 (3.47 to 7.11)	4.34 (2.93 to 6.42)	0.91 (0.529 to 1.565)
Stroke	3.60 (2.88 to 4.50)	5.61 (4.77 to 6.60)	1.546 (1.17 to 2.043)	4.49 (2.55 to 7.91)	6.90 (4.22 to 11.2)	0.921 (0.361 to 2.35)
Hypertension	3.65 (3.40 to 3.92)	6.31 (5.96 to 6.68)	1.735 (1.582 to 1.903)	3.95 (3.12 to 4.99)	6.27 (5.12 to 7.67)	1.497 (1.096 to 2.044)
TBI	3.36 (2.42 to 4.66)	7.04 (5.59 to 8.86)	1.879 (1.226 to 2.88)	5.36 (2.01 to 14.3)	4.84 (2.01 to 11.6)	0.656 (0.132 to 3.27)
COPD	3.43 (2.98 to 3.95)	6.42 (5.78 to 7.12)	1.882 (1.578 to 2.245)	4.76 (3.16 to 7.16)	6.58 (4.48 to 9.67)	1.300 (0.731 to 2.311)
CVD	3.82 (3.42 to 4.27)	6.45 (5.93 to 7.02)	1.729 (1.505 to 1.988)	4.02 (2.74 to 5.91)	6.31 (4.59 to 8.68)	1.459 (0.876 to 2.43)
Hyperlipidemia	3.63 (3.25 to 4.06)	7.25 (6.67 to 7.88)	2.004 (1.744 to 2.303)	3.36 (2.21 to 5.11)	8.29 (6.19 to 11.1)	2.37 (1.406 to 3.995)

COPD, chronic obstructive pulmonary disease; CVD, cardiovascular diseases; DM, diabetes mellitus; HR, hazard ratio; TBI, traumatic brain injury.

Risk of Developing PD in Patients with AD

Among the 606 patients with AD, 160 cases developed PD. The relative risk of PD in patients with AD was significantly higher than the non-AD cohort (1.531, 95% CI = 1.209 to 1.939; Table 2). The Cox proportional hazard model revealed that after adjusting for the confounding variables, patients with AD had a significantly higher risk of developing PD than the non-AD group (adjusted HR = 1.667, 95% CI = 1.244 to 2.232, P = 0.0006; Table 3). The cumulative probability of PD in the Kaplan-Meier curves (Fig. F) of the AD group was significantly higher than that of the non-AD group over a 144-mo follow-up (P = 0.0004).

Subgroup Analysis on the Effect of Dementia and AD on the Risk of PD

Patients with dementia were at high risk of PD if they had underlying diabetes mellitus (adjusted HR = 1.784, 95% CI = 1.533 to 2.076), stroke (adjusted HR = 1.546, 95% CI = 1.17 to 2.043), hypertension (adjusted HR = 1.735, 95% CI = 1.582 to 1.903), traumatic brain injury (adjusted HR = 1.879, 95% CI = 1.226 to 2.88), COPD (adjusted HR = 1.882, 95% CI = 1.578 to 2.245), CVD (adjusted HR = 1.729, 95% CI = 1.505 to 1.988), or HLD (adjusted HR = 2.004, 95% CI = 1.744 to 2.303; Table 4). Likewise, patients with AD who had underlying hypertension (adjusted HR = 1.497, 95% CI = 1.096 to 2.044) or HLD (adjusted HR = 2.37, 95% CI = 1.406 to 3.995) presented with significantly high risk of PD (Table 4).

Discussion

The results of this study suggested that patients with dementia presented with an elevated risk of PD. After adjusting for confounding factors such as diabetes, CVDs, metabolic syndrome, pulmonary diseases (including pulmonary complications caused by smoking), and low oral hygiene awareness, as defined by cumulative dental visiting behavior, the association persisted. All age groups of patients with dementia had a significantly higher risk for developing PD, while PD severity increased in patients who developed dementia earlier in life, suggesting a cumulative and detrimental effect of dementia on periodontal health. Moreover, the effect of AD was more significant than that of vascular dementia, and the effect of dementia/AD on PD was even stronger in patients with concomitant HLD. Although similar associations between dementia and PD were shown in cross-sectional studies (Martande et al. 2014), the directionality was not clear due to low sample sizes and a lack of time-dependent variables. In this study, we have clearly demonstrated that dementia is associated with increased PD risk in a population that has no previous periodontal pathologies, and this effect was independent of other potential confounding factors. The incidence rates derived from longitudinal studies allow for causality due to the established temporal sequence, which presents one of the major strengths of our findings.

AD is one of the primary forms of dementia and the most studied. As we excluded PD that developed before AD onset, the findings of this cohort study strongly suggested a causal relationship between AD and increased PD risk. This effect was more pronounced among younger patients with dementia, in agreement with a previous case-control study suggesting that periodontal infections are more prevalent in early-stage AD (de Souza Rolim et al. 2014). As patients with comorbidities and risk factors for dementia (e.g., low physical activity, malnutrition, smoking and alcohol, CVDs, and HLD) are predisposed to higher risks for dementia and AD, it is likely that individuals who had these factors earlier in life were at higher risks for both geriatric syndromes, including cognitive impairment (Livingston et al. 2020), and periodontal disease. In the same sense, these factors symmetrically made younger patients with PD more susceptible to dementia (Demmer et al. 2020). This may explain the observed age-dependent effect of dementia on PD in our study, in which patients who had dementia earlier in life also presented with higher risks of PD.

The findings of the present study complemented our knowledge on the relationship between AD and PD. Although previous studies implicated PD as a risk factor or early sign of dementia or AD (Demmer et al. 2020; Nadim et al. 2020), whether the relationship was similar to that between PD and diabetes or was bidirectional was not determined. Specifically, in patients enrolled through the Atherosclerosis Risk in Communities Study, Demmer et al. (2020) reported significantly increased risk of dementia among patients with PD and patients who were edentulous; moreover, the effect of PD on dementia or mild cognitive impairment was stronger among younger patients with PD than that among patients aged >62 y. Interestingly, this finding was symmetrical to that in our study indicating that the effect of dementia on PD was stronger in younger patients. Likewise, in a meta-analysis on all types of dementia, Nadim et al. (2020) found significantly higher pooled relative risk of dementia following PD from reviewed cohort studies and case-controls studies. These results highlighted a likely association between PD and risk of AD but were limited in their assessment of the possible reverse causation that AD may also initiate PD. Juxtaposing the findings of the present study and those from previous studies on the relationship between PD and the risk of dementia/AD (Demmer et al. 2020; Nadim et al. 2020), PD and dementia/AD may possess a bidirectional relationship, in which the link was strongest among younger patients (Demmer et al. 2020).

Several models have been proposed to explain the coexistence of AD-associated neuroinflammation as a risk for other inflammatory diseases, including PD. For example, genome-wide association studies suggested that innate immune genes were the risk factors and aided in the primary role for the inflammatory elements of AD pathology via the inappropriate activation of the complement system (Harold et al. 2009). Innate immune responses and hyperactivated complement cascades are also involved in PD development (Hajishengallis et al. 2019). Deregulation of inflammatory response due to immunosenescence is also a plausible mechanism linking neuroinflammatory processes and PD by reducing the inflammatory immune response to periodontal pathogens (Goronzy and Weyand 2013; Fulop et al. 2017).

As most previous studies did not elaborate on the time effect of dementia on PD development, we included all age groups over an extensive observational period, up to 10 y. Aging served as a common denominator that drives PD and dementia, and PD increased as a factor of aging and dementia. In this study, since age was matched for each group before propensity score matching, the effect of aging has been separated from the effect of dementia. Thus, the likelihood of aging as a confounding factor for PD was minimized. Our study demonstrated that the younger the patients with dementia were, their likelihood of developing PD increased. This finding is significant and may highlight other age-related mechanisms in the dementia-induced risk of PD. Since chronic low-grade inflammation may serve as mutual pathogenesis underlying dementia (particularly AD) and PD (Bonham et al. 2018), the resolution of peripheral and central nervous system inflammation is critical (Heppner et al. 2015). As age progresses, the resolution power to end inflammation may get compromised and become a determining factor for AD and PD. The resolution of inflammation appears to be an attractive therapeutic target for AD and PD (Serhan 2014; Dunn et al. 2015; Krashia et al. 2019).

There were 2 limitations in this study. First, the overestimation of dementia might not be avoided from our longitudinal database and warrants further studies to validate whether our findings could be generalized to patients with dementia of low medical compliance or without access to health care. Second, as the LHID did not include the raw data of full mouth examination, such as periodontal indices and distribution of numbers of remaining teeth, it was impossible to elucidate the association between dementia and PD severity. However, the primary strength of the study is a large sample size. Also, propensity score matching was adopted to minimize the effect of potential confounding factors on the presented association.

To our knowledge, this is the first large-scale longitudinal study identifying the potential causal relationship between neurodegenerative diseases, including dementia, and PD. In conclusion, after adjusting for confounding factors such as CVDs, diabetes mellitus, rheumatoid arthritis, metabolic syndrome, chronic kidney disease, respiratory diseases, and oral health awareness, the associations between preexisting dementia or AD and the development of PD were significant.

Clinical Relevance and Policy Recommendations

Clinical implications of this study include the advocacy of oral health promotion and implementation of health care policies by authorities and clinicians to partner in reducing the burden of PD in patients with AD or dementia; moreover, as the observed strong effect of dementia/AD on PD was even stronger in patients with concomitant HLD, appropriate management for dementia and common comorbidities with dementia were desirable. Periodontal disease may be managed as part of geriatric syndromes or sequelae and complications following dementia.

Author Contributions

K.S. Ma, contributed to conception, design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; H. Hasturk, I. Carreras, A. Kantarci, J.C. Wei, contributed to conception, design, and data interpretation, drafted and critically revised the manuscript; A. Dedeoglu, J.J. Veeravalli, contributed to design, data acquisition, and interpretation, drafted and critically revised the manuscript; J.Y. Huang, contributed to design, data acquisition and analysis, drafted and critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.