Association between oral health and dementia in the elderly: a population-based study in Korea

We have investigated the association of oral health with development of dementia in elderly Koreans. Data for subjects aged ≥65 years who underwent regular National Health Insurance Service check-ups and received dental care in 2017 were analysed. Those treated for dementia in 2017 served as the dementia group and those with no record of dementia in 2002–2017 as the control group. Explanatory variables were sex, age, household income, place of residence, smoking status, alcohol consumption, and presence of diabetes mellitus and hypertension, periodontitis, and removable dentures. Regression analysis showed that dementia was significantly more common in women than in men and in those aged ≥81 years than in those aged 65–70 years. The risk of dementia was highest in the ‘upper-middle’ income group and in the rural population. Smokers and those who consumed alcohol were less likely to develop dementia. Subjects with diabetes were more likely to have dementia than those without it, as were those with hypertension. Dementia was less likely in subjects with periodontitis and more likely in those with removable dentures. Therefore, loss of teeth may contribute to development of dementia.

. Association between denture wear and dementia by sex and age. a Model 1 was unadjusted; b model 2 was adjusted for age, household income, and residence; c model 3 was adjusted for age, household income, residence, smoking, alcohol consumption, diabetes mellitus, and hypertension. Response variable: dementia (ref. no). *P < 0.05, **P < 0.01, ***P < 0.001. CI, confidence interval; OR, odds ratio.
response variable, i.e., the prevalence of dementia. All explanatory variables were found to have a statistically significant effect on the likelihood of developing dementia (P < 0.001) and were therefore included in the regression model. The fitted regression model was statistically significant (P < 0.001) and the Nagelkerke R 2 value was 0.109. Dementia was more common in women than in men (OR 1.43; P < 0.001) and more common in those aged ≥81 years than in those aged 65-70 years (OR 8.00; P < 0.001). Subjects with an 'upper-middle' household income were more likely to have dementia (OR 1.09; P < 0.001), as were residents of rural areas when compared with those from urban areas (OR 1.35; P < 0.001). Dementia was less common in smokers than in non-smokers (OR 0.92; P < 0.001) and in those who consumed alcohol than in those who did not (OR 0.63; P < 0.001). Subjects with diabetes were more likely to have dementia than those who did not (OR 1.46; P < 0.001), as were those with hypertension (OR 1.17; P < 0.001). Dementia was less common in subjects with periodontitis (OR 0.80; P < 0.001) but was more common in those with removable dentures (OR 1.26; P < 0.001).

Discussion
Dementia is a disease that progresses slowly over a period of 5-20 years after onset, and prevention is most important 13 . Health determinants that act as risk factors for dementia, including sociodemographic characteristics, health behaviours, and systemic diseases, should be recognized in an effort to prevent the disease 6 . This study investigated a large number of variables, including sex, age, household income, and place of residence as sociodemographic characteristics, smoking and drinking as health behaviours, and diabetes and hypertension as systemic diseases.
Advanced age is the strongest risk factor for dementia. In a previous study 14 , the prevalence of dementia at the age of 65-69 years was 1.3%, with a two-fold increase every 5 years; thereafter, the prevalence in the population aged ≥85 years has been reported to be 30.5% 14 . In the present study, the prevalence of dementia was 14.3% and 21.4% for men and women in the group aged ≥81 years and 1.8% and 2.7% in the group aged 65-70 years, which is consistent with a previous report 14 . Moreover, in a study reported in 2018, the risk of dementia was significantly higher in women than in men, in smokers than in non-smokers, and in those who consumed alcohol than in those who did not 15 . Furthermore, Kim 16 reported that female sex, age >70 years, a low educational level, and alcohol abuse were risk factors for cognitive impairment. The results of multiple logistic regression analysis in the present study also showed that women were more likely to develop dementia than men (OR 1.43; P < 0.001) while subjects aged ≥81 years were more likely to have dementia than those aged 65-70 years (OR 8.00; P < 0.001). However, in this study, smokers were less likely to have dementia than non-smokers and the same was true for alcohol consumption. These findings may reflect the fact that subjects who had drinking or smoking habits may have quit them after the onset of dementia either on their own or under the care of nursing homes and caregivers. Therefore, it was difficult to determine the effects of smoking and alcohol consumption on dementia in this study.
Those with an 'upper-middle' household income were more likely to develop dementia (OR 1.09; P < 0.001) in this study. This finding is different from that in the study by Ribeiro et al., who showed that patients with AD had a lower average monthly income 17 , and from that in a study of urban-dwelling African-Americans that  www.nature.com/scientificreports www.nature.com/scientificreports/ investigated factors influencing cognitive function, which showed that a low income level was associated with poor cognitive function 18 . Further studies are needed to verify these findings.
Rural residents were more likely to have dementia than their urban counterparts (OR 1.35; P < 0.001). This finding may reflect the effects of social isolation on development of dementia found by Lee 19 in a study that investigated dementia and social isolation among the rural elderly in Korea. Hence, elderly individuals residing in rural areas may have an increased risk of dementia because of their limited social networks when compared with elderly residents of urban areas.
Hypertension and diabetes mellitus are known to be risk factors for dementia 20,21 , with an increase in the prevalence of the disease as age increases. Being elderly is an uncontrollable risk factor for dementia; however, hypertension and diabetes mellitus are controllable risk factors 22 . In this study, patients with diabetes mellitus were also more likely to have dementia than healthy elderly subjects (OR 1.46; P < 0.001) and the same was true for patients with hypertension (OR 1.17; P < 0.001). Therefore, continuous health care before the onset of old age to prevent chronic illnesses such as hypertension and diabetes mellitus may help to lower the risk of dementia.
Periodontitis is a chronic inflammatory disease that has been identified as a potential risk factor for a variety of systemic disorders, including cardiovascular disease, diabetes mellitus, hypertension, respiratory disease, osteoporosis, metabolic syndrome, and rheumatoid arthritis 23,24 . An association between periodontitis and dementia has also been suggested [25][26][27][28] . The direct and indirect causes of dementia are thought to involve inflammatory reactions, damage caused by oxygen-derived free radicals that damage cells, and toxic substances; periodontitis, a chronic inflammatory disease, is thought to impair cognition via this mechanism 25,26 . Periodontitis has been reported to be a risk factor for the onset and progression of AD 27 and has been suggested to exacerbate inflammatory conditions in the elderly, thereby accelerating the progression of neurodegenerative diseases 28 . Furthermore, a longitudinal study investigating the risk of dementia following chronic periodontitis in Koreans by Cox proportional hazards regression recently suggested that chronic periodontitis could increase the risk of dementia 29 . However, in the present study, subjects with periodontitis were less likely to have dementia (OR 0.80; P < 0.001), and the prevalence of dementia was lower in subjects with periodontitis even when analysed by sex and age. We expected that the elderly with dementia would have more tooth loss than their counterparts without dementia, as previously reported [30][31][32] , and in this study considered that tooth loss would reduce the need for treatment of periodontitis and decrease its prevalence. Using the NHIS data, we determined the presence of chronic disease by using a diagnostic code for medical and dental care 33 , but the true prevalence of periodontal disease might not have been reflected if individuals with a given disease stopped treatment. Therefore, there is a possibility of underestimation of prevalence. However, it is also possible that the prevalence of periodontitis is low in patients with dementia. This possibility needs investigation in the future.
Subjects using a removable partial denture or full dentures are very likely to have lost a high number of teeth. Our results showed that patients with removable dentures were more likely to have dementia (OR 1.26; P < 0.001) regardless of sex and age. Syrjälä et al. 30 reported that subjects with dementia are likely to be edentulous (OR 5.2; 95% CI 1.0-26.6). Moreover, Cho 31 reported that for persons with 0-10 residual teeth, the OR for probability of dementia was 3.527 (95% CI 1.382-8.997) based on the Mini-Mental State Examination score in individuals with a residual tooth number of ≥21. Other studies have also found that individuals with cognitive impairment have fewer remaining teeth than those without cognitive impairment 32 . However, in a study examining the relationship between dental health and dementia in the elderly using the Korean Dementia Screening Questionnaire 3,34 , neither the number of functional teeth, including implants and fixed prostheses, nor use of dentures had a significant effect on the risk of dementia, unlike in previous studies.
This study has several limitations. It is difficult to infer a causal relationship between dementia and the risk factors for dementia using a cross-sectional study design. Furthermore, the NHIS data do not reveal the number of teeth lost, so there is a limit to estimation of tooth loss when using data for subjects treated with removable dentures. However, despite these limitations, it was possible to control recall bias, which is common in questionnaire-based research, because of use of NHIS data, which is meaningful for confirmation of the effect of dental health on dementia in the elderly. Therefore, we are planning a longitudinal cohort study to investigate the causal relationship between dementia and oral health. In the meantime, we hope that our findings will contribute to the development of health promotion programs and effective strategies for prevention of dementia.

participants. The study protocol was approved by the Institutional Review Board of Shinhan University
(SHIRB-201806-HR-082-01) and conducted following the guidelines of the Declaration of Helsinki. The requirement for written informed consent was waived because the study used the NHIS database, which contains only anonymised secondary data.
Almost all Korean citizens (approximately 98%) are covered by Korean national health insurance, and their sociodemographic and medical records, prescription medication history, and history of medical and dental health screening are stored in the NHIS database. This database contains data for patients aged ≥65 years who received dental treatment in 2017, who was an eligibility criterion for this study. Patients who also underwent regular health check-ups performed by the NHIS were selected as the final subjects for this study (Fig. 1). www.nature.com/scientificreports www.nature.com/scientificreports/ The explanatory variables used in this study were sex, age, household income, place of residence, smoking and drinking status, presence of diabetes mellitus and hypertension, periodontitis, and wearing of removable dentures. Age was classified as 65-70, 71-75, 76-80, and ≥81 years. Household income was classified into quintiles (low, lower-middle, middle, upper-middle, high) based on the cost of the health insurance paid, and place of residence was classified as urban or rural. Furthermore, current smoking status and drinking status were examined based on the responses to the NHIS health screening questionnaire. Ex-smokers were classified as non-smokers, and current drinking status was judged based on drinking alcohol more than once a week.

Study variables.
Diabetes, hypertension, periodontitis, and wearing a denture were assessed according to the medical records of dental and medical institutions. First, based on the ICD-10, we looked at the medical records of the subjects to see if there were any records corresponding to the diagnostic code for diabetes mellitus (E10-E14) or hypertension (I10-I15). The definition of periodontitis was modified for the purposes of this study with reference to a previous definition 35 . Patients who had had periodontal treatment (U1010; subgingival curettage, U1051-1052; periodontal flap operation, U1060; root conditioning, U1071-1072; bone graft for alveolar bone defects, U1081-1083; guided tissue regeneration) with a diagnostic code for periodontitis (K05.2; acute periodontitis, K05.3; chronic periodontitis, K05.4; other periodontal disease, K05.6; periodontal disease unspecified) were defined as patients with periodontitis. Furthermore, subjects were deemed to use a removable denture if there were any medical records including the treatment code related to the manufacturing process of a removable partial denture (UA301-UA359, UA401-UA419), the treatment code related to the manufacturing process for a full denture (UA101-UA149, UA201-UA209, UA501-UA549), and the treatment code related to the denture repair (U1511-U1542).
Statistical analysis. The differences in prevalence of dementia according to subject characteristics were analysed using the chi-square test. We also divided the study participants into 8 groups according to sex and age and analysed the prevalence of dementia according to the oral health of each group. Furthermore, multiple regression analyses were conducted to determine the associations of periodontitis and denture wear with dementia in each group, with sequential inclusion of confounding factors in the models 1-3. Finally, multiple logistic regression analysis was performed with all confounding factors using periodontitis and denture wear as explanatory variables and dementia as the response variable. All statistical analyses were performed using SAS software (version 9.4; SAS Institute Inc., Cary, NC, USA). A P-value of 0.05 was considered statistically significant.

Data Availability
The data that support the findings of this study are available from the National Health Insurance Service-National Sample Cohort but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available.