ABBREVIATIONS

CIND, Cognitive Impairment, Not Demented

HRS, Health and Retirement Study

ISR, Institute for Social Research

OH, Oral status or condition of the mouth as measured by self-report

OHRQoL, Oral Health Related Quality of Life

TICS-m, modified Telephone Interview for Cognitive Status

INTRODUCTION

With life expectancy nearly doubled during the 20th century, there has been a ten-fold increase in the number of middle to older aged Americans [1,2]. An estimated 50 million older individuals have dementia, however, dementia it is not a normal part of ageing [3,4]. Recent studies associate poor oral health and periodontitis with greater cognitive decline and worse cognitive outcomes [5–13].

Dental pain and tooth loss may affect nutrition, functional statuses, self-confidence, precipitate stress and ultimately affect an individual’s overall quality of life [5–7]. Previous studies have examined the relationship between cognition and oral health and oral health related quality of life (OHRQoL), however these studies predominantly do so with cognition as the predictor rather than the outcome [5,8–11]. Other studies examining poor oral health as an outcome, primarily focus on the physical aspect of oral function and have evaluated the association of cognitive function with tooth loss, edentulism and denture use [8,12,13].

The aim of the current study is to examine the association between status or condition of the mouth as measured by self-report (OH) and OHRQoL on cognitive function in a nationally representative population based sample of the Health and Retirement Study (HRS) in the United States of America (USA).

METHODS

The present study utilized publicly available data from the USA HRS, a nationally representative longitudinal study of community dwelling adults. The HRS data is collected by the Institute for Social Research (ISR) at the University of Michigan (grant number NIA U01AG009740) and sponsored by the National Institute on Aging in USA who administer a combination of longitudinal household surveys, and conduct in-person and telephone based interviews. A detailed description of HRS has been published elsewhere [14–16].

The present study is a cross-sectional analysis of participants from the HRS 2018 Core data. The ‘Dental Health and Dental Health Services module’ was administered as an experimental module by the HRS to a randomly-selected sub-sample of HRS self-respondents, after the main (core) interview was completed [17]. Individuals who participated in the Dental Health and Dental Health Services module between April 2018 and June 2019 and responded to at least one question each from the sections assessing OH, OHRQoL, and Dental Utilization, were included in the study. The final study sample consisted of 1268 participants, after excluding those who were below 50 years of age or had missing cognitive scores on immediate and delayed recall, backward counting and serial seven subtraction tests.

Institutional Review Board (IRB) at the University of Michigan (UM) approved the study protocol (UM Health Sciences/Behavioral Sciences IRB Protocol: HUM00061128 Approved through 10/18/2018 Associated protocols: HUM00056464, HUM00002562, HUM00074501, HUM00079949, HUM00080925, HUM00085942, HUM00099822, HUM00103072, HUM00106904, HUM00122335, REP00000046).

Self-reported Oral Health (OH) to assess the condition of the mouth was measured by three indicators: self-reported condition of mouth and teeth, number of teeth and use of dentures. Participants were asked to report if the condition of their mouth and teeth was “excellent”, “very good”, “good”, “fair”, or “poor”. A binary variable was created grouping (i) ‘excellent, very good, good’ and (ii)’fair/poor’. The number of teeth lost were reported as (i) ‘lost more than four teeth’, (ii) lost less than four teeth/all teeth present”; Denture use was recorded as “yes” or “no”.

Self- reported Oral health related quality of life (OHRQoL) constituted five questions from the “Section on Quality of Life” from the HRS dental module (2018) [21]. These questions pertain to the participant experiencing any of the following challenges/conditions due to problems with their teeth, mouth, or dentures. These included (i) avoidance of food, (ii) going out, (iii) difficulty to relax, (iv) being nervous or self-consciousness and (iv) experiencing pain or distress due to the condition of their teeth, mouth, or dentures. Data was collected as an ordinal variable scored 1–5 (never, hardly ever, occasionally, fairly often, very often) across five items related to avoidance of food, going out, difficulty to relax, nervous or self-consciousness or pain and distress due to the condition of their teeth, mouth, or dentures. A summary (composite) score was calculated by adding the item response to the five OHRQoL questions.

Two self-reported questions from the “Section on Dental Utilization” of the HRS dental module (2018) were evaluated. These included the questions “wanted to see a dentist but didn't see the dentist” and dental insurance coverage; the responses were recorded as “Yes and No” responses.

Cognitive assessment was conducted using the modified Telephone Interview for Cognitive Status (TICS-m) [18] The TICS-m comprises of three subscales that capture cognitive function in the domains of episodic memory, attention, and working memory: (i) immediate and delayed recall of 10 words from a word list randomly assigned for each participant (0–20 points), (ii) backward counting (0–2 points), and (iii) serial seven subtraction (0–5 points) [19]. The sum of scores from these subscales was calculated and is referred to as the ‘global cognitive score’. The global cognitive score ranges from 0 to 27, with higher scores indicating better overall cognitive function [20].

Next, using the global cognitive function score three validated cognitive score categories were created: (i) (≥12) representing normal cognitive functioning (ii) (6–11) “cognitive impairment, not demented” (CIND), and (iii) (<6) dementia. These cut-points have been previously validated in the Aging, Demographics, and Memory Study, a national population-based study of dementia whose participants are a sub-population of the HRS [20].

Descriptive statistical tests were used to compare the participant characteristics across cognitive score categories. Multivariable linear regression models were used to investigate associations between global cognitive scores (dependent variable), and OHRQoL score (independent variable). Lastly, a series of multinomial logistic regression were employed to evaluate the association of cognitive outcomes (cognitive score <6 as reference category) with OH, and Dental Utilization.

We considered potential co-variates including age, sex, education and current smoking, ever alcohol use and current diabetes mellitus based on previous literature [22–30]. We evaluated these variables using a stepwise forward selection process using the ‘GLMSELECT’ function in SAS software and appropriately adjusted our statistical models for the same. All analysis were conducted using a 0.05 level of significance as the reference.

Analyses were performed using SAS version 9.4 (SAS Institute Inc. Cary, NC, USA).

RESULTS

The descriptive characteristics of the study participants (N = 1268) by cognitive score categories (Normal, CIND and dementia) are shown in Table 1. In general, the individuals in the dementia category were older, consumed alcohol, were Caucasian and fewer participants in this category had attained higher education. A greater percentage of individuals with dementia had fair/poor self-reported OH. A greater percentage of individuals with CIND wore dentures. With respect to the individual OHRQoL items, a greater percentage of participants with dementia avoided eating certain foods, had difficulty in relaxing and were nervous or self-conscious due to the conditions of their mouth, teeth or gums.

TABLE 1

Table 1. Characteristics of study sample by Cognitive Score Categories.

Table 2. shows the results of the multivariate linear regression model testing the association between self-reported OHRQoL and Global cognitive scores. The model was adjusted for sex, age, race, low education attainment (less than high school completed), current smoking, ever alcohol use and current diabetes mellitus. The adjusted model shows that individuals with worse self-reported OHRQoL had 0.15 units lower global cognitive scores (β = −0.15, SE = 0.03, p ≤ 0.0001).

TABLE 2

Table 2. Adjusted Multivariate linear regression model and 95% confidence intervals of the association between self-reported OHRQoL and global cognitive scores.

Lastly, in order to provide results in the form of commonly employed clinically relevant cognitive status groupings, we used a series of multinomial logistic regression analyses to evaluate the association between cognitive outcomes defined above (normal cognition, CIND and dementia) with OH and Dental Utilization. Global cognitive score <6 representing Dementia was taken as the reference group. Compared to participants with dementia, those in both Normal and CIND groups had greater odds of having better self-reported OH conditions (OR_Normal = 1.76, (95%CI: 1.09–2.84); OR_CIND = 2.86, (95%CI: 1.60–5.09)). Neither participants in the Normal nor CIND were associated with having ≥ 4 teeth, however, participants in the CIND group had greater odds of using dentures (OR_CIND = 2.71, (95%CI: 1.49–4.93)) compared to those with dementia. With regard to dental utilization, individuals with normal cognition had slightly lower odds of wanting to go to the dentist but not going for a dental visit, when compared to those in the dementia category (OR_Normal = 0.98, (95% CI: 0.94–0.999)), however, these findings did not reach statistical significance. No association was found between cognitive score categories and dental insurance coverage.

DISCUSSION

The present study assessed the association between self-reported OH, OHRQoL, dental utilization and cognitive function in a nationally representative sample of adults over the age of 50 years. This study adds to the oral health and cognitive function literature by investigating the role of OH and OHRQoL as risk factors of cognitive function.

A number of studies have focused on the ‘oral function’ aspect of oral health and have evaluated the association between cognitive function with surrogates of oral conditions by measuring tooth loss, edentulism and denture use [5,8,12,13]. Ge et al. reported that worse oral health resulted in faster decline of cognitive functioning in racially diverse populations [17]. Consistent with previous literature our study showed that participants in the Normal and CIND categories had better self-reported OH compared to the participants in the dementia category. A meta-analysis focusing on tooth loss, diminished cognitive function and risk of dementia showed that that individuals with fewer teeth were at a 20% higher risk for developing cognitive decline [8]. Paradoxically, the present study did not find any association between tooth loss and dementia (Table 3). Though further validation is needed, a potential explanation of these results could be that individuals with dementia may indeed have worse conditions of mouth and teeth, which is consistent with our findings, but they may not be able to get the periodontally compromised or decayed teeth extracted due to lack of patient cooperation or due to extensive inflammation and infection.

TABLE 3

Table 3. Multinomial logistic regression between self-reported OH, Dental Utilization and cognitive categories.

Two investigators have utilized HRS data from previous waves (2008) to study the associations between edentulism, and cognitive functioning [7,12]; The study by Han et al. found that edentulism and dental care service utilization were independently associated with cognitive decline [12]. Similarly, we found that cognitively normal participants were not associated with denture use while those with CIND were more likely to use dentures than participants with dementia, and these findings are consistent with previous literature [14,16]. Based on our findings, we think that distinguishing between participants with CIND and dementia is key from a clinical perspective. A potential explanation for lower denture use in participants with dementia might be that fabrication of dentures requires multiple long dental appointments and losing dentures is commonly seen in individuals with dementia. Future investigation tracking denture use followed by clinical dental examination would be required to validate these findings.

OHRQoL includes a subjective evaluation of the individual’s oral health, functional status, emotional well-being, and a sense of self [31]. While most investigators focus predominantly on oral health status, the assessment of the latter components may help gain valuable insights into an individual’s behavioral, emotional and psychosocial functioning that may be impacted by dementia [32]. Fewer studies have evaluated the association between OHRQoL and cognition. A study by Wong et al. reviewed twenty-five surveys (or study series) from 19 countries and found that among institutionalized older adults, those with worse cognition (mild or above cognitive impairment)had poor OH and OHRQoL [14]. Another study by Jing et al. found a positive reciprocal association between cognition and OH and OHRQoL [33]. Consistent with these studies, we found that individuals with worse self-reported composite OHRQoL score had 0.15 units lower global cognitive scores.

The strengths of this study include representation of community dwelling middle to older aged adults, utilization of items from validated cognitive and OHRQoL measures [34–36]. Furthermore, previous studies show that the TICS-m may relate to more common cognitive scales such as the Montreal Cognitive Assessment (MoCA) or the Mini-Mental State Examination (MMSE) [37,38]. Although these findings are valuable, our study has some limitations. Some of the limitations of our study include small sample size, the use of dental parameters collected from a sub-sample of the HRS participants, and self-report of dental conditions. Although HRS randomly selected participants who would be administered the dental module, it is possible that severely demented individuals may not respond to paper-based surveys, thus underestimating the effect of the intervention. Additionally, clinical oral examination and self-reported OH may not always be concordant, especially in individuals with CIND or dementia and thus, further investigation verifying this concordance is needed. Furthermore, although previously validated, relying on population-based cut-offs to create cognitive score categories (normal, CIND, dementia) may introduce systematic errors. From a life-course perspective, a number of factors such as low education, cardiovascular disease, smoking, diabetes may influence the potential association between poor oral conditions and cognitive functioning. To some extent, we have accounted for the potential confounding factors in the analysis, however, effects of residual confounding may remain. Furthermore, due to the cross-sectional nature of our study the exposure and outcome have been simultaneously assessed, and thus further study would be needed to assess the potential temporal relationship between OH, OHRQoL and cognitive function.

Despite these limitations, our findings provide valuable information associating poor OH and lower OHRQoL with worse cognitive function. If validated, through longitudinal studies and clinical dental examination, these methods of employing self-reported OH and OHRQoL as well as the use of modified Telephone Interview for Cognitive Status might be practical screening modalities for middle to older aged individuals who may be home-bound, medically compromised or may have limited access to care.

AUTHOR CONTRIBUTIONS

Prajakta Joshi designed the study and analyzed the data. Howard Cabral helped design the statistical analysis section, interpreting and writing the results, and discussion sections. Prajakta Joshi and Laura B. Kaufman wrote the paper with input from all authors.

CONFLICTS OF INTEREST

The authors declare that they have no conflicts of interest.

FUNDING

This work was in part supported by the Boston University Clinical and Translational Science Award (NIH grant number: 1UL1-TR001430).

Data used in preparation of this article were obtained from the Health and Retirement Study. The data is generated and distributed by the University of Michigan with funding from the National Institute on Aging (grant number NIA U01AG009740, U01 MH064887 Treatment of Early Age Mania (TEAM)) Ann Arbor, MI, 2018.