Abstract
Background
Several mechanisms may associate tooth loss and related oral inflammation with cognitive impairment. The authors studied the relationship between tooth loss and cognitive function.
Methods
The REasons for Geographic And Racial Differences in Stroke study is a national longitudinal study of more than 30,000 African American and white adults 45 years or older. Data for tooth loss, cognitive function and potential confounding variables were available for 9,853 participants at the time of analysis. The authors used incremental linear regression modeling to investigate the cross-sectional association between self-reported tooth loss and cognitive function.
Results
In unadjusted analysis (mean learning followed by recall; α level of significance of .05), the loss of six to 16 teeth and the loss of more than 16 teeth were associated with poorer cognitive function compared with the loss of no teeth. Attenuated associations persisted after the authors adjusted for demographic and systemic risk factors. The full model, which was adjusted for socioeconomic status (SES), revealed no association between tooth loss and cognitive function.
Conclusion
Tooth loss may be associated with cognitive function; however, this association is mediated by age and SES.
Clinical Implications
Tooth loss due to periodontal disease may be a marker for low SES, and the interplay of these factors with advanced age may confer risk of having poorer cognitive function. Further studies are needed to clarify these associations.
Keywords: Cognitive function, tooth loss
Investigators in some studies have hypothesized or demonstrated an association between cognitive impairment and poor oral health, with higher levels of cognitive impairment in populations with worse oral health.1–11 The investigators have suggested that cognitive impairment causes poor oral health, possibly owing to a lack of interest in or forgetting about oral hygiene. Results from a prospective study demonstrated an increased risk of experiencing cognitive decline with increased tooth loss, caries incidence and periodontal disease progression.12 However, this relationship has been inconsistent across studies.13–15 Poor oral health is associated with many diseases such as diabetes, cardiovascular disease, stroke and Alzheimer disease.9 Tooth loss and associated factors due to periodontitis might be associated with cognitive impairment through at least three other potential, but not equally plausible, pathways.
The oral cavity is colonized by numerous species of bacteria. Evidence suggests that bacterial products16 or whole bacteria can colonize the brain directly via the bloodstream17 or the trigeminal nerve18 and cause cognitive impairment.18–22 The results of some studies refute these ideas.23–27
The primary culprits in the pathogenesis of periodontitis are gram-negative oral bacteria.9 Their presence can stimulate the localized production of inflammatory cytokines28–35 and systemic production of C-reactive protein (CRP),33,34,36 all of which may injure and impair the brain directly.37,38 Conversely, this disease process may be a local phenomenon in the brain.
Ischemic stroke and cerebrovascular health are risk factors for dementia39 and cognitive impairment because they diminish short-term memory.40 Severe periodontitis, edentulism and partial edentulism have been associated with increased risk of ischemic stroke.41–45
Memory and learning impairment are aspects of declining cognitive function.9 We studied the association between tooth loss and cognitive function as measured by means of the word list learning (WLL) assessment in a cross-sectional analysis of data from the REasons for Geographic And Racial Differences in Stroke (REGARDS) study. We hypothesized that people with more teeth would have better cognitive function and that this association would persist with incremental linear regression modeling, after adjusting for demographics. We also theorized that covariate adjustment for risk factors would mediate this association.
PARTICIPANTS, MATERIALS AND METHODS
The REGARDS study is a national longitudinal study of African American and white adults 45 years or older that has the primary objective of determining the causes for excess stroke mortality among blacks and residents of the “Stroke Belt.” The Stroke Belt is defined as the following eight states with high stroke mortality: Alabama, Arkansas, Georgia, Louisiana, Mississippi, North Carolina, South Carolina and Tennessee.46 Study enrollment ended in 2007, and the follow-up is ongoing.
Participants
From 2003 through 2007, we used commercially available lists of U.S. residents to select participants randomly, and we recruited them by mail and telephone. The resulting cohort consisted of 30,239 people of whom 42 percent were black and 58 percent were white. Fifty-six percent of the participants lived in the Stroke Belt and 44 percent lived in the remaining 40 contiguous states (non–Stroke Belt states) and the District of Columbia. People who lived in Alaska and Hawaii were not included because these states were “disconnected” from the contiguous states for which geographic patterns initially were described, and the number of blacks living in those states was so small that recruitment would have been a challenge. Staff members of the Survey Research Unit of the University of Alabama at Birmingham conducted a computer-assisted telephone interview to collect demographic data, administer medical history questionnaires and complete cognitive screenings. They obtained written informed consent from the participants during an in-home visit during which they collected blood and urine samples and obtained electrocardiographic, blood pressure and body mass index (BMI) measures. We contact participants by telephone every six months to monitor for stroke and coronary events, annually for cognitive screening status and biennially for a more extensive cognitive assessment. The institutional review boards of all participating institutions approved the REGARDS study.
We introduced the cognitive assessment outcome used in this study to the follow-up telephone calls made in February 2006. As of June 2008, data for 12,661 participants were available for analysis. We excluded participants if they did not have values for income, CRP, tooth loss, hyperlipidemia, BMI, depressive symptoms, diabetes, hypertension, smoking or education (these variables are defined below). A total of 2,808 (22.2 percent) participants were missing one or more of these values (primarily income), which left 9,853 participants for the final analyses.
Outcome measure (cognitive function—learning and recall)
The WLL assessment from the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) battery47 involves a standardized protocol of presentation, learning and recall to obtain scores for two primary domains: sum learning (attention and learning) and delay recall (memory). We administered the WLL assessment via telephone by playing a recording of the standardized CERAD instructions and the list of 10 words presented in the standardized CERAD order, which differs across the three learning trials (the three opportunities for participants to hear and repeat back the 10 words). To avoid problems that could be introduced by variations in regional accents, a professional orator recorded the instructions and word list presentations. As we assessed all of the participants in our study by telephone, we were not able to show participants the word lists. We instructed the participants at the beginning of each trial not to write anything down.
For each of the three learning trials, we asked the participants to repeat as many of the 10 words as possible. We calculated the sum learning score by totaling the number of correctly recalled responses across the three trials; the scores ranged from zero to 30. After an interval of approximately five minutes, during which we asked noncognitive interview questions, we obtained the delay recall score by asking the participant to recall as many of the 10 words as possible; the scores ranged from zero to 10. Finally, we calculated a mean learning score by averaging the scores for the three learning trials; the scores ranged from zero to 10. This mean learning score retained the properties of the sum learning score and had identical associations with both tooth loss and confounding factors. We calculated the mean learning score to facilitate comparisons to the delay recall score, which has the same range.
Primary exposure variable (tooth loss)
We asked the participants the following questions to assess tooth loss: “Have you lost any of your teeth owing to gum disease?” and “How many teeth have you lost owing to gum disease?” Although neither question has been validated, responses to the following questions have shown high specificity when compared with a radiographic verification of periodontal disease in a German study: “Have you ever had a tooth extracted due to bone loss?” and “Have you ever had a tooth extracted because it was loose?”48 We think that both of our survey questions encompass these two domains of periodontal disease. We related our latter survey question to systemic inflammation in another REGARDS analysis.49 We included third molars reported as lost owing to periodontal disease. We categorized the answers as none, one to five, six to 16 and more than 16 teeth lost.
Potential confounding factors
Demographics
We categorized self-reported residence as the Stroke Belt or non–Stroke Belt states. We categorized self-reported race as black or white and self-reported sex as male or female. We calculated age from date of birth and coded it into one of five categories: 45 to 54, 55 to 64, 65 to 74, 75 to 84 and 85 years or older.
Risk factors
We calculated BMI as kilograms per square meter and categorized it as normal (18.5–24.9), underweight (< 18.5), overweight (25–29.9) or obese (> 30). We took the logarithm of CRP (logCRP) to normalize the distribution of data points and analyzed it as a continuous variable, logCRP. We defined having diabetes as a participant’s having a fasting glucose level of more than 126 milligrams per deciliter, a nonfasting glucose level of more than 200 mg/dL, or taking medicine or insulin for diabetes. We defined having hypertension as a participant’s having systolic blood pressure of 140 millimeters of mercury or greater, diastolic blood pressure of 90 mm Hg or greater or self-reported hypertension confirmed by means of medication use. We defined having hyperlipidemia as a participant’s having a total cholesterol level of 240 mg/dL or greater, a low-density lipoprotein cholesterol level of 160 mg/dL or greater, a high-density lipoprotein cholesterol level of 40 mg/dL or lower, or current use of lipid-lowering medication. We categorized smoking status as never, past smoker and current smoker.
Depressive symptoms
We used the four-item version of the Center for Epidemiological Studies-Depression scale to evaluate depressive symptoms. Scores range from zero to 12; a score of four or greater indicates an elevated level of psychological distress.50
Socioeconomic status (SES)
We categorized income as less than $20,000, $20,000 to $34,999, $35,000 to $74,999 and $75,000 or higher per year. We categorized education as no high school diploma, high school diploma or general equivalency diploma, some college but no degree, and college degree or higher.
Statistical analysis
We assessed the associations between the exposure variable (tooth loss) and outcomes (mean WLL and delay recall scores) in incremental series of linear regression models by first examining the univariate or unadjusted association and then adjusting for risk factors in steps to see if the association might have been due to other factors. This approach has been called a “mediation analysis” and has been championed by Mackinnon and colleagues.51 Mediation analysis approaches focus on the change in the magnitude of the coefficient associated with tooth loss on the addition of these demographic or risk factors to the model. If the demographic or risk factors are responsible for the observed univariate relationship, then the magnitude of the coefficient for tooth loss will be attenuated completely to 1.0 with the addition of these mediating factors. Given the strong association between SES and tooth loss,49 we thought that SES would represent the greatest potential for confounding, so we added it last. We used the Student-Newman-Keuls multiple range test in a univariate analysis of the following variables: logCRP, Center for Epidemiological Studies-Depression scale score, mean learning score and delay recall score. We performed post hoc analyses to assess potential age-by-tooth-loss and smoking-by-tooth-loss interactions. We performed all analyses with software (SAS for Windows, Version 9.1) by using two-tailed P values and an α level of significance of .05.
RESULTS
Among the 9,853 participants, 8,546 participants (86.7 percent) reported having lost no teeth, 424 (4.3 percent) reported having lost one to five teeth, 314 (3.2 percent) reported having lost six to 16 teeth, and 569 (5.8 percent) reported having lost more than 16 teeth owing to periodontal disease. Table 1 shows the distribution of teeth lost according to the participants’ characteristics. We noted significant differences for all variables except for region.
TABLE 1.
Participant characteristics, according to number of teeth lost.
| CHARACTERISTIC | NUMBER OF TEETH LOST*
|
P VALUE† | |||
|---|---|---|---|---|---|
| 0 | 1–5 | 6–16 | > 16 | ||
|
| |||||
| Total (No. [%]) | 8,546 (86.7) | 424 (4.3) | 314 (3.2) | 569 (5.8) | |
|
| |||||
| Region | .212 | ||||
| Non–Stroke Belt | 3,932 (86.6) | 208 (4.6) | 153 (3.4) | 245 (5.4) | |
| Stroke Belt | 4,614 (86.8) | 216 (4.1) | 161 (3.0) | 324 (6.1) | |
|
| |||||
| Race (No. [%]) | < .001 | ||||
| White | 5,603 (89.9) | 228 (3.7) | 134 (2.2) | 266 (4.3) | |
| Black | 2,943 (81.3) | 196 (5.4) | 180 (5.0) | 303 (8.4) | |
|
| |||||
| Sex (No. [%]) | < .001 | ||||
| Male | 4,046 (88.1) | 192 (4.2) | 135 (2.9) | 219 (4.8) | |
| Female | 4,500 (85.5) | 232 (4.4) | 179 (3.4) | 350 (6.7) | |
|
| |||||
| Age (Years) (No. [%]) | < .001 | ||||
| 45–54 | 867 (90.9) | 41 (4.3) | 28 (2.9) | 18 (1.9) | |
| 55–64 | 3,654 (87.4) | 194 (4.6) | 141 (3.4) | 192 (4.6) | |
| 65–74 | 2,830 (85.4) | 124 (3.7) | 108 (3.3) | 250 (7.5) | |
| 75–84 | 1,092 (85.0) | 63 (4.9) | 34 (2.6) | 95 (7.4) | |
| 85 or older | 103 (84.4) | 2 (1.6) | 3 (2.5) | 14 (11.5) | |
|
| |||||
| BMI‡ (No. [%]) | .024 | ||||
| Normal | 2,041 (88.9) | 87 (3.8) | 62 (2.7) | 107 (4.7) | |
| Underweight | 75 (84.3) | 5 (5.6) | 2 (2.2) | 7 (7.9) | |
| Overweight | 3,238 (86.7) | 167 (4.5) | 126 (3.4) | 203 (5.4) | |
| Obese | 3,192 (85.5) | 165 (4.4) | 124 (3.3) | 252 (6.8) | |
|
| |||||
| Income ($) (No. [%]) | < .001 | ||||
| < 20,000 | 1,402 (79.3) | 77 (4.4) | 82 (4.6) | 207 (11.7) | |
| 20,000–34,999 | 2,282 (84.9) | 113 (4.2) | 96 (3.6) | 197 (7.3) | |
| 35,000–74,999 | 3,164 (88.1) | 172 (4.8) | 110 (3.1) | 144 (4.0) | |
| 75,000 or higher | 1,698 (94.0) | 62 (3.4) | 26 (1.4) | 21 (1.2) | |
|
| |||||
| Education Level (No. [%]) | < .001 | ||||
| No high school diploma | 770 (80.8) | 33 (3.5) | 45 (4.7) | 105 (11.0) | |
| High school diploma or general equivalency degree | 2,117 (83.8) | 97 (3.8) | 92 (3.6) | 220 (8.7) | |
| Some college but no degree | 2,342 (86.8) | 127 (4.7) | 88 (3.3) | 141 (5.2) | |
| College degree or higher | 3,317 (90.2) | 167 (4.5) | 89 (2.4) | 103 (2.8) | |
|
| |||||
| Diabetes (No. [%]) | < .001 | ||||
| No | 7,013 (88.1) | 331 (4.2) | 234 (2.9) | 386 (4.8) | |
| Yes | 1,533 (81.2) | 93 (4.9) | 80 (4.2) | 183 (9.7) | |
|
| |||||
| Hypertension (No. [%]) | < .001 | ||||
| No | 3,811 (89.0) | 176 (4.1) | 117 (2.7) | 179 (4.2) | |
| Yes | 4,735 (85.0) | 248 (4.5) | 197 (3.5) | 390 (7.0) | |
|
| |||||
| Hyperlipidemia (No. [%]) | .001 | ||||
| No | 3,636 (87.8) | 171 (4.1) | 138 (3.3) | 194 (4.7) | |
| Yes | 4,910 (85.9) | 253 (4.4) | 176 (3.1) | 375 (6.6) | |
|
| |||||
| Smoking Status (No. [%]) | < .001 | ||||
| Never | 3,975 (90.5) | 145 (3.3) | 100 (2.3) | 170 (3.9) | |
| Past smoker | 3,535 (85.1) | 201 (4.8) | 146 (3.5) | 271 (6.5) | |
| Current smoker | 1,036 (79.1) | 78 (6.0) | 68 (5.2) | 128 (9.8) | |
|
| |||||
| logCRP§ (Mean ± SD¶) | 0.74 ± 1.16A | 0.87 ± 1.14A | 1.05 ± 1.18B | 1.18 ± 1.15B | < .001 |
|
| |||||
| CES-D# Scale Score (Mean ± SD) | 0.95 ± 1.79A | 1.32 ± 2.10B | 1.34 ± 2.26B | 1.33 ± 2.17B | < .001 |
|
| |||||
| Mean Learning Score** (Mean ± SD) | 5.81 ± 1.64A | 5.82 ± 1.57A | 5.55 ± 1.67B | 5.29 ± 1.56C | < .001 |
|
| |||||
| Delay Recall Score†† (Mean ± SD) | 6.39 ± 2.14A | 6.41 ± 2.12A | 6.08 ± 2.20B | 5.82 ± 2.08C | < .001 |
Superscripted capital letters A, B or C indicate the grouping based on Student-Newman-Keuls multiple range test.
All P values were based on χ2 test or analysis of variance (for the last four variables with mean ± standard deviation).
BMI: Body mass index.
logCRP: Logarithm of C-reactive protein.
SD: Standard deviation.
CES-D: Center for Epidemiological Studies-Depression.
Consortium to Establish a Registry for Alzheimer’s Disease word list learning, average of responses across three trials (range, 0–10).
Consortium to Establish a Registry for Alzheimer’s Disease word list delay recall, number of responses to single recall trial (range, 0–10).
Results for incremental linear regression modeling for the mean learning and delay recall outcome variables were similar and are presented in Table 2 and Table 3 (pages 386 and 387), respectively. Regression coefficients represent the association of each risk factor with a one-item (word) higher score from the recall list. The results of our unadjusted analysis showed that higher tooth loss was associated with poorer cognitive function. Participants who reported that they had lost six to 16 teeth (β = −0.26; 95 percent confidence interval [CI], −0.45 to −0.08; P = .005) and more than 16 teeth (β = −0.52; 95 percent CI, −0.65 to −0.38; P < .001) performed significantly worse than did participants who reported that they had lost no teeth (reference group), whereas those who reported that they had lost one to five teeth were not significantly different from the reference group in terms of mean learning score. Results for delay recall were similar to those for mean learning. These associations were attenuated after we adjusted for demographic factors, but significantly lower cognitive scores remained for those who reported that they had lost six or more teeth.
TABLE 2.
Association between mean learning score* and other factors with incremental regression models† (n = 9,853).
| PREDICTOR VARIABLE | UNADJUSTED | ADJUSTED FOR DF‡ | ADJUSTED FOR DF PLUS BMI§ | ADJUSTED FOR DF PLUS BMI PLUS logCRP¶ | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF# | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF PLUS CES-D** | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF PLUS CES-D PLUS SES†† | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||||
| Regression Coefficient (β [95% CI‡‡]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | |
|
| ||||||||||||||
| Teeth Lost (No.) | ||||||||||||||
| 0 (ref.) | ||||||||||||||
| 1–5 | 0.01 (−0.15 to 0.17) | .87 | 0.06 (−0.08 to 0.21) | .408 | 0.06 (−0.08 to 0.21) | .395 | 0.07 (−0.08 to 0.21) | .362 | 0.09 (−0.05 to 0.24) | .197 | 0.12 (−0.02 to 0.26) | .101 | 0.07 (−0.06 to 0.21) | .297 |
| 6–16 | −0.26 (−0.45 to −0.08) | .005 | −0.17 (−0.34 to 0.00) | .048 | −0.17 (−0.34 to 0.00) | .044 | −0.16 (−0.33 to 0.01) | .060 | −0.12 (−0.28 to 0.05) | .163 | −0.10 (−0.26 to 0.07) | .249 | −0.03 (−0.19 to 0.13) | .689 |
| > 16 | −0.52 (−0.65 to −0.38) | < .001 | −0.28 (−0.41 to −0.15) | < .001 | −0.27 (−0.39 to −0.14) | < .001 | −0.25 (−0.38 to −0.12) | .001 | −0.18 (−0.31 to −0.05) | .005 | −0.16 (−0.29 to −0.04) | .011 | −0.01 (−0.13 to 0.12) | .913 |
|
| ||||||||||||||
| Region | ||||||||||||||
| Non–Stroke Belt (ref.) | ||||||||||||||
| Stroke Belt | NA§§ | NA | −0.19 (−0.25 to −0.13) | < .001 | −0.18 (−0.24 to −0.12) | < .001 | −0.18 (−0.24 to −0.12) | < .001 | −0.16 (−0.22 to −0.10) | < .001 | −0.16 (−0.21 to −0.10) | < .001 | −0.10 (−0.15 to −0.04) | .010 |
|
| ||||||||||||||
| Sex | ||||||||||||||
| Male (ref.) | ||||||||||||||
| Female | NA | NA | 0.75 (0.69 to 0.81) | < .001 | 0.76 (0.70 to 0.82) | < .001 | 0.78 (0.72 to 0.84) | < .001 | 0.75 (0.69 to 0.81) | < .001 | 0.79 (0.73 to 0.85) | < .001 | 0.90 (0.84 to 0.96) | < .001 |
|
| ||||||||||||||
| Age (Years) | ||||||||||||||
| 45–54 (ref.) | ||||||||||||||
| 55–64 | NA | NA | −0.37 (−0.47 to −0.26) | < .001 | −0.36 (−0.47 to −0.26) | < .001 | −0.35 (−0.46 to −0.25) | < .001 | −0.32 (−0.43 to −0.22) | < .001 | −0.33 (−0.44 to −0.23) | < .001 | −0.27 (−0.37 to −0.17) | < .001 |
| 65–74 | NA | NA | −1.01 (−1.11 to −0.90) | < .001 | −1.01 (−1.12 to −0.91) | < .001 | −1.01 (−1.11 to −0.90) | < .001 | −0.96 (−1.07 to −0.85) | < .001 | −0.99 (−1.09 to −0.88) | < .001 | −0.80 (−0.90 to −0.69) | < .001 |
| 75–84 | NA | NA | −1.73 (−1.85 to −1.61) | < .001 | −1.76 (−1.88 to −1.63) | < .001 | −1.75 (−1.87 to −1.62) | < .001 | −1.71 (−1.83 to −1.58) | < .001 | −1.72 (−1.85 to −1.60) | < .001 | −1.44 (−1.57 to −1.32) | < .001 |
| 85 or older | NA | NA | −2.35 (−2.63 to −2.07) | < .001 | −2.39 (−2.67 to −2.11) | < .001 | −2.38 (−2.66 to −2.10) | < .001 | −2.39 (−2.67 to −2.11) | < .001 | −2.38 (−2.66 to −2.10) | < .001 | −2.12 (−2.38 to −1.85) | < .001 |
|
| ||||||||||||||
| Income ($) | ||||||||||||||
| < 20,000 (ref.) | ||||||||||||||
| 20,000–34,999 | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.30 (0.21 to 0.39) | < .001 |
| 35,000–74,999 | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.48 (0.39 to 0.57) | < .001 |
| 75,000 or higher | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.70 (0.59 to 0.81) | < .001 |
|
| ||||||||||||||
| Education Level | ||||||||||||||
| No high school diploma (ref.) | ||||||||||||||
| High school diploma or general equivalency degree | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.44 (0.33 to 0.55) | < .001 |
| Some college but no degree | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.67 (0.56 to 0.78) | < .001 |
| College degree or higher | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.91 (0.80 to 1.02) | < .001 |
Mean learning score: Consortium to Establish a Registry for Alzheimer’s Disease word list learning, average of responses across three learning trials (range, 0–10).
All P values are based on a likelihood ratio test.
DF: Demographic factors (region, race, sex, age).
BMI: Body mass index.
logCRP: Logarithm of C-reactive protein.
RF: Risk factors (diabetes, hypertension, hyperlipidemia, smoking).
CES-D: Center for Epidemiological Studies-Depression. The scale ranges from zero to 12; a score of four or higher indicates an elevated level of psychological distress.
SES: Socioeconomic status (income, education).
CI: Confidence interval.
NA: Not applicable.
TABLE 3.
Association between delay recall score* and other factors with incremental regression models† (n = 9,853).
| PREDICTOR VARIABLE | UNADJUSTED | ADJUSTED FOR DF‡ | ADJUSTED FOR DF PLUS BMI§ | ADJUSTED FOR DF PLUS BMI PLUS logCRP¶ | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF# | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF PLUS CES-D** | ADJUSTED FOR DF PLUS BMI PLUS logCRP PLUS RF PLUS CES-D PLUS SES†† | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
| ||||||||||||||
| Regression Coefficient (β [95% CI‡‡]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | Regression Coefficient (β [95% CI]) | P Value | |
|
| ||||||||||||||
| Teeth Lost (No.) | ||||||||||||||
| 0 (ref.) | ||||||||||||||
| 1–5 | 0.02 (−0.19 to 0.23) | .86 | 0.07 (−0.12 to 0.26) | .458 | 0.07 (−0.12 to 0.27) | .451 | 0.08 (−0.11 to 0.27) | .424 | 0.10 (−0.09 to 0.29) | .296 | 0.13 (−0.06 to 0.32) | .183 | 0.08 (−0.11 to 0.27) | .401 |
| 6–16 | −0.30 (−0.54 to −0.06) | .013 | −0.20 (−0.43 to 0.02) | .072 | −0.21 (−0.43 to 0.01) | .067 | −0.20 (−0.42 to 0.03) | .085 | −0.16 (−0.38 to 0.06) | .164 | −0.13 (−0.35 to 0.09) | .235 | −0.06 (−0.28 to 0.15) | .569 |
| > 16 | −0.56 (−0.75 to −0.38) | < .001 | −0.29 (−0.46 to −0.13) | .001 | −0.28 (−0.45 to −0.12) | .001 | −0.27 (−0.44 to −0.10) | .002 | −0.20 (−0.37 to −0.03) | .021 | −0.18 (−0.35 to −0.01) | .036 | −0.01 (−0.17 to 0.16) | .924 |
|
| ||||||||||||||
| Region | ||||||||||||||
| Non–Stroke Belt (ref.) | ||||||||||||||
| Stroke Belt | NA§§ | NA | −0.16 (−0.24 to −0.08) | < .001 | −0.16 (−0.23 to −0.08) | .001 | −0.15 (−0.23 to −0.07) | .001 | −0.13 (−0.21 to −0.05) | .001 | −0.12 (−0.20 to −0.05) | .002 | −0.05 (−0.13 to 0.02) | .164 |
|
| ||||||||||||||
| Sex | ||||||||||||||
| Male (ref.) | ||||||||||||||
| Female | NA | NA | 0.92 (0.84 to 1.00) | < .001 | 0.93 (0.85 to 1.01) | < .001 | 0.95 (0.87 to 1.03) | < .001 | 0.92 (0.84 to 1.01) | < .001 | 0.97 (0.89 to 1.05) | < .001 | 1.10 (1.01 to 1.18) | < .001 |
|
| ||||||||||||||
| Age (Years) | ||||||||||||||
| 45–54 (ref.) | ||||||||||||||
| 55–64 | NA | NA | −0.48 (−0.62 to −0.34) | < .001 | −0.47 (−0.61 to −0.34) | < .001 | −0.47 (−0.60 to −0.33) | < .001 | −0.43 (−0.57 to −0.29) | < .001 | −0.44 (−0.58 to −0.30) | < .001 | −0.36 (−0.50 to −0.23) | < .001 |
| 65–74 | NA | NA | −1.23 (−1.38 to −1.09) | < .001 | −1.24 (−1.39 to −1.10) | < .001 | −1.23 (−1.38 to −1.09) | < .001 | −1.16 (−1.31 to −1.02) | < .001 | −1.20 (−1.34 to −1.05) | <. 001 | −0.98 (−1.12 to −0.83) | < .001 |
| 75–84 | NA | NA | −2.16 (−2.33 to −2.00) | < .001 | −2.19 (−2.35 to −2.02) | < .001 | −2.18 (−2.34 to −2.01) | < .001 | −2.11 (−2.28 to −1.94) | < .001 | −2.12 (−2.29 to −1.96) | < .001 | −1.80 (−1.97 to −1.63) | < .001 |
| 85 or older | NA | NA | −2.74 (−3.11 to −2.37) | < .001 | −2.78 (−3.15 to −2.40) | < .001 | −2.77 (−3.14 to −2.40) | < .001 | −2.74 (−3.11 to −2.37) | < .001 | −2.74 (−3.11 to −2.37) | < .001 | −2.42 (−2.79 to −2.06) | < .001 |
|
| ||||||||||||||
| Income ($) | ||||||||||||||
| < 20,000 (ref.) | ||||||||||||||
| 20,000–34,999 | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.38 (0.26 to 0.50) | < .001 |
| 35,000–74,999 | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.60 (0.48 to 0.72) | < .001 |
| 75,000 or higher | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.83 (0.68 to 0.98) | < .001 |
|
| ||||||||||||||
| Education Level | ||||||||||||||
| No high school diploma (ref.) | ||||||||||||||
| High school diploma or general equivalency degree | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.55 (0.40 to 0.70) | < .001 |
| Some college but no degree | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.75 (0.50 to 0.90) | < .001 |
| College degree or higher | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | NA | 0.97 (0.82 to 1.13) | < .001 |
Delay recall score: Consortium to Establish a Registry for Alzheimer’s Disease word list learning, number of responses to sole recall trial (range, 0–10).
All P values are based on a likelihood ratio test.
DF: Demographic factors (region, race, sex, age).
BMI: Body mass index.
logCRP: Logarithm of C-reactive protein.
RF: Risk Factors (diabetes, hypertension, hyperlipidemia, smoking).
CES-D: Center for Epidemiological Studies-Depression. The scale ranges from zero to 12; a score four or higher indicates an elevated level of psychological distress.
SES: Socioeconomic status (income, education).
CI: Confidence interval.
NA: Not applicable.
After we adjusted for BMI, we found further attenuation of the association between tooth loss and mean learning score for participants who reported that they had lost six or more teeth and between tooth loss and delay recall score for participants who reported that they had lost more than 16 teeth. However, adjustment for BMI revealed no attenuation of the association between tooth loss and mean learning score (before adjusting for BMI: β = −0.17; 95 percent CI, −0.34 to 0.00; P = .048; after adjusting for BMI: β = −0.17; 95 percent CI, −0.34 to 0.00; P = .044) for the participants who reported that they had lost six to 16 teeth. Only among participants who reported that they had lost more than 16 teeth did we note significant associations between tooth loss and mean learning scores (β = −0.25; 95 percent CI, −0.38 to −0.12; P = .001) and delay recall scores (β = −0.27; 95 percent CI, −0.44 to −0.10; P = .002) after adjusting for logCRP.
After we adjusted for risk factors and depressive symptoms, we found that the association was further attenuated, with those who reported that they had lost more than 16 teeth remaining significantly different from the reference group for both cognitive measures (mean learning and delayed recall). In the final model, which included adjustments for income and education (SES variables), the association between tooth loss and cognitive function was extinguished (mean learning score: one to five teeth lost [β = 0.07; 95 percent CI, −0.06 to 0.21; P = .30]; six to 16 teeth lost [β = −0.03; 95 percent CI, −0.19 to 0.13; P = .69]; more than 16 teeth lost [β = −0.01; 95 percent CI, −0.13 to 0.12; P = .91]; delay recall score: one to five teeth lost [β = 0.08; 95 percent CI, −0.11 to 0.27; P = .40]; six to 16 teeth lost [β = −0.06; 95 percent CI, −0.28 to 0.15; P = .57]; more than 16 teeth lost [β = −0.01; 95 percent CI, −0.17 to 0.16; P = .92]). We ran separate models, which included participants who were missing income data (n = 12,661). The results were similar to those from the original models (data not shown).
After we adjusted for demographics, we found approximately 46 percent and 48 percent reductions in regression coefficients for mean learning and delay recall scores, respectively, for the group of participants who reported that they had lost more than 16 teeth. We determined that the group of participants 65 years or older was largely responsible for this attenuation of the association. Therefore, we evaluated the interaction of age and tooth loss in relation to cognitive function. The interaction term was significant (mean learning score [β = 0.06; 95 percent CI, 0.01 to 0.10; P = .015]; delayed recall score [β = 0.07; 95 percent CI, 0.01 to 0.13; P = .022]) after we adjusted for demographics, BMI and logCRP, with people older than 54 years displaying a higher likelihood of poor cognitive function with tooth loss than did people aged 45 to 54 years. However, this finding became nonsignificant after we adjusted for risk factors, and it remained so after we adjusted for SES (data not shown). The results of our smoking-by-tooth-loss interaction analysis were similar; current smokers had a significantly higher likelihood of poor cognitive function with tooth loss after we adjusted for systemic risk factors than did those who never smoked. This association, however, became nonsignificant after we adjusted for depressive symptoms and for SES (data not shown).
DISCUSSION
The main finding of our study was that tooth loss due to periodontitis was not associated independently with lower scores for learning and recall in our full regression models. We accounted for the associations seen in earlier models by adjusting for other risk factors, mainly age and SES.
Interpreting the attenuating effect of SES can be complex, because SES might be associated with tooth loss and cognitive function via multiple pathways. In our study, we were unable to address whether tooth loss is a mechanism whereby SES relates to cognitive function. Tooth loss may be a surrogate marker for SES. Thus, the strong association between tooth loss and SES shown in Table 1 may make it impossible to assess their independent roles. For example, only 1.2 percent of participants with an annual family income of $75,000 or higher had lost 16 or more teeth compared with 11.7 percent of participants with an annual family income less than $20,000. The association of tooth loss and education was nearly as strong. This association of tooth loss and SES might make the interpretation of the association between the exposure and outcome variables problematic. Therefore, we do not suggest that tooth loss due to periodontal disease is an independent risk factor for cognitive impairment. Furthermore, nondental factors associated with SES may affect cognition. In this scenario SES would be a classic confounding factor in the relationship between tooth loss and cognitive function and thereby introduce an apparent association when there is none. Alternatively, tooth loss may be in the direct pathway of action between SES and cognitive function and may be one of the mechanisms through which lower SES is associated with lower cognitive function. Therefore, one should interpret the complete mediation of the effect of tooth loss with the addition of SES measures with caution.
We observed little confounding of the association of tooth loss and cognitive function owing to CRP concentration. This finding suggests that inflammation associated with tooth loss was not a mediator of lower cognitive function among those who had lost teeth in our study.
The limitations of our study were inherent in its cross-sectional design, which made determinations of directionality impossible, and in the self-reported tooth loss data, which could have resulted in low reliability. The cross-sectional design is relevant to the potential effect of CRP as a mediator. Although the progression of periodontitis is time dependent,52 the severity of the disease may not be age dependent.53–59 For example, a 25-year-old and a 65-year-old could have periodontitis of the same severity. Although the lack of significant age-by-tooth-loss and smoking-by-tooth-loss interactions in the full models in our study was consistent with findings from other studies,11 this information was inconclusive because our results cannot be used to rule out smoking as a cause of tooth loss or clearly determine the relationship between age and tooth loss. It is, however, reasonable to hypothesize that the severity of periodontitis will be affected by the number of teeth remaining, as well as other factors such as hygiene and professional care. As teeth are lost and their sockets heal, these areas of the oral cavity theoretically no longer contribute to periodontitis and should lead to a reduction in oral inflammation. The design of our study cannot account for these likely changes in CRP concentrations and their potential effects on cognitive function across time.
The specificity and sensitivity of the survey items used to measure tooth loss may have added to measurement uncertainty. We cannot guarantee that all of the tooth loss we analyzed was due to periodontitis and not caries, trauma or elective extraction. Dietrich and colleagues48 determined that some self-reported periodontitis items had low sensitivity and positive predictive value, although many of them had more than 90 percent specificity. This finding may be a bigger issue because some participants in our study may have had mild cognitive impairment when they were surveyed about tooth loss despite efforts to exclude them from the study. Blicher and colleagues,60 however, suggested that self-reported measures of periodontitis show “good potential.”
Stein and colleagues61 assessed periodontitis by measuring alveolar bone loss on dental radiographs. Although they found an association between a low number of remaining teeth and an increased risk of developing dementia, they failed to find a significant association between periodontitis, measured by means of radiographic evidence of alveolar bone loss, and dementia, measured by means of cognitive assessment. We recommend that more robust measures of periodontitis—such as levels of antibodies against periodontal pathogens, periodontal probing depths, attachment levels and standardized radiographs—be used to explore this potential relationship further.
A longitudinal study design with methods that Watts and colleagues62 suggested may help elucidate the elements directly in the causal pathway. Investigators in future studies must consider age and SES carefully as potential confounding factors and incorporate the proper measurement of the proper variables into the design. Finding a measure of oral inflammation related to tooth loss that is independent of SES would be ideal given the complex role SES seems to play in the potential association. Although our assessment of possible effect modification and interaction was not hypothesis driven, we recommend incorporating these covariate assessments into the a priori analysis plan as Hyman63 suggested.
CONCLUSIONS
In this analysis, the association of tooth loss and cognitive function as assessed by means of WLL was mediated largely by SES, by age of 65 years or older and to a lesser degree by other risk factors, including CRP as an inflammation marker. Therefore, we do not reject the null hypothesis that there is no association between tooth loss and cognitive function. We are considering future analyses.
Acknowledgments
This research project was supported by cooperative agreement U01 NS041588 from the National Institute of Neurological Disorders and Stroke (NINDS), National Institutes of Health, Department of Health and Human Services. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the National Institutes of Health. Representatives of NINDS have been involved in the review of the manuscript but not directly involved in the collection, management, analysis or interpretation of the data.
The authors thank the other investigators, the staff and the participants of the Reasons for Geographic and Racial Differences in Stroke study for their valuable contributions. A full list of participating investigators and institutions is available at “www.regardsstudy.org”.
ABBREVIATION KEY
- BMI
Body mass index
- CERAD
Consortium to Establish a Registry for Alzheimer’s Disease
- CES-D
Center for Epidemiological Studies-Depression
- CRP
C-reactive protein
- DF
Demographic factors
- logCRP
Logarithm of CRP
- REGARDS
REasons for Geographic And Racial Differences in Stroke
- RF
Risk factors
- SES
Socioeconomic status
- WLL
Word list learning
Footnotes
Disclosure. None of the authors reported any disclosures.
Contributor Information
Dr. Jonathan C. Matthews, Postdoctoral trainee, Institute of Oral Health Research, School of Dentistry, University of Alabama at Birmingham, when this article was written. He now is in private practice in Birmingham, Ala.
Dr. Zhiying You, Assistant professor, Division of Preventive Medicine, Department of Medicine, University of Alabama at Birmingham.
Dr. Virginia G. Wadley, Associate professor, Department of Medicine, Division of Gerontology, Geriatrics, and Palliative Care, University of Alabama at Birmingham.
Dr. Mary Cushman, Professor of medicine, University of Vermont—Burlington, Vt.
Dr. George Howard, Chair and a professor, Department of Biostatistics, School of Public Health, University of Alabama at Birmingham.
References
- 1.Chalmers JM, Carter KD, Spencer AJ. Oral diseases and conditions in community-living older adults with and without dementia. Spec Care Dentist. 2003;23(1):7–17. doi: 10.1111/j.1754-4505.2003.tb00283.x. [DOI] [PubMed] [Google Scholar]
- 2.Gordon SR. Survey of dental need among veterans with severe cognitive impairment. Gerodontics. 1988;4(4):158–159. [PubMed] [Google Scholar]
- 3.Weyant RJ, Pandav RS, Plowman JL, Ganguli M. Medical and cognitive correlates of denture wearing in older community-dwelling adults. J Am Geriatr Soc. 2004;52(4):596–600. doi: 10.1111/j.1532-5415.2004.52168.x. [DOI] [PubMed] [Google Scholar]
- 4.Jones JA, Lavallee N, Alman J, Sinclair C, Garcia RI. Caries incidence in patients with dementia. Gerodontology. 1993;10(2):76–82. doi: 10.1111/j.1741-2358.1993.tb00086.x. [DOI] [PubMed] [Google Scholar]
- 5.Ship JA, Puckett SA. Longitudinal study on oral health in subjects with Alzheimer’s disease. J Am Geriatr Soc. 1994;42(1):57–63. doi: 10.1111/j.1532-5415.1994.tb06074.x. [DOI] [PubMed] [Google Scholar]
- 6.Wu B, Plassman BL, Crout RJ, Liang J. Cognitive function and oral health among community-dwelling older adults. J Gerontol A Biol Sci Med Sci. 2008;63(5):495–500. doi: 10.1093/gerona/63.5.495. [DOI] [PubMed] [Google Scholar]
- 7.Kamer AR, Craig RG, Dasanayake AP, Brys M, Glodzik-Sobanska L, de Leon MJ. Inflammation and Alzheimer’s disease: possible role of periodontal diseases. Alzheimers Dement. 2008;4(4):242–250. doi: 10.1016/j.jalz.2007.08.004. [DOI] [PubMed] [Google Scholar]
- 8.Kamer AR, Dasanayake AP, Craig RG, Glodzik-Sobanska L, Bry M, de Leon MJ. Alzheimer’s disease and peripheral infections: the possible contribution from periodontal infections, model and hypothesis. J Alzheimers Dis. 2008;13(4):437–449. doi: 10.3233/jad-2008-13408. [DOI] [PubMed] [Google Scholar]
- 9.Stein P, Scheff S, Dawson DR. Alzheimer’s disease and periodontal disease: mechanisms underlying a potential bidirectional relationship. Grand Rounds Oral Systemic Med. 2006;1:14–24. [Google Scholar]
- 10.Grabe HJ, Schwahn C, Völzke H, et al. Tooth loss and cognitive impairment. J Clin Periodontol. 2009;36(7):550–557. doi: 10.1111/j.1600-051X.2009.01426.x. [DOI] [PubMed] [Google Scholar]
- 11.Noble JM, Borrell LN, Papapanou PN, Elkind MS, Scarmeas N, Wright CB. Periodontitis is associated with cognitive impairment among older adults: analysis of NHANES-III. J Neurol Neurosurg Psychiatry. 2009;80(11):1206–1211. doi: 10.1136/jnnp.2009.174029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kaye EK, Valencia A, Baba N, Spiro A, 3rd, Dietrich T, Garcia RI. Tooth loss and periodontal disease predict poor cognitive function in older men. J Am Geriatr Soc. 2010;58(4):713–718. doi: 10.1111/j.1532-5415.2010.02788.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kondo K, Niino M, Shido K. A case-control study of Alzheimer’s disease in Japan: significance of life-styles. Dementia. 1994;5(6):314–326. doi: 10.1159/000106741. [DOI] [PubMed] [Google Scholar]
- 14.Shimazaki Y, Soh I, Saito T, et al. Influence of dentition status on physical disability, mental impairment, and mortality in institutionalized elderly people. J Dent Res. 2001;80(1):340–345. doi: 10.1177/00220345010800010801. [DOI] [PubMed] [Google Scholar]
- 15.Gatz M, Mortimer JA, Fratiglioni L, et al. Potentially modifiable risk factors for dementia in identical twins. Alzheimers Dement. 2006;2(2):110–117. doi: 10.1016/j.jalz.2006.01.002. [DOI] [PubMed] [Google Scholar]
- 16.Le Y, Murphy PM, Wang JM. Formyl-peptide receptors revisited. Trends Immunol. 2002;23(11):541–548. doi: 10.1016/s1471-4906(02)02316-5. [DOI] [PubMed] [Google Scholar]
- 17.Little JW, Falace DA, Miller CS, Rhodus NL. Dental Management of the Medically Compromised Patient. 6. St. Louis: Mosby; 2002. p. 33. [Google Scholar]
- 18.Riviere GR, Riviere KH, Smith KS. Molecular and immunological evidence of oral Treponema in the human brain and their association with Alzheimer’s disease. Oral Microbiol Immunol. 2002;17(2):113–118. doi: 10.1046/j.0902-0055.2001.00100.x. [DOI] [PubMed] [Google Scholar]
- 19.Balin BJ, Appelt DM. Role of infection in Alzheimer’s disease. J Am Osteopath Assoc. 2001;101(12 suppl pt 1):S1–S6. [PubMed] [Google Scholar]
- 20.Balin BJ, Gérard HC, Arking EJ, et al. Identification and localization of Chlamydia pneumoniae in the Alzheimer’s brain. Med Microbiol Immunol. 1998;187(1):23–42. doi: 10.1007/s004300050071. [DOI] [PubMed] [Google Scholar]
- 21.Itzhaki RF, Wozniak MA, Appelt DM, Balin BJ. Infiltration of the brain by pathogens causes Alzheimer’s disease. Neurobiol Aging. 2004;25(5):619–627. doi: 10.1016/j.neurobiolaging.2003.12.021. [DOI] [PubMed] [Google Scholar]
- 22.Wozniak MA, Shipley SJ, Combrinck M, Wilcock GK, Itzhaki RF. Productive herpes simplex virus in brain of elderly normal subjects and Alzheimer’s disease patients. J Med Virol. 2005;75(2):300–306. doi: 10.1002/jmv.20271. [DOI] [PubMed] [Google Scholar]
- 23.Gieffers J, Reusche E, Solbach W, Maass M. Failure to detect Chlamydia pneumoniae in brain sections of Alzheimer’s disease patients. J Clin Microbiol. 2000;38(2):881–882. doi: 10.1128/jcm.38.2.881-882.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.McLaughlin R, Kin NM, Chen MF, Nair NP, Chan EC. Alzheimer’s disease may not be a spirochetosis. Neuroreport. 1999;10(7):1489–1491. doi: 10.1097/00001756-199905140-00018. [DOI] [PubMed] [Google Scholar]
- 25.Nochlin D, Shaw CM, Cmpbell LA, Kuo CC. Failure to detect Chlamydia pneumoniae in brain tissues of Alzheimer’s disease. Neurology. 1999;53(8):1888. doi: 10.1212/wnl.53.8.1888-a. [DOI] [PubMed] [Google Scholar]
- 26.Ring RH, Lyons JM. Failure to detect Chlamydia pneumoniae in the late-onset Alzheimer’s brain. J Clin Microbiol. 2000;38(7):2591–2594. doi: 10.1128/jcm.38.7.2591-2594.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Wozniak MA, Cookson A, Wilcock GK, Itzhaki RF. Absence of Chlamydia pneumoniae in brain of vascular dementia patients. Neurobiol Aging. 2003;24(6):761–765. doi: 10.1016/s0197-4580(02)00236-1. [DOI] [PubMed] [Google Scholar]
- 28.Ebersole JL, Cappelli D. Acute-phase reactants in infections and inflammatory diseases. Periodontol 2000. 2000;23(1):19–49. doi: 10.1034/j.1600-0757.2000.2230103.x. [DOI] [PubMed] [Google Scholar]
- 29.Page RC. The role of inflammatory mediators in the pathogenesis of periodontal disease. J Periodontal Res. 1991;26(3 pt 2):230–242. doi: 10.1111/j.1600-0765.1991.tb01649.x. [DOI] [PubMed] [Google Scholar]
- 30.Grossi SG, Genco RJ. Periodontal disease and diabetes mellitus: a two-way relationship. Ann Periodontol. 1998;3(1):51–61. doi: 10.1902/annals.1998.3.1.51. [DOI] [PubMed] [Google Scholar]
- 31.Iacopino AM, Cutler CW. Pathophysiological relationships between periodontitis and systemic disease: recent concepts involving serum lipids. J Periodontol. 2000;71(8):1375–1384. doi: 10.1902/jop.2000.71.8.1375. [DOI] [PubMed] [Google Scholar]
- 32.Loesche WJ. Periodontal disease: link to cardiovascular disease. Compend Contin Educ Dent. 2000;21(6):463–466. 468. [PubMed] [Google Scholar]
- 33.Roberts FA, Richardson GJ, Michalek SM. Effects of Porphyromonas gingivalis and Escherichia coli lipopolysaccharides on mononuclear phagocytes. Infect Immun. 1997;65(8):3248–3254. doi: 10.1128/iai.65.8.3248-3254.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Offenbacher S. Periodontal diseases: pathogenesis. Ann Periodontol. 1996;1(1):821–878. doi: 10.1902/annals.1996.1.1.821. [DOI] [PubMed] [Google Scholar]
- 35.Amar S, Han X. The impact of periodontal infection on systemic diseases. Med Sci Monit. 2003;9(12):RA291–RA299. [PubMed] [Google Scholar]
- 36.Page RC. The pathobiology of periodontal diseases may affect systemic diseases: inversion of a paradigm. Ann Periodontol. 1998;3(1):108–120. doi: 10.1902/annals.1998.3.1.108. [DOI] [PubMed] [Google Scholar]
- 37.Akiyama H, Barger S, Barnum S, et al. Inflammation and Alzheimer’s disease. Neurobiol Aging. 2000;21(3):383–421. doi: 10.1016/s0197-4580(00)00124-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.McGreer EG, McGreer PL. Role of inflammatory processes and microglial activation in Alzheimer disease. In: Terry RD, Katzman R, Bick KL, Sisodia SS, editors. Alzheimer Disease. 2. Philadelphia: Lippincott Williams & Wilkins; 1999. pp. 389–397. [Google Scholar]
- 39.Arvanitakis Z, Hachinski V. Vascular cognitive impairment: what else do we need to learn? In: Terry RD, Katzman R, Bick KL, Sisodia SS, editors. Alzheimer Disease. 2. Philadelphia: Lippincott Williams & Wilkins; 1999. pp. 148–149. [Google Scholar]
- 40.Reitz C, Luchsinger JA, Tang MX, Manly J, Mayeux R. Stroke and memory performance in elderly persons without dementia. Arch Neurol. 2006;63(4):571–576. doi: 10.1001/archneur.63.4.571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Beck J, Garcia R, Heiss G, Vokonas PS, Offenbacher S. Periodontal disease and cardiovascular disease. J Periodontol. 1996;67 (10 suppl):1123–1137. doi: 10.1902/jop.1996.67.10s.1123. [DOI] [PubMed] [Google Scholar]
- 42.Elter JR, Offenbacher S, Toole JF, Beck JD. Relationship of periodontal disease and edentulism to stroke/TIA. J Dent Res. 2003;82(12):998–1001. doi: 10.1177/154405910308201212. [DOI] [PubMed] [Google Scholar]
- 43.Grau AJ, Becher H, Ziegler CM, et al. Periodontal disease as a risk factor for ischemic stroke. Stroke. 2004;35(2):496–501. doi: 10.1161/01.STR.0000110789.20526.9D. [DOI] [PubMed] [Google Scholar]
- 44.Grau AJ, Buggle F, Ziegler C, et al. Association between acute cerebrovascular ischemia and chronic and recurrent infection. Stroke. 1997;28(9):1724–1729. doi: 10.1161/01.str.28.9.1724. [DOI] [PubMed] [Google Scholar]
- 45.Joshipura KJ, Hung HC, Rimm EB, Willett WC, Ascherio A. Periodontal disease, tooth loss, and incidence of ischemic stroke. Stroke. 2003;34(1):47–52. doi: 10.1161/01.str.0000052974.79428.0c. [DOI] [PubMed] [Google Scholar]
- 46.Borhani NO. Changes and geographic distribution of mortality from cerebrovascular disease. Am J Public Health Nations Health. 1965;55(5):673–681. doi: 10.2105/ajph.55.5.673. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Morris JC, Heyman A, Mohs RC, et al. The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), part I: clinical and neuropsychological assessment of Alzheimer’s disease. Neurology. 1989;39(9):1159–1165. doi: 10.1212/wnl.39.9.1159. [DOI] [PubMed] [Google Scholar]
- 48.Dietrich T, Stosch U, Dietrich D, Schamberger D, Bernimoulin JP, Joshipura K. The accuracy of individual self-reported items to determine periodontal disease history. Eur J Oral Sci. 2005;113(2):135–140. doi: 10.1111/j.1600-0722.2004.00196.x. [DOI] [PubMed] [Google Scholar]
- 49.You Z, Cushman M, Jenny NS, Howard G. REGARDS. Tooth loss, systemic inflammation, and prevalent stroke among participants in the reasons for geographic and racial difference in stroke (REGARDS) study. Atherosclerosis. 2009;203(2):615–619. doi: 10.1016/j.atherosclerosis.2008.07.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Melchior LA, Huba GJ, Brown VB, Reback CJ. A short depression index for women. Educ Psychological Meas. 1993;53(4):1117–1125. [Google Scholar]
- 51.Mackinnon DP, Warsi G, Dwyer JH. A simulation study of mediated effect measures. Multivariate Behav Res. 1995;30(1):41. doi: 10.1207/s15327906mbr3001_3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Schätzle M, Löe H, Lang NP, et al. Clinical course of chronic periodontitis, III: patterns, variations and risks of attachment loss. J Clin Periodontol. 2003;30(10):909–918. doi: 10.1034/j.1600-051x.2003.00401.x. [DOI] [PubMed] [Google Scholar]
- 53.Armitage GC. Development of a classification system for periodontal diseases and conditions. Ann Periodontol. 1999;4(1):1–6. doi: 10.1902/annals.1999.4.1.1. [DOI] [PubMed] [Google Scholar]
- 54.Papapanou PN. Periodontal diseases: epidemiology. Ann Periodontol. 1996;1(1):1–36. doi: 10.1902/annals.1996.1.1.1. [DOI] [PubMed] [Google Scholar]
- 55.Aass AM, Albandar J, Aasenden R, Tollefsen T, Gjermo P. Variation in prevalence of radiographic alveolar bone loss in subgroups of 14-year-old schoolchildren in Oslo. J Clin Periodontol. 1988;15(2):130–133. doi: 10.1111/j.1600-051x.1988.tb01006.x. [DOI] [PubMed] [Google Scholar]
- 56.Van der Velden U, Abbas F, Van Steenbergen TJ, et al. Prevalence of periodontal breakdown in adolescents and presence of Actinobacillus actinomycetemcomitans in subjects with attachment loss. J Periodontol. 1989;60(11):604–610. doi: 10.1902/jop.1989.60.11.604. [DOI] [PubMed] [Google Scholar]
- 57.Cappelli DP, Ebersole JL, Kornman KS. Early-onset periodontitis in Hispanic-American adolescents associated with A. actinomycetemcomitans. Community Dent Oral Epidemiol. 1994;22(2):116–121. doi: 10.1111/j.1600-0528.1994.tb01585.x. [DOI] [PubMed] [Google Scholar]
- 58.Clerehugh V, Worthington HV, Lennon MA, Chandler R. Site progression of loss of attachment over 5 years in 14- to 19-year-old adolescents. J Clin Periodontol. 1995;22(1):15–21. doi: 10.1111/j.1600-051x.1995.tb01766.x. [DOI] [PubMed] [Google Scholar]
- 59.Timmerman MF, Van der Weijden GA, Armand S, et al. Untreated periodontal disease in Indonesian adolescents: clinical and microbiological baseline data. J Clin Periodontol. 1998;25(3):215–224. doi: 10.1111/j.1600-051x.1998.tb02431.x. [DOI] [PubMed] [Google Scholar]
- 60.Blicher B, Joshipura K, Eke P. Validation of self-reported periodontal disease: a systematic review. J Dent Res. 2005;84(10):881–890. doi: 10.1177/154405910508401003. [DOI] [PubMed] [Google Scholar]
- 61.Stein PS, Desrosiers M, Donegan SJ, Yepes JF, Kryscio RJ. Tooth loss, dementia and neuropathology in the Nun study. JADA. 2007;138(10):1314–1322. doi: 10.14219/jada.archive.2007.0046. [DOI] [PubMed] [Google Scholar]
- 62.Watts A, Crimmins EM, Gatz M. Inflammation as a potential mediator for the association between periodontal disease and Alzheimer’s disease. Neuropsychiatr Dis Treat. 2008;4(5):865–876. doi: 10.2147/ndt.s3610. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Hyman J. The importance of assessing confounding and effect modification in research involving periodontal disease and systemic diseases. J Clin Periodontol. 2006;33(2):102–103. doi: 10.1111/j.1600-051X.2005.00881.x. [DOI] [PubMed] [Google Scholar]