Abstract
Aim:
Examine the association of dietary patterns with periodontal disease (PD) and its progression over 5 years.
Materials and Method:
Analyses involved 1,197 postmenopausal women from OsteoPerio cohort. Dietary patterns assessed include Healthy Eating Index-2015 (HEI), Alternative HEI (AHEI), Dietary Approaches to Stop Hypertension (DASH), and alternate Mediterranean Diet (aMed) at baseline (the average of two food frequency questionnaires administered between 1993–2001). At baseline and the 5-year follow-up, periodontal assessments evaluated alveolar crestal height (ACH), probing pocket depth (PPD), clinical attachment loss (CAL), percentage of gingival sites bleeding on probing (%BOP) and missing teeth due to PD. Linear and logistic regression were used to examine the associations.
Results:
Cross-sectionally, HEI and aMed were associated with smaller CAL and %BOP; along with DASH, they were associated with a decreased odds of teeth missing due to PD. AHEI and aMed were associated with a decreased odds of severe PD. Prospectively, AHEI was associated with greater ACH progression. This association was attenuated to the null after loss of ACH was imputed for teeth lost due to PD over follow-up, or after excluding participants with diabetes, osteoporosis, hypertension, or heart disease at baseline.
Conclusion:
Better adherence to healthy dietary patterns was associated with better PD measures cross-sectionally but greater progression of ACH over 5 years. The latter might be explained by incident tooth loss due to PD and pre-existing comorbidities.
Keywords: dietary pattern, nutritional epidemiology, periodontal disease, tooth loss, women’s health
INTRODUCTION
Periodontal disease (PD) results from an infection that occurs in the supporting tissues of the teeth and is a leading cause of tooth loss. Approximately forty-two percent of American adults aged 30 years and older are affected by PD (1). PD can provide entry points within the gingiva for periodontal bacteria invasion and trigger inflammatory and host immune responses (2). Evidence has shown that PD is associated with an increased risk of diabetes (2), cardiovascular disease (3), adverse pregnancy outcomes (4), and respiratory disease (5).
In recent years, several studies have examined the impact of score-based dietary patterns, which refer to the combination of foods and nutrients consumed, on PD. Better adherence to the Healthy Eating Index (HEI) (6), Alternative HEI (AHEI) (7), Dietary Approaches to Stop Hypertension (DASH) (8), Mediterranean diet (9), and pro-/anti-inflammatory diets (10,11) was found to be associated with a lower prevalence or severity of PD from cross-sectional studies. However, cross-sectional studies are limited in their ability to infer causality or to track the progression of PD. While some observational studies have prospectively examined incident PD, these studies did not monitor the progression of quantitative clinical measures of PD over time and only included subjects free from PD at enrollment (12–14). Furthermore, a randomized controlled trial indicated the Mediterranean diet’s protective effect against gingivitis but excluded individuals with pre-existing periodontitis (15).
In this study, we used data from the Buffalo Osteoporosis and Periodontal Disease (OsteoPerio) Study of older postmenopausal women to examine the association of four dietary patterns, including the HEI-2015, AHEI, DASH, and the alternate Mediterranean Diet (aMed), with the prevalence and severity of periodontal disease and its progression over five years.
MATERIAL AND METHODS
Study Sample
The Buffalo OsteoPerio cohort comprises postmenopausal women recruited from the Women’s Health Initiative Observational Study (WHI-OS) at the Buffalo, New York clinical center (16). The OsteoPerio Study explores risk factors for the presence, severity, and progression of osteoporosis and PD in postmenopausal women aging 51–83 years at enrollment. Women with fewer than six teeth, bone disease, or a serious illness such as cancer were not eligible to enroll. The baseline visit of the OsteoPerio cohort coincided with participants’ WHI-OS year three clinic visit. Of the 2,249 women in the Buffalo WHI-OS cohort, 1,342 women further enrolled in the OsteoPerio study and completed whole mouth clinical PD examinations at baseline (1997–2001). The OsteoPerio 5-year follow-up visits took place between 2002–2005 (n=1,026). Individuals with complete data for dietary assessment, PD measures, and covariates, and with reported energy intake between 600 to 5,000 kcals at each assessment were included in the analytical sample (n = 1,197 at baseline; n=894 at 5-year follow-up) for the present study (Figure 1). This study was reviewed and approved by the Institutional Review Board of the University at Buffalo.
Figure 1.
Flowchart for the Study Sample for the Analysis of the Association between Dietary Patterns and Periodontal Disease in OsteoPerio cohort
Dietary Assessment
At the WHI-OS baseline (1993–1998), OsteoPerio baseline, and the OsteoPerio 5-year follow-up visit, dietary intake was assessed by a modified Block food frequency questionnaires (FFQ) containing 122 main questions and four summary questions (17,18). Participants self-reported the frequency and portion size they consumed of the queried foods and beverages in the past three months.
The Nutrition Assessment Shared Resource (NASR) of the Fred Hutchinson Cancer Research Center (Seattle, WA) estimated MyPyramid Equivalent intake for 32 MyPyramid food groups for each participant from their FFQ data using the MyPyramid Equivalents Database (MPED) (19). MyPyramid Equivalents, measured in cups, ounces, and teaspoons, were estimated for each of the 32 MyPyramid major groups and subgroups categorized into eight groups, including grain group, vegetable group, fruit group, milk group, meat and beans group, discretionary fats, added sugar, and alcoholic beverages (19). The estimated intake, in MyPyramid equivalents, of each of the 32 MPED groups/subgroups were used, along with dietary data on nutrients and food items, to calculate dietary pattern scores, including HEI, AHEI, DASH, and aMed. HEI primarily focuses on maintaining a healthy weight, promoting overall health, and preventing disease among the US population (20); AHEI, a modified version of HEI, was designed for the prevention of chronic disease (21); DASH diet emphasizes the prevention and control of hypertension (22); and aMed was modeled after the typical dietary styles of people living near the Mediterranean Sea, where chronic disease rates are considerably lower than in populations consuming westernized diets (23,24). Dietary pattern scores at baseline were calculated from the averaged intake of MPED group, food, and nutrient intakes at WHI baseline and OsteoPerio baseline, assumed to represent usual dietary intake over three years leading up to OsteoPerio baseline.
Periodontal Assessment
Periodontal assessments, including intraoral radiographic measures of alveolar crestal height (ACH), probing pocket depth (PPD) and clinical attachment loss (CAL), were conducted at each clinical OsteoPerio follow-up visit by trained and calibrated dental examiners (16). ACH, the distance from the cemental-enamel junction to the alveolar crest in millimeters (mm), was measured at interproximal sites in seven anterior periapical and four posterior vertical bitewing radiographs using a single radiographic unit (Bennett HFQ 300 high-frequency x-ray generator, Bennett X-ray Corp., Copiague, NY). For prospective analyses, paired radiograph pictures (from baseline and the respective follow-up) were exhibited on the same monitor, and the second image was aligned with the first image using a flicker system. This approach makes it possible to compute the difference in ACH over time using the same anatomical landmark on each tooth, which reduces measurement error and enhances the accuracy of determining true ACH (25). PPD, the distance from the gingival margin to the base of the sulcus/pocket in mm, was measured at six surfaces around each tooth using a constant force (20g) electronic probe (The Florida Probe System, Florida Probe, Gainesville, FL.). CAL, the length from the cementoenamel junction to the depth of the pocket base in mm, is an assessment of the pocket depth and gingival recession (26). CAL was measured using a manual periodontal probe on six surfaces of each tooth (Michigan O periodontal probe, Hu-Friedy, Chicago, IL.). Whole mouth mean ACH, PPD, and CAL (mm, each), were summarized for each participant, focusing exclusively on interproximal sites. The percentage of bleeding on probing (%BOP) was assessed by the presence or absence of bleeding on manual periodontal probing at three sites of all teeth and reported as a percentage of the whole mouth sites measured.
At each visit, the dental status of each of the 28 tooth sites was examined (the 3rd molars excluded) for each woman, and the number of decayed, missing, and filled teeth (DMFT) were recorded. For each missing tooth, the reason for the missing tooth was recorded, and PD was one possible reason. We derived a dichotomous outcome variable (yes/no) for new tooth loss due to PD over follow-up by subtracting the number of teeth recorded as missing from PD at the 5-year visit from the number of teeth recorded as missing from PD at the baseline examination.
The presence and severity of PD were defined via two approaches. One approach, the OsteoPerio definition, is based on alveolar crestal height (ACH) and tooth loss as previously described (16). CAL and PD measures at interproximal sites were used to define prevalent periodontal disease status based on the definition from Centers for Disease Control and Prevention/American Academy of Periodontology (CDC/AAP) (27). We also computed a dichotomous variable (yes/no) for PD progression from baseline to follow-up. Participants with ≥2 teeth with ≥1 mm ACH increment from baseline to year 5, or ≥1 new tooth loss due to PD during this time, were defined as having “progressed”.
Covariates
Participants completed self-administered questionnaires at baseline (16) to collect data on their demographic and lifestyle information. Data collected included variables for age at baseline OsteoPerio visit, race, ethnicity, education level, household income, neighborhood socioeconomic status (nSES; higher score indicates better nSES), smoking status and history, alcohol consumption, physical activity, menopausal hormone therapy use, medication use by inventory, chronic and other disease history, frequency of brushing teeth, frequency of flossing, and frequency of dental visits. Trained and certified technicians measured standing height (cm) and body weight (kg) using a calibrated clinical scale and stadiometer, which was used to calculate body mass index (BMI; kg/m2). At OsteoPerio baseline, women were asked to report their history of the comorbidities of diabetes, osteoporosis, hypertension, and heart disease. They were asked if they had ever been diagnosed or treated for diabetes, osteoporosis, hypertension, or heart disease.
Statistical Analysis
All statistical analyses were conducted in SAS software, version 9.4 (SAS Institute Inc., Cary, NC). The mean and standard deviation (SD) of each dietary pattern score at baseline were summarized by categories of participant characteristics; similarly, the distribution for each continuous periodontal measurement at baseline was summarized. T- tests were used to examine the differences for characteristics with two categories and ANOVA F-tests were used for characteristics with more than two categories. P-values were deemed statistically significant at α=0.05.
Cross-sectional associations at each clinic visit
Linear regression models were used to examine the cross-sectional associations at baseline between each dietary pattern score and PD measurements (the whole mouth mean ACH, PPD, CAL, and % BOP). The four dietary pattern scores in this study have different ranges but all have 10 components, so we assigned different units to the four dietary pattern scores to facilitate the comparison of their estimates as predictors. Analyses on AHEI and HEI were performed in units of 10, on DASH in units of 4, and on aMed in units of 1, which are equal to the range of each component in the dietary pattern scores and can indicate the possible increase within one component in a dietary pattern. Binary and multinomial logistic regression models were used to examine the cross-sectional association of continuous measures of dietary pattern scores with the history of having missing teeth reported as due to PD (yes/no) and with the definition-based PD (none/mild-moderate/severe) outcome, respectively. Confounders were selected based on existing literature that suggested their associations with both diet and periodontal health (28,29). Models adjusted for age, race, recreational physical activity, education level, nSES, smoking status, menopausal hormone therapy use, dental hygiene variables (frequency of brushing teeth, flossing, and dental visits) and total calorie intake.
Prospective associations
Endpoints for prospective analyses consist of chronic indicators including whole mouth mean ACH and CAL, the progression of periodontitis variable, and incident new tooth loss due to PD (yes or no) at year 5. Linear regression was used to examine associations between continuous measures of baseline dietary pattern scores and measures of ACH and CAL at the 5-year follow-up with the adjustment for the respective baseline periodontal measurements. Examining measures of ACH and CAL at the 5-year follow-up, while adjusting for the baseline measures of ACH and CAL, is equivalent to examining changes in PD measures over time based on differences in ACH and CAL. Binary logistic regression was used to examine the association of dietary pattern scores with our dichotomous outcomes of incident new tooth loss due to PD (yes/no) and our definition of progression of periodontitis (yes/no). Models were adjusted for the same set of covariates described for cross-sectional associations.
Sensitivity Analyses
Tooth loss due to periodontal disease during follow-up may spuriously attenuate associations with periodontal disease measurements by possibly omitting the most severely affected sites when computing the whole mouth averages for ACH and CAL. Therefore, we conducted sensitivity analyses to account for the effect of the loss of teeth due to periodontal disease. As we do not have an exact estimate of the amount or oral bone of soft tissue corresponding to each lost tooth, we ran a series of sensitivity analyses assuming a loss of 1, 2, 3 or 4 mm in ACH or CAL for that tooth over follow-up. With each imputation we reran the prospective linear regression models.
It is also possible that women with self-reported PD related comorbidities (diabetes, osteoporosis, hypertension, and heart disease) at OsteoPerio baseline may have improved their diets prior the start of WHI-OS, biasing the association between diet and PD. To examine the effect of women changing their diet because of comorbidities before enrollment in WHI, we conducted sensitivity analyses by rerunning our analyses among participants without these self-reported comorbidities at OsteoPerio baseline (30–32).
Furthermore, given the potential for participants to alter their dietary habits between the baseline and 5-year follow-up visit, we performed a sensitivity analysis to evaluate the prospective associations after excluding individuals who shifted by more than one quartile from the average baseline dietary score to the 5-year dietary pattern score.
RESULTS
The baseline analytic sample consist of 1,197 (mean age 66.7 years) women, after excluding participants with incomplete data for diet, PD measures, or key covariates.
Mean and SD of each dietary pattern scores were described according to participant baseline characteristics (Table 1). The overall mean (±SD) for each score was 68.6 (±9.6) for HEI, 53.7 (±9.7) for AHEI, 24.4 (±4.8) for DASH, and 4.3 (±1.9) for aMed. Women who were older, better educated, had higher nSES, lower BMIs, and greater MET-hours/week of recreational physical activity had better mean adherence to dietary patterns than their comparison groups. Former smokers, current users of hormone therapy, more frequent flossers, more frequent brushers, and those attending dental visits more frequently had better mean adherence to the dietary patterns than their comparison groups. Women who had ever versus never been diagnosed with osteoporosis had a higher mean AHEI score. Women who had ever versus never been diagnosed or treated for hypertension or high blood pressure had a lower mean HEI and AHEI scores, and women who had ever been diagnosed or treated for heart disease, had a higher DASH score. Women in the 2nd tertile for whole mouth mean ACH and CAL have higher mean dietary pattern scores compared to those in tertiles 1 or 3. Women in the 3rd tertile of PPD had lower mean aMed scores, and women in the 3rd tertile for %BOP had lower means for all dietary pattern scores, compared to those in tertiles 1 or 2. Furthermore, women who reported losing one or more teeth due to PD were more likely to have lower mean DASH and aMed scores compared to those who had not lost any teeth. Baseline PD measures by characteristics are presented in Supplementary Table 1.
Table 1.
Description of Dietary Pattern Scores at OsteoPerio Baseline by Participants Characteristics
| Sub-groups | N (%) | Mean±SD |
|||
|---|---|---|---|---|---|
| HEI | AHEI | DASH | aMed | ||
| Overall | 1197 (100.0) | 68.6±9.6 | 53.7±9.7 | 24.4±4.8 | 4.3±1.9 |
| Age at Baseline | |||||
| <60 | 228 (19.0) | 67.1±10.3 | 52.7±10.0 | 23.8±4.8 | 4.3±1.9 |
| 60–70 | 554 (46.3) | 68.3±9.5 | 53.5±9.6 | 24.0±4.7 | 4.1±1.9 |
| >=70 | 415 (34.7) | 69.8±9.2 | 54.6±9.7 | 25.2±4.8 | 4.5±1.9 |
| Pa | 0.002** | 0.047* | <0.001*** | 0.027* | |
| Race/Ethnicity | |||||
| American Indian or Alaskan Native | 4 (0.3) | 63.9±14.1 | 48.5±11.5 | 22.0±5.2 | 3.5±1.3 |
| Asian or Pacific Islander | 3 (0.3) | 73.4±10.9 | 48.5±10.6 | 21.3±7.5 | 3.3±1.2 |
| Black or African-American | 17 (1.4) | 68.6±11.4 | 52.0±11.4 | 23.2±5.5 | 4.1±1.7 |
| Hispanic/Latino | 4 (0.3) | 65.6±10.3 | 53.5±10.5 | 25.4±8.6 | 4.3±2.9 |
| White (not of Hispanic origin) | 1169 (97.7) | 68.6±9.5 | 53.8±9.7 | 24.4±4.8 | 4.3±1.9 |
| Pa | 0.711 | 0.626 | 0.478 | 0.819 | |
| Education | |||||
| High School | 252 (21.1) | 66.6±10.0 | 51.4±9.7 | 22.9±4.7 | 3.7±1.8 |
| College | 523 (43.7) | 68.2±9.5 | 53.3±9.8 | 24.3±4.8 | 4.2±1.9 |
| Post-graduate | 422 (35.3) | 70.2±9.1 | 55.6±9.3 | 25.5±4.7 | 4.7±1.8 |
| Pa | <0.001*** | <0.001*** | <0.001*** | <0.001*** | |
| Neighborhood SESc | |||||
| <75 | 352 (29.4) | 68.2±9.8 | 52.7±9.7 | 24.1±4.9 | 4.2±1.9 |
| >=75 | 845 (70.6) | 68.7±9.5 | 54.1±9.7 | 24.5±4.7 | 4.3±1.8 |
| Pb | 0.017* | 0.156 | 0.333 | 0.592 | |
| Body Mass Index (kg/m2) | |||||
| Underweight/ Normal (<25) | 533 (44.5) | 69.9±9.7 | 55.2±9.6 | 25.0±4.9 | 4.4±1.8 |
| Overweight (25–29.9) | 414 (34.6) | 68.7±9.0 | 53.5±9.8 | 24.4±4.6 | 4.2±1.9 |
| Obese (≥30) | 250 (20.9) | 65.5±9.7 | 50.8±9.3 | 23.1±4.8 | 3.9±1.8 |
| Pa | <0.001*** | <0.001*** | <0.001*** | 0.001** | |
| Total MET Hours | |||||
| 0 | 170 (14.2) | 63.6±10.4 | 49.2±9.6 | 22.3±4.5 | 3.4±1.8 |
| <12.5 | 484 (40.4) | 67.7±9.3 | 52.1±9.4 | 23.8±4.8 | 4.1±1.9 |
| >=12.5 | 543 (45.4) | 70.9±8.9 | 56.6±9.2 | 25.6±4.6 | 4.7±1.8 |
| Pa | <0.001*** | <0.001*** | <0.001*** | <0.001*** | |
| Smoking Status | |||||
| Never | 625 (52.2) | 68.3±9.7 | 53.1±9.9 | 24.3±4.8 | 4.2±1.9 |
| Former | 537 (44.9) | 69.5±9.2 | 55.0±9.3 | 24.8±4.7 | 4.4±1.8 |
| Current | 35 (2.9) | 59.1±9.0 | 45.9±9.7 | 20.4±4.0 | 3.1±1.5 |
| Pa | <0.001*** | <0.001*** | <0.001*** | <0.001*** | |
| Hormone Therapya | |||||
| Never | 385 (32.2) | 67.8±10.0 | 52.7±9.7 | 23.9±4.7 | 4.1±1.8 |
| Former | 248 (20.7) | 68.1±10.0 | 53.0±10.3 | 24.1±5.1 | 4.2±1.9 |
| Current | 564 (47.1) | 69.3±9.0 | 54.7±9.4 | 24.9±4.7 | 4.4±1.8 |
| Pa | 0.045* | 0.004** | 0.006** | 0.156 | |
| Frequency of Flossing | |||||
| Not every week | 220 (18.4) | 66.1±9.5 | 51.6±10.0 | 23.4±4.6 | 3.9±1.8 |
| Once a week | 111 (9.3) | 68.2±9.3 | 51.5±9.3 | 23.9±4.9 | 4.0±1.8 |
| More than once a week | 342 (28.6) | 67.6±9.4 | 53.3±9.9 | 23.8±4.8 | 4.2±1.9 |
| Everyday | 524 (43.8) | 70.3±9.5 | 55.3±9.4 | 25.3±4.7 | 4.5±1.9 |
| Pa | <0.001*** | <0.001*** | <0.001*** | <0.001*** | |
| Frequency of Brushing Teeth | |||||
| Once a day or less | 263 (22.0) | 66.6±9.8 | 51.9±9.7 | 23.5±4.7 | 4.0±1.8 |
| Twice a day | 671 (56.1) | 68.9±9.3 | 53.9±9.7 | 24.4±4.8 | 4.3±1.9 |
| More than twice a day | 263 (22.0) | 69.7±9.8 | 54.9±9.6 | 25.2±4.6 | 4.5±1.9 |
| Pa | 0.001** | 0.002** | <0.001*** | 0.005** | |
| Frequency of Visiting Dentist | |||||
| Only with problem /never | 95 (7.9) | 63.8±10.9 | 50.9±10.1 | 22.9±4.7 | 3.8±1.8 |
| Once a year | 172 (14.4) | 68.0±9.6 | 52.7±10.1 | 23.9±4.7 | 4.1±1.8 |
| More than once a year | 930 (77.7) | 69.2±9.3 | 54.2±9.6 | 24.6±4.8 | 4.3±1.9 |
| Pa | <0.001*** | 0.002** | 0.001** | 0.017* | |
| Self-Reported as Having or Treated for Diabetes | |||||
| No | 1143 (95.5) | 68.5±9.6 | 53.7±9.8 | 24.4±4.8 | 4.3±1.9 |
| Yes | 54 (4.5) | 69.2±9.5 | 53.7±8.6 | 24.8±4.5 | 4.2±1.6 |
| Pb | 0.968 | 0.503 | 0.626 | 0.918 | |
| Self-Reported Ever Diagnosed with Osteoporosis | |||||
| No | 1032 (86.2) | 68.5±9.5 | 53.6±9.8 | 24.3±4.8 | 4.2±1.9 |
| Yes | 163 (13.6) | 69.2±9.9 | 54.5±9.2 | 25.2±5.0 | 4.5±1.8 |
| Pb | 0.247 | 0.018* | 0.397 | 0.078 | |
| Self-Reported Diagnosed or Treated for Hypertension or High Blood Pressure | |||||
| No | 825 (68.9) | 68.8±9.7 | 54.2±9.7 | 24.6±4.8 | 4.3±1.9 |
| Yes | 372 (31.1) | 68.0±9.4 | 52.6±9.6 | 24.0±4.7 | 4.2±1.8 |
| Pb | 0.008** | 0.048* | 0.191 | 0.567 | |
| Self-Reported Diagnosed or Treated for Heart Disease | |||||
| No | 1080 (90.2) | 68.4±9.5 | 53.6±9.7 | 24.3±4.8 | 4.2±1.9 |
| Yes | 117 (9.8) | 70.4±9.9 | 54.8±9.8 | 25.0±4.5 | 4.4±1.8 |
| Pb | 0.190 | 0.120 | 0.027* | 0.302 | |
| Whole Mouth Mean Alveolar Crest Height (ACH) | |||||
| Tertile 1 (1.08–2.07 mm) | 406 (33.9) | 67.9±9.5 | 53.4±9.7 | 24.0±4.5 | 4.2±1.9 |
| Tertile 2 (2.08–2.62 mm) | 405 (33.8) | 69.7±9.3 | 54.6±10.3 | 25.1±4.8 | 4.5±1.9 |
| Tertile 3 (2.63–8.90 mm) | 386 (32.2) | 68.2±9.9 | 53.1±9.1 | 24.1±5.0 | 4.1±1.8 |
| Pa | 0.017* | 0.082 | 0.001** | 0.014* | |
| Whole Mouth Mean Clinical Attachment Level (CAL) | |||||
| Tertile 1 (0.85–2.06 mm) | 395 (33.0) | 68.5±9.4 | 53.9±10.0 | 24.4±4.6 | 4.4±1.8 |
| Tertile 2 (2.06–2.51 mm) | 405 (33.8) | 69.4±9.3 | 54.4±9.5 | 24.8±4.8 | 4.4±1.9 |
| Tertile 3 (2.51–7.80 mm) | 397 (33.2) | 67.8±10.0 | 52.9±9.7 | 24.0±5.0 | 4.1±1.9 |
| Pa | 0.053 | 0.097 | 0.067 | 0.029* | |
| Whole Mouth Mean Pocket Probing Depth (PPD) | |||||
| Tertile 1 (1.24–1.99 mm) | 392 (32.7) | 68.7±9.1 | 54.3±9.7 | 24.5±4.5 | 4.4±1.8 |
| Tertile 2 (1.99–2.31 mm) | 405 (33.8) | 68.9±9.9 | 53.6±9.8 | 24.5±4.9 | 4.4±1.9 |
| Tertile 3 (2.31–4.28 mm) | 400 (33.4) | 68.0±9.7 | 53.3±9.7 | 24.1±5.0 | 4.1±1.9 |
| Pa | 0.385 | 0.333 | 0.438 | 0.046* | |
| Percentage of Sites that Bled on Probing (%BOP) | |||||
| Tertile 1 (0–19%) | 387 (32.3) | 69.7±9.5 | 54.8±9.8 | 24.9±4.7 | 4.5±1.8 |
| Tertile 2 (20–40%) | 413 (34.5) | 68.8±9.3 | 53.8±9.6 | 24.2±4.8 | 4.3±1.8 |
| Tertile 3 (41–100%) | 397 (33.2) | 67.3±9.8 | 52.6±9.7 | 24.1±4.8 | 4.0±1.9 |
| Pa | 0.001** | 0.009** | 0.026* | 0.002** | |
| Number of Missing Teeth Self-reported as lost due to Periodontal Disease at OsteoPerio Baseline | |||||
| 0 | 1096 (91.6) | 68.8±9.5 | 53.8±9.7 | 24.5±4.7 | 4.3±1.9 |
| 1 | 25 (2.1) | 66.0±11.9 | 50.1±11.0 | 22.6±5.7 | 3.4±1.9 |
| >=2 | 76 (6.3) | 66.6±10.0 | 53.1±10.0 | 23.5±5.2 | 4.0±1.8 |
| Pa | 0.069 | 0.142 | 0.039* | 0.029* | |
ANOVA F-tests were used to examine the difference of dietary pattern scores by characteristics with >=3 categories.
T-tests were used to examine the difference of dietary pattern scores by characteristics with 2 categories.
SES: Socioeconomic Status
P<0.05
P<0.01
P<0.001
Cross-sectional associations
No associations between dietary patterns and ACH or PPD were found cross-sectionally (Table 2). However, a 10-point higher HEI score and a 1-point higher aMed score were associated with significantly lower mean CAL (β-coefficient = −0.056 mm and −0.029 mm, respectively; P<0.05 each) and significantly lower %BOP (β-coefficient = −1.526 and −0.836, respectively; P<0.05 each). The HEI, DASH, and aMed scores were inversely associated with lower odds (ORs 0.77 to 0.86; all P<0.05) of having teeth missing due to PD at baseline. Compared to participants free from OsteoPerio defined PD, each 10-point higher AHEI was associated with 0.82 times lower odds (P<0.05) of severe PD. Compared to participants free from CDC/AAP defined PD, AHEI and aMed were associated with lower odds (OR=0.81 and 0.88, respectively; both P<0.05) of severe PD.
Table 2.
Cross-Sectional Associationsa between Dietary Pattern Scores and Periodontal Measures at Baseline (N=1,197)
| HEI (per 10 pts) | AHEI (per 10 pts) | DASH (per 4 pts) | aMed (per 1 pt) | |
|---|---|---|---|---|
|
|
||||
| β (95% CI) |
||||
| ACH (mm) | −0.027 (−0.073, 0.018) | −0.034 (−0.078, 0.010) | −0.019 (−0.054, 0.017) | −0.011 (−0.034, 0.013) |
| CAL (mm) | −0.056 (−0.099, −0.014) | −0.038 (−0.080, 0.003) | −0.025 (−0.059, 0.008) | −0.029 (−0.051, −0.006) |
| PPD (mm) | −0.009 (−0.035, 0.016) | −0.010 (−0.034, 0.015) | 0.001 (−0.019, 0.021) | −0.009 (−0.022, 0.004) |
| %BOP | −1.526 (−2.966, −0.086) | −1.059 (−2.455, 0.337) | −0.107 (−1.246, 1.032) | −0.836 (−1.581, −0.090) |
|
| ||||
| OR (95%) | ||||
|
| ||||
| Missing Teeth Self-reported as lost due to Periodontal Disease at OsteoPerio Baseline | ||||
| None (n=101) | Ref. | Ref. | Ref. | Ref. |
| Any (n=1096) | 0.77 (0.61, 0.96) | 0.83 (0.66, 1.05) | 0.79 (0.65, 0.95) | 0.86 (0.76, 0.98) |
| OsteoPerio Defined Periodontal Disease | ||||
| None (n=290) | Ref. | Ref. | Ref. | Ref. |
| Mild/Moderate (n=592) | 1.10 (0.92, 1.30) | 0.94 (0.80, 1.11) | 1.05 (0.92, 1.20) | 1.03 (0.95, 1.13) |
| Severe (n=315) | 0.88 (0.72, 1.07) | 0.82 (0.68, 1.00) | 0.90 (0.77, 1.05) | 0.92 (0.83, 1.02) |
| CDC/AAP Defined Periodontal Disease | ||||
| None (281) | Ref. | Ref. | Ref. | Ref. |
| Mild/Moderate (n=730) | 0.98 (0.83, 1.15) | 0.98 (0.84, 1.15) | 0.94 (0.83, 1.06) | 0.98 (0.90, 1.06) |
| Severe (n=186) | 0.84 (0.67, 1.04) | 0.81 (0.65, 1.00) | 0.88 (0.74, 1.05) | 0.88 (0.78, 0.98) |
Models adjusted for age, race, physical activity, neighborhood SES, smoking, education level, hormone therapy, frequency of brushing teeth, frequency of flossing, frequency of visiting dentists and total calorie intake.
Bold associations are statistically significant at an α-level of 0.05.
Prospective associations
Prospective analyses consist of 894 women with complete data at the follow-up visit. Characteristics of OsteoPerio participants included and excluded for these analyses are shown in Supplemental Table 2. During the 5 years of follow-up, a 10-point higher AHEI was positively associated with a 0.038 mm increment of ACH (Table 3). No statistically significant associations were observed with CAL, PD progression, or new tooth loss (Table 3).
Table 3.
Prospective Associations between Baseline Dietary Pattern Scores and Periodontal Measures after 5 Years of Follow-up (N=894)
| HEI (per 10 pts) | AHEI (per 10 pts) | DASH (per 4 pts) | AMED (per 1 pt) | |
|---|---|---|---|---|
|
|
||||
| β (95% CI) |
||||
| ACH (mm) a | 0.004 (−0.032, 0.041) | 0.038 (0.003, 0.073) | 0.021 (−0.007, 0.050) | 0.016 (−0.003, 0.035) |
| CAL (mm) a | −0.018 (−0.058, 0.021) | 0.007 (−0.031, 0.044) | −0.014 (−0.044, 0.017) | −0.001 (−0.021, 0.020) |
|
| ||||
| OR (95% CI) | ||||
|
|
||||
| Periodontal Disease Progression [Yes (n=191) Vs. No (n=701)] b | ||||
| No (n=701) | Ref. | Ref. | Ref. | Ref. |
| Yes (n=191) | 0.97 (0.80, 1.17) | 1.11 (0.92, 1.33) | 0.99 (0.86, 1.15) | 1.06 (0.96, 1.17) |
| Incident Tooth Loss Due to Periodontal Disease [Yes (n=38) Vs. No (n=856)] b | ||||
| No (n=856) | Ref. | Ref. | Ref. | Ref. |
| Yes (n=38) | 0.86 (0.59, 1.27) | 0.84 (0.57, 1.25) | 1.00 (0.73, 1.37) | 0.86 (0.70, 1.06) |
Models adjusted for age, race, physical activity, neighborhood SES, smoking, education level, hormone therapy, frequency of brushing teeth, frequency of flossing, frequency of visiting dentists, total calorie intake, and for the respective baseline periodontal disease measures (i.e., ACH or CAL).
Models adjusted for age, race, physical activity, neighborhood SES, smoking, education level, hormone therapy, frequency of brushing teeth, frequency of flossing, frequency of visiting dentists, and total calories intake.
Bold associations are statistically significant at an α-level of 0.05.
Sensitivity Analyses
After imputing ≥1 mm loss in ACH for each tooth lost due to PD over the course of the 5-year follow-up period, the positive association between AHEI and ACH became null (e.g., β (95% CI) =0.030 (95% CI: −0.002, 0.062) for 1 mm imputation of ACH loss). Other prospective associations with ACH remained null. Furthermore, the associations between all dietary patterns and CAL remained null with imputations of ≥1 mm (details not provided).
After excluding individuals with any comorbidities at baseline, the cross-sectional associations between higher dietary pattern scores and better periodontal measures at baseline were strengthened, and significant inverse associations were observed between all dietary pattern scores and ACH. No significant association was found in the prospective 5-year analysis; but the beta-coefficients and OR’s moved toward the negative direction (Supplementary Table 3).
After removing individuals with unstable diet, conclusions from the prospective associations remained the same (results not shown).
DISCUSSION
Using data from the OsteoPerio cohort of women aged 53–81 at baseline, we examined the cross-sectional and 5-year prospective associations of four widely used score-based dietary patterns (HEI, AHEI, aMed, and DASH) with periodontal measures and tooth loss due to PD cross-sectionally, and during 5 years of follow-up. Cross-sectional analyses showed that higher (healthier) dietary pattern scores were associated with better clinical periodontal measures at baseline; prospective analyses showed that higher baseline dietary pattern scores were associated with greater progression (worsening) of ACH during the follow-up. However, these associations no longer existed after accounting for tooth loss due to PD during follow-up or removing participants with pre-existing PD-related comorbidities.
A number of studies have examined associations between dietary patterns and PD, with most conducted in cross-sectional studies. Higher adherence to the HEI (6), AHEI (7), DASH diet (8), and the Mediterranean diet (9), were found to be cross-sectionally associated with lower odds of PD. A handful of studies, with diets scored according to their inflammatory nature, have examined their associations with PD. Li A et al. found that consuming a pro‐inflammatory diet was associated with an increased odds of periodontitis (10), while Lieske B et al reported that an anti-inflammatory diet was associated with lower odds of periodontitis (11). Our findings support a protective association between dietary patterns and periodontal disease in our cross-sectional associations.
The prospective association between dietary patterns and PD was investigated in recent years. Alhassani et al. found that a Western diet and an inflammatory diet were associated with a higher risk of self-reported periodontitis among obese men. No statistically significant association was found in men who were not obese (12,13). Additionally, Jauhiainen et al. found that diet quality assessed by Baltic Sea Diet Score (BSDS) and the Recommended Finnish Diet Score (RFDS) was associated with less teeth with deepened gingival pockets over 11 years (14).
In our prospective analysis, we observed that participants with better compared to worse adherence to healthy dietary patterns tended to have greater progression of ACH during the 5 years of follow-up, indicating more progression of alveolar bone loss. These findings contradicted our hypothesis, but there are two possible explanations that we should consider:
First, when severe PD leads to tooth loss, the overall whole mouth mean PD measures will improve since poor sites were removed from mean measurements as a result of tooth loss. To rule out the impact of tooth loss in our prospective analyses, we conducted a sensitivity analysis to determine the impact of missing teeth due to PD at follow-up. When we assumed (imputed) a progression of 1 mm or greater loss of bone above the baseline ACH values, the association between dietary patterns and loss in ACH was attenuated and became null. It is possible that unaccounted tooth loss likely contributed to the positive association we observed, and future studies need to account for loss of teeth.
Furthermore, women with self-reported comorbidities of diabetes, osteoporosis, hypertension, and cardiovascular disease at OsteoPerio baseline may be at increased risk for PD progression (30–32). These women may have reported a dietary intake at WHI-OS and OsteoPerio baseline that did not reflect their usual dietary intake during their earlier adult years (prior to age 50). A diagnosis with one of these chronic diseases may have led to diet change prior to the baseline dietary assessment. Therefore, we may be misclassifying women’s long-term dietary exposures, especially if the critical window of exposure for dietary intake on PD spans a longer period. For this reason, we reran our analyses after excluding women with these comorbidities at OsteoPerio baseline. As assumed, all the associations moved toward the negative direction, and the positive significant association observed between dietary pattern score and ACH progression no longer remained.
Strengths
A strength of this study was its set of comprehensive assessments on periodontal health, including chronic indicators (ACH, CAL, and the number of teeth lost due to PD), which tracked the cumulative PD progression, and acute indicators (%BOP and PPD), which can be reversed by good oral hygiene practices (33). The dietary data was collected through FFQs which allowed us to assess habitual dietary intake patterns. Moreover, we had more than one measure of diet over time which allowed us to average the WHI-OS and OsteoPerio baseline dietary pattern scores, estimating participant’s dietary habits over three years leading up to the OsteoPerio baseline visit. This reflects a longer period of usual intake than one FFQ alone. In this study, we used hypothesis-based dietary pattern scores instead of data-driven dietary patterns to examine the association with PD so that the findings could be compared with other studies and are reproducible. Lastly, we conducted this study among postmenopausal women, who are at high risk of PD as aging and decreased estrogen concentrations are both risk factors. Therefore, the predictive and preventive significance of dietary patterns for PD is critical for this population from a public health perspective.
Limitations
Our study sample consisted of women who were highly educated, practiced good oral hygiene, and were nonsmokers, so they may be a group with a relatively good diet and less likely to show clinically relevant differences in this cohort. Besides, FFQs are prone to measurement error from limitations in the respondents’ memory and from seasonal variation not captured in a 3 month dietary recall (34,35). It is important to recognize that the loss of posterior teeth can lead to difficulties in chewing certain types of healthy foods, such as fruits and vegetables. Most of the tooth loss in our sample pertains to molars (87% of participants with tooth loss at baseline lost at least 1 molar), which may raise the possibility of reverse causality in the cross-sectional analyses. As in all observational studies, the possibility of residual confounding exists. Additionally, as our sample consists of primarily non-Hispanic white postmenopausal white women, samples including young adults, men, and racial minorities may find different results due to differences in the distribution of diet pattern scores as well as in rate and severity of progression of PD over time. Future studies may be conducted among individuals with more diverse dietary behaviors.
In conclusion, data from OsteoPerio cohort study consisting of postmenopausal women indicated that better adherence to HEI, AHEI, DASH, and aMed is cross-sectionally associated with better periodontal health. The AHEI was found to be associated with greater progression of oral bone loss in our prospective analysis, however after accounting for tooth loss due to PD or pre-existing comorbidities, the significant positive association no longer existed.
Supplementary Material
CLINICAL RELEVANCE.
Scientific Rationale for Study
While limited studies have prospectively examined the association between dietary patterns and periodontal disease (PD), this study is the first to utilize multiple clinical measures of PD reflecting different stages of the disease, the first study with a measure of alveolar bone loss and expands the research on older women.
Principal Findings
Cross-sectionally, healthier dietary patterns were associated with better PD measures, but prospective associations after 5 years of follow-up did not support this observation.
Practical Implications
Among postmenopausal women, consumption of healthier dietary patterns does not appear to offer protective benefits against the development or progression of PD.
ACKNOWLEDGEMENTS
YY, MJL, JWW, CAA, and AEM conceptualized this study; YY and KMH analyzed the data; YY and AEM wrote the first draft; All authors reviewed and commented on subsequent drafts of the manuscript.
We acknowledge our collaborators, OsteoPerio study group, and all participants who joined in the WHI and OsteoPerio cohort.
FUNDING
The Women’s Health Initiative program is funded by the National Heart, Lung, and Blood Institute (NHLBI), National Institutes of Health (NIH), U.S. Department of Health and Human Services through contracts HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C; Other funding sources include NIH/NHLBI contract N01WH32122; NIH grants R01DE024523 and R01DE13505 from the National Institute of Dental and Craniofacial Research (NIDCR) and funding from the Department of Defense (DAMD179616319).
Footnotes
CONFLICT OF INTEREST:
The authors declared no conflict of interest.
DATA AVAILABILITY
Data described in the manuscript and analytic code will be made available upon request.
REFERENCE
- 1.Eke PI, Borgnakke WS, Genco RJ. Recent epidemiologic trends in periodontitis in the USA. Periodontology 2000. 2020;82(1):257–67. [DOI] [PubMed] [Google Scholar]
- 2.Chapple ILC, Genco R, Workshop working group 2 of the joint E. Diabetes and periodontal diseases: consensus report of the Joint EFP/AAP Workshop on Periodontitis and Systemic Diseases. Journal of Clinical Periodontology. 2013;40(s14):S106–12. [DOI] [PubMed] [Google Scholar]
- 3.Carrizales-Sepúlveda EF, Ordaz-Farías A, Vera-Pineda R, Flores-Ramírez R. Periodontal Disease, Systemic Inflammation and the Risk of Cardiovascular Disease. Heart Lung Circ 2018. Nov;27(11):1327–34. [DOI] [PubMed] [Google Scholar]
- 4.Daalderop LA, Wieland BV, Tomsin K, Reyes L, Kramer BW, Vanterpool SF, et al. Periodontal Disease and Pregnancy Outcomes: Overview of Systematic Reviews. JDR Clin Trans Res 2018. Jan;3(1):10–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kelly N, Winning L, Irwin C, Lundy FT, Linden D, McGarvey L, et al. Periodontal status and chronic obstructive pulmonary disease (COPD) exacerbations: a systematic review. BMC Oral Health. 2021. Sep 3;21(1):425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Li X yu, Wen M zhe, Xu Y hua, Shen Y chen, Yang X tao. The association of healthy eating index with periodontitis in NHANES 2013–2014. Front Nutr 2022. Aug 9;9:968073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Salazar CR, Laniado N, Mossavar-Rahmani Y, Borrell LN, Qi Q, Sotres-Alvarez D, et al. Better-quality diet is associated with lower odds of severe periodontitis in US Hispanics/Latinos. J Clin Periodontol 2018/05/21 ed. 2018. Jul;45(7):780–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Altun E, Walther C, Borof K, Petersen E, Lieske B, Kasapoudis D, et al. Association between Dietary Pattern and Periodontitis—A Cross-Sectional Study. Nutrients. 2021. Nov 21;13(11):4167. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sáenz-Ravello G, Matamala L, Cisternas P, Gamonal J, Hernández P, Santos NCD, et al. Association between the Mediterranean Diet Index and self-reported Gingival Health Status Indicators in a population of Chilean adults: a cross-sectional study. J Appl Oral Sci 2023;31:e20230100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Li A, Chen Y, Schuller AA, van der Sluis LWM, Tjakkes GHE. Dietary inflammatory potential is associated with poor periodontal health: A population-based study. J Clin Periodontol 2021. Jul;48(7):907–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lieske B, Moszka N, Borof K, Petersen EL, Jagemann B, Ebinghaus M, et al. Association between an Anti-Inflammatory Dietary Score and Periodontitis-Evidence from the Population-Based Hamburg City Health Study. Nutrients. 2023. Jul 21;15(14):3235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Alhassani AA, Hu FB, Li Y, Rosner BA, Willett WC, Joshipura KJ. The associations between major dietary patterns and risk of periodontitis. J Clin Periodontol 2021. Jan;48(1):2–13. [DOI] [PubMed] [Google Scholar]
- 13.Alhassani AA, Hu FB, Rosner BA, Tabung FK, Willett WC, Joshipura KJ. The relationship between inflammatory dietary pattern and incidence of periodontitis. Br J Nutr 2021. Dec 14;126(11):1698–708. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Jauhiainen LM, Ylöstalo PV, Knuuttila M, Männistö S, Kanerva N, Suominen AL. Poor diet predicts periodontal disease development in 11‐year follow‐up study. Community Dent Oral Epidemiol 2020. Apr;48(2):143–51. [DOI] [PubMed] [Google Scholar]
- 15.Bartha V, Exner L, Schweikert D, Woelber JP, Vach K, Meyer AL, et al. Effect of the Mediterranean diet on gingivitis: A randomized controlled trial. J Clin Periodontol 2022. Feb;49(2):111–22. [DOI] [PubMed] [Google Scholar]
- 16.Banack HR, Genco RJ, LaMonte MJ, Millen AE, Buck MJ, Sun Y, et al. Cohort profile: the Buffalo OsteoPerio microbiome prospective cohort study. BMJ Open. 2018. Dec 4;8(12):e024263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement characteristics of the Women’s Health Initiative food frequency questionnaire. Ann Epidemiol 1999. Apr;9(3):178–87. [DOI] [PubMed] [Google Scholar]
- 18.Wactawski-Wende J, LaMonte MJ, Hovey K, Banack H. The Buffalo OsteoPerio Studies: Summary of our findings and the unique contributions of Robert J. Genco, DDS, PhD. Curr Oral Health Rep 2020. Mar;7(1):29–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Bowman Shanthy A., Friday James E., Moshfegh Alanna J.. MyPyramid Equivalents Database, 2.0 for USDA Survey Foods, 2003–2004: Documentation and User Guide [Internet]. Food Surveys Research Group. Beltsville Human Nutrition Research Center, Agricultural Research Service, U.S. Department of Agriculture, Beltsville, MD.; 2008. Available from: http://www.ars.usda.gov/ba/bhnrc/fsrg [Google Scholar]
- 20.Krebs-Smith SM, Pannucci TE, Subar AF, Kirkpatrick SI, Lerman JL, Tooze JA, et al. Update of the Healthy Eating Index: HEI-2015. J Acad Nutr Diet 2018. Sep;118(9):1591–602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.McCullough ML, Feskanich D, Stampfer MJ, Giovannucci EL, Rimm EB, Hu FB, et al. Diet quality and major chronic disease risk in men and women: moving toward improved dietary guidance. Am J Clin Nutr 2002. Dec;76(6):1261–71. [DOI] [PubMed] [Google Scholar]
- 22.Sacks FM, Obarzanek E, Windhauser MM, Svetkey LP, Vollmer WM, McCullough M, et al. Rationale and design of the Dietary Approaches to Stop Hypertension trial (DASH): A multicenter controlled-feeding study of dietary patterns to lower blood pressure. Annals of Epidemiology. 1995. Mar 1;5(2):108–18. [DOI] [PubMed] [Google Scholar]
- 23.Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med 2003. Jun 26;348(26):2599–608. [DOI] [PubMed] [Google Scholar]
- 24.Fung TT, Hu FB, McCullough ML, Newby PK, Willett WC, Holmes MD. Diet quality is associated with the risk of estrogen receptor-negative breast cancer in postmenopausal women. J Nutr 2006. Feb;136(2):466–72. [DOI] [PubMed] [Google Scholar]
- 25.Hausmann E, Allen K, Carpio L, Christersson LA, Clerehugh V. Computerized methodology for detection of alveolar crestal bone loss from serial intraoral radiographs. J Periodontol 1992. Aug;63(8):657–62. [DOI] [PubMed] [Google Scholar]
- 26.Chung YM, Jeong SN. Analysis of periodontal attachment loss in relation to root form abnormalities. J Periodontal Implant Sci 2013. Dec;43(6):276–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Page RC, Eke PI. Case definitions for use in population-based surveillance of periodontitis. J Periodontol 2007. Jul;78(7 Suppl):1387–99. [DOI] [PubMed] [Google Scholar]
- 28.Eke PI, Dye BA, Wei L, Slade GD, Thornton-Evans GO, Borgnakke WS, et al. Update on Prevalence of Periodontitis in Adults in the United States: NHANES 2009 to 2012. J Periodontol 2015. May;86(5):611–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Genco RJ, Borgnakke WS. Risk factors for periodontal disease. Periodontology 2000. 2013;62(1):59–94. [DOI] [PubMed] [Google Scholar]
- 30.Czesnikiewicz-Guzik M, Osmenda G, Siedlinski M, Nosalski R, Pelka P, Nowakowski D, et al. Causal association between periodontitis and hypertension: evidence from Mendelian randomization and a randomized controlled trial of non-surgical periodontal therapy. Eur Heart J 2019. Nov 1;40(42):3459–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liccardo D, Cannavo A, Spagnuolo G, Ferrara N, Cittadini A, Rengo C, et al. Periodontal Disease: A Risk Factor for Diabetes and Cardiovascular Disease. Int J Mol Sci 2019. Mar 20;20(6):1414. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Megson E, Kapellas K, Bartold PM. Relationship between periodontal disease and osteoporosis. Int J Evid Based Healthc 2010. Sep;8(3):129–39. [DOI] [PubMed] [Google Scholar]
- 33.Kotsakis GA, Lian Q, Ioannou AL, Michalowicz BS, John MT, Chu H. A network meta-analysis of interproximal oral hygiene methods in the reduction of clinical indices of inflammation. J Periodontol 2018. May;89(5):558–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Horner NK, Patterson RE, Neuhouser ML, Lampe JW, Beresford SA, Prentice RL. Participant characteristics associated with errors in self-reported energy intake from the Women’s Health Initiative food-frequency questionnaire. Am J Clin Nutr 2002. Oct;76(4):766–73. [DOI] [PubMed] [Google Scholar]
- 35.Yang JJ, Park SK, Lim HS, Ko KP, Ahn Y, Ahn YO. [Seasonal variation of food intake in food frequency questionnaire among workers in a nuclear power plant]. J Prev Med Public Health. 2007. May;40(3):239–48. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the manuscript and analytic code will be made available upon request.