Diet Quality and Periodontal Disease: Results from The Oral Infections, Glucose Intolerance and Insulin Resistance Study (ORIGINS)

Abstract

Aims:

This study examined the cross-sectional association between diet quality and periodontal disease.

Methods:

In the Oral Infections, Glucose Intolerance and Insulin Resistance Study (ORIGINS), 923 individuals completed the National Cancer Institute’s validated Diet History Questionnaire 1, from which the Alternative Healthy Eating Index (AHEI) scores and A Priori Diet Quality Scores (APDQS) were calculated. Mean probing depth (MPD), mean clinical attachment loss (MAL) and % of sites bleeding on probing (%BOP) were derived from full-mouth periodontal exams. Multivariable adjusted linear and logistic regression models assessed the associations between diet quality and MPD, MAL, %BOP, and the odds of periodontitis (defined via the CDC/AAP classification).

Results:

AHEI and APDQS were not associated with MPD, MAL, or periodontitis. While AHEI was also not associated with %BOP, the APDQS was associated with %BOP (p=0.03). Higher nut consumption was related to lower MPD (p=0.03) and periodontitis odds (p=0.03). Higher red meat consumption was associated with higher MPD (p=0.01) and %BOP (p=0.05). Higher trans-fatty acid consumption was also associated with increased %BOP (p=0.05).

Conclusion:

Overall diet quality scores were not associated with periodontal status. Future studies are necessary to replicate the associations observed in this study to minimize the risk of false discovery.

INTRODUCTION

It is well-established that dietary patterns are associated with chronic inflammatory diseases including cardiometabolic disease (Guasch-Ferre et al., 2017; Mueller & Appel, 2017). Periodontal disease is known to have inflammatory underpinnings; however, the influence of diet on this condition is poorly understood (Milward & Chapple, 2013). This association is of particular importance, given the prevalence of periodontal disease. Approximately 50–90% of the global adult population is affected by gingivitis, with 47% of adults aged 30 or older in the US having some form of periodontitis (P. Eke et al., 2015; Pihlstrom, Michalowicz, & Johnson, 2005). Furthermore, 5 to 10% of total health care expenditures in industrialized countries are attributed to dental diseases (Listl, Galloway, Mossey, & Marcenes, 2015; P. Moynihan & Petersen, 2004) and globally, lost economic productivity from severe periodontitis is estimated to cost $54 billion per year (Listl et al., 2015; P. Moynihan & Petersen, 2004). Given the expectation that periodontal disease prevalence will increase as populations age (Jepsen et al., 2017), a better understanding of prevention strategies, such as dietary behaviors, can help to minimize population disease burden.

Diet may play an important role in the progression, management, and prevention of periodontal disease (Nunn, 2003). Diets rich in whole grains with moderate energy intake were associated with reduced periodontitis risk among men in a prospective cohort study using validated semiquantitative food-frequency questionnaires (FFQs) (Merchant, Pitiphat, Franz, & Joshipura, 2006). In a cross-sectional study, US adults aged 30 years or older who were in the lowest quartile of dietary fiber consumption had increased odds of moderate to severe periodontitis compared to those in the highest quartile of fiber consumption (Nielsen, Trak-Fellermeier, Joshipura, & Dye, 2016). Lastly, a review examining the association between diet and oral disease concluded that diets that are high in fruits, vegetables, and whole grain starchy foods, and low in free sugars and fats have been associated with reduced periodontal disease (P. J. Moynihan, 2005).

In addition, certain dietary supplementations may play a role in periodontal health. It has been suggested that supplementation with vitamin C may be a dietary component that can improve periodontal health by promoting type 1 collagen synthesis, a major component of periodontal ligaments, by reducing inflammatory markers, such as interleukin-8 and TNF-α, and by reducing gingival inflammation (Shimabukuro et al., 2015; Tsutsumi et al., 2012). Adequate intake of calcium in the diet may influence bone remodeling, which is pivotal in the health of the periodontium (Varela-Lopez, Giampieri, Bullon, Battino, & Quiles, 2016b). However, there are many methodological limitations with current studies to interpret current data examining the association between vitamin supplementation, different food and vitamin consumption, and periodontal health status (Najeeb, Zafar, Khurshid, Zohaib, & Almas, 2016; Varela-Lopez et al., 2016b). Therefore, further analysis is required.

The American Dental Association (ADA) currently states that “the effect of nutritional status on the body’s immune response may modify factors affecting management of periodontal disease; however, the multifactorial nature of periodontal disease and nutritional status makes it difficult to determine such effects”. Therefore, the purpose of this study is to investigate the cross-sectional associations of overall diet quality (Alternative Healthy Eating Index (AHEI) and A Priori Diet Quality Scores (APDQS)) and specific dietary subcomponents with periodontal disease among participants in The Oral Infections, Glucose Intolerance and Insulin Resistance Study (ORIGINS). This will fill the gap in information regarding the association between periodontal disease and diet. With a clearer understanding of this association, dental practitioners may have a better ability to modulate and prevent periodontal disease in their patients.

METHODS

Study Population

ORIGINS is a prospective cohort study at the Columbia University Irving Medical Center in New York City (Demmer et al., 2015). Wave 1 of this study enrolled 300 participants from February 2011 to May 2013, and wave 2 enrolled 800 additional participants between January 2016 to January 2020. The inclusion criteria for ORIGINS has been described in previous studies (Demmer et al., 2015). Participants were included in the study if they were i) aged 20–55 years; ii) had no diabetes mellitus (type 1 or type 2) based on participant self-report of no previously diagnosed disease, HbA1c values <6.5% and fasting plasma glucose < 126 mg/dl; and iii) had no history of myocardial infarction, congestive heart failure, stroke, or chronic inflammatory conditions based on participant self-report. Among those enrolled, 281 and 782 participants completed a food frequency questionnaire (FFQ) at waves 1 and 2, respectively (Figure 1). Participants from waves 1 and 2 displayed similar demographic characteristics (Supplemental Table 14). After combining these waves for a total of 1063 participants, another 11 participants were excluded for missing baseline data, withdrawal of consent, or an error in their participant ID (Figure 1). Furthermore, 100 participants were removed from this study because of incomplete diet data creating outlier values and another 29 participants were excluded from the study because of missing periodontal data (n=8) and demographic data (n=21) to produce a total study population of 923 participants (Figure 1). Included and excluded participants displayed similar demographic characteristics.

Figure 1:

Study flow diagram

Dietary Assessment

Dietary data were collected using the National Cancer Institute’s Diet History Questionnaire 1 (DHQ-1) that queries frequency of consumption and portion size for 124 food items (Institute, 2016). This FFQ has been previously validated and found to provide valid nutrient intake estimates (Institute, 2016). The dietary data was subsequently operationalized using the AHEI and APDQS diet quality measures as previously described (Institute, 2016; Nettleton et al., 2008). Briefly, the AHEI was scored based on intake of eleven dietary components (fruits, vegetables, nuts, red meat, sugar sweetened beverages, omega-3 fatty acids, polyunsaturated fats, trans fats, alcohol, whole grains, and refined grains), which were summed for a total AHEI score ranging from 16.4 (indicating lower diet quality) to 83.0 (indicating higher diet quality); higher scores reflected hypothesized better diets such that scores for red meat, sugar sweetened beverages and refined grains reflected lower consumption. The APDQS (range: 27.0–96.0) was calculated based on intake of “positive foods” (i.e. those postulated to be associated with reduced cardiovascular disease risk), such as green vegetables, fruits, lean fish, low fat dairy products and whole grains; “neutral foods” (i.e. those irrelevant to cardiovascular disease risk), such as eggs, lean meat, shellfish and potatoes; and “negative foods” (i.e. those postulated to be associated with higher cardiovascular disease risk), such as fried potatoes, high fat processed meat, desserts, pastries, full fat dairy products and soft drinks (Nettleton et al., 2008).

Periodontal Examination

Each participant received a detailed periodontal examination by a trained examiner. All examiners were calibrated by Dr. Papapanou, a study investigator (Demmer et al., 2015). Intraexaminer reliability data for examiners for intraclass correlations was 0.97 and 0.94 between repeated measures of probing depth and attachment loss and mean absolute differences between repeated examinations 0.09 and 0.17mm, respectively (Demmer et al., 2015). Interexaminer reliability data revealed mean differences between repeat measurements for probing depth and attachment loss to be 0.5 (0.02) mm and 0.9 (0.04) mm, respectively among 1,118 sites (Demmer et al., 2015). Periodontal probing depth (PD) and attachment loss (AL) were measured at six sites per tooth, including mesial buccal, direct buccal, distal buccal, distal lingual, direct lingual, and mesial lingual sites with a UNC-15 periodontal probe manufactured by Hu-Friedy. Bleeding on probing was assessed up to 192 sites. Using site level data, periodontitis was defined according to the Centers for Disease Control and Prevention and the American Academy of Periodontology (CDC/AAP) recommendations (Page & Eke, 2007). Additionally, we defined the following variables from site level information: i) mean probing depth (MPD) was defined as the mean PD across all measured sites; ii) mean attachment loss (MAL) was defined as the mean AL across all measures sites; iii) percent of sites bleeding on probing (%BOP) was defined as the percent of measured sites that bled following probing.

Laboratory Measures

Blood was collected following an overnight fast. Plasma glucose, serum lipids, and hemoglobin A1c (HbA1c) from whole blood, were measured at baseline using standard methods (Demmer et al., 2017; Demmer et al., 2015) that were consistent between study waves.

Risk Factors

Systolic and diastolic blood pressures were measured in triplicate and the last two measurements were averaged. Body mass index (BMI) was calculated as weight in kilograms/height in meters². Questionnaires assessed information on age, sex (male, female), race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, Other), educational level (high school completion, college or vocational training, advanced degrees), cigarette smoking (never smoked, former smoker, current smoker and duration/intensity of smoking), leisure-time physical activity (categorized into 4 LTPA categories (Thai, Papapanou, Jacobs, Desvarieux, & Demmer, 2014)) and total caloric intake. Prediabetes was defined according to The American Diabetes Association (ADA) definition as follows: HbA1c ranging from 5.7%−6.4% or impaired fasting glucose (IFG) defined as fasting plasma glucose (FPG) ranging from 100 mg/dl to 125 mg/dl.

Statistical Analyses

Participant characteristics were described using means ± standard deviation (SD), frequencies and percentages. Exposure variables consisted of AHEI scores, APDQ scores, fruits, vegetables, nuts, red meat, sugar sweetened beverages, omega-3 fatty acids, polyunsaturated fats, trans fats, alcohol, whole grains, and refined grains consumption, and positive/negative APDQ scores. The outcome variables measured were mean probing depth (MPD), mean clinical attachment loss (MAL), percent of sites bleeding on probing (%BOP), and periodontitis. Covariables consisted of age, sex, race/ethnicity, education, smoking, BMI, total caloric intake, and prediabetes. Some of these covariates can also be considered mediators, such as BMI and prediabetes; we included these variables in a final model so readers can interpret models with and without these adjustments. Multivariable linear regression models were then used to regress full mouth MPD, MAL, and %BOP across quartiles of diet quality. Next, logistic regression models regressed the odds of periodontitis across quartiles of diet quality (using quartile 1 as the referent group). Periodontal disease was categorized as none/mild vs. moderate/severe (P. I. Eke, Page, Wei, Thornton-Evans, & Genco, 2012).

Three models were constructed for each diet quality score as follows: i) Model 1 (M1) = wave adjusted; ii) M2= M1+ age (years), gender (male, female), race/ethnicity (Hispanic, non-Hispanic black, non-Hispanic white, other – race is included as a proxy for social, not biological, risk factors as recently discussed (Boyd, Lindo, Weeks, & McLemore, 2020)), education (less than college education, four years of college education, greater than college education), smoking status (additional adjustment for pack-years did not change results), and total caloric intake (kcal); iii) M3= M2+ body mass index (underweight/normal weight, overweight, and obesity based on World Health Organization (WHO) standards) and prediabetes (present, absent) for primary analysis. A fourth model was used only for analysis of dietary subcomponents adjusting for AHEI or APDQS within that model. P-values were calculated for type-3 sum of squares F-Test analysis of effect across AHEI or APDQS quartiles and to determine any difference in the odds of periodontitis across quartiles of AHEI or APDQS in a 3d.f test. All analyses were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

RESULTS

Participant Characteristics

Participant characteristics are described in Table 1. The study cohort was 73% female with a mean age of 32.0±9.5 years. Participants were Hispanic (32%), non-Hispanic white (29%), non-Hispanic black (15%), or other race (23%). More than half of the participants (76%) had four years of a college education or more, with 23% receiving less than 4 years of a college education. The majority of participants never smoked (85%) and were normal or underweight (55%) with an average BMI value of 25.7±5.8. The prevalence of pre-diabetes in this cohort was 11.4% and had an average caloric intake of 1798.1±97 kcal. The mean overall AHEI and APDQ scores were 47.2±12.2 and 61.1±11.8, respectively, with ranges of 16.4–83.0 and 27.0–96.0, respectively. The Pearson correlation coefficient between AHEI and APDQS was 0.61.

Table 1:

General demographics for ORIGINS participants by AHEI and APDQS Score

	Overall Mean ± Standard deviation or percent2	1st quartile AHEI (n= 230)	2nd quartile AHEI (n= 231)	3rd quartile AHEI (n= 231)	4th quartile AHEI (n= 231)	1st quartile APDQS (n= 231)	2nd quartile APDQS (n= 242)	3rd quartile APDQS (n= 215)	4th quartile APDQS (n= 235)
Overall/Range of AHEI values	47.2±12.2	16.4 – 38.5	38.6 – 46.8	46.81– 55.9	56 – 83	n/a	n/a	n/a	n/a
Overall/Range of APDQS values	61.1±11.8	n/a	n/a	n/a	n/a	27–53	53–61	61–70	70–96
Age (years)	32.0±9.5	32.5±9.5	32.5±9.7	31.8±9.4	31.2±9.4	32.5±9.6	31.7±9.3	32.2±9.8	31.6±9.4
Sex
Male (%)	27	41	28	17	141	34	29	20	171
Female (%)	73	19	24	28	29	22	25	25	28
Race/Ethnicity
Hispanic (%)	32	33	32	21	141	31	30	22	191
Black (%)	15	35	29	18	18	44	24	17	15
White (%)	29	18	17	27	39	13	19	28	39
Other (%)	24	16	24	33	28	20	32	23	24
Education
Less than College (%)	24	35	32	20	131	36	28	18	181
College (%)	50	25	27	23	25	24	27	24	25
Graduate (%)	26	16	15	34	36	17	23	27	33
Smoking
Never (%)	85	24	26	25	251	26	26	24	251
Former (%)	8	14	17	35	35	19	17	24	40
Current (%)	7	49	23	15	12	26	40	19	15
BMI
Underweight/Normal (%)	55	22	22	28	291	23	26	24	27
Overweight (%)	27	25	31	21	23	26	24	25	26
Obese (%)	18	35	28	20	17	32	29	20	20
BMI Mean	25.7±5.8	27.0±6.6	26.4±6.1	24.9±5.3	24.6±4.91	26.4±6.3	26.0±6.0	25.3±5.5	25.2±5.4
Total Caloric Intake	1798.1±978.4	1837.9±1128.3	1770.2±1065.8	1801.5±935.9	1782.9±745.51	1843.1±1082.9	1707.4±1029.3	1825.3±924.5	1822.3±857.2
Prediabetes
No (%)	89	24	25	25	26	24	26	24	26
Yes (%)	11	32	25	25	18	33	27	21	19
Mean Probing Depth	1.99±0.38	1.99±0.37	2.00±0.45	2.01±0.33	1.98±0.36	2.05±0.39	1.94±0.32	2.01±0.42	1.97±0.381
Mean Attachment Loss	0.79±0.76	0.77±0.75	0.83±0.82	0.80±0.69	0.75±0.78	0.92±0.77	0.72±0.71	0.77±0.82	0.75±0.721
%BOPA	34.1±19.6	34.8±18.7	34.9±21.1	33.4±18.1	33.3±20.6	38.3±21.7	31.8±16.9	33.6±18.5	32.9±20.61
PeriodontitisB
None/Mild (%)	35	24	24	25	28	22	27	24	27
Moderate/Severe (%)	65	28	26	26	20	31	25	22	23

Results for mean probing depth, mean attachment loss and %BOP are Mean±SE; results for periodontitis are reported as odds ratios[95% confidence intervals]

¹p<0.05, for differences across diet score quartiles, based on chi-square tests (r x c) for categorical variables with df=(r-1)(c-1) and based on 3-df F-test for continuous variables

²Percents reflect column percentages.

^APercent bleeding on probing

^BDefined as AAP/CDC defined mild or no periodontitis vs. moderate/severe periodontitis

AHEI and Periodontal Disease

The overall AHEI score was not associated with MPD, MAL, %BOP, or CDC/AAP defined periodontitis after adjusting for risk factors (Table 2). In analyses by food groups, nut consumption levels were associated with probing depth such that MPD decreased significantly from 2.02±0.02 in quartile 1 to 1.93±0.03 in quartile 4 (p=0.03) of nut consumption after multivariable adjustment (Supplemental Table 1). Accordingly, periodontitis odds were lower among groups with higher nut consumption, such that the odds ratio (OR) for having moderate/severe periodontitis among participants in the 4^th vs 1^st quartile was 0.43(95% Confidence Interval [CI]: 0.24,0.77) (p-value= 0.03) after multivariable adjustment (Supplemental Table 1). In contrast, MPD (p-value= 0.01) and %BOP (p-value= 0.05) increased with increased levels of red meat consumption (Supplemental Table 2). Similarly, %BOP increased with increasing levels of trans-fat consumption in the fully adjusted models (p-value=0.05, Supplemental Table 3). No statistically significant associations were found for fruits, vegetables, polyunsaturated fats, whole grains, alcohol, refined grains, sugar-sweetened beverages, and omega-3 fatty acids (see Supplemental Tables 4–11).

Table 2:

Association between alternative healthy eating index score and periodontal disease.

	1st quartile AHEI (n= 230)	2nd quartile AHEI (n= 231)	3rd quartile AHEI (n= 231)	4th quartile AHEI (n= 231)	p-valueC
Mean Probing Depth
Model 1	2.03±0.02	2.01±0.02	1.99±0.02	1.96±0.02	0.12
Model 2	2.02±0.02	2.01±0.02	1.99±0.02	1.97±0.02	0.38
Model 3	2.02±0.02	2.01±0.02	1.99±0.02	1.97±0.02	0.43
Mean Attachment Loss
Model 1	0.85±0.04	0.84±0.04	0.76±0.04	0.71±0.04	0.05
Model 2	0.78±0.04	0.81±0.04	0.79±0.04	0.78±0.04	0.94
Model 3	0.78±0.04	0.81±0.04	0.79±0.04	0.79±0.04	0.95
%BOPA
Model 1	36±1	35±1	33±1	33±1	0.18
Model 2	35±1	35±1	33±1	34±1	0.76
Model 3	35±1	35±1	33±1	34±1	0.74
PeriodontitisB
Model 1	REF.	0.83[0.56,1.23]	0.75[0.50,1.11]	0.52[0.35,0.78]	0.02
Model 2	REF.	0.91[0.60,1.38]	0.97[0.63,1.51]	0.76[0.47,1.20]	0.61
Model 3	REF.	0.93[0.61,1.42]	1.01[0.65,1.58]	0.77[0.48,1.23]	0.61

Results for mean probing depth, mean attachment loss and %BOP are Mean±SE; results for periodontitis are reported as odds ratios[95% confidence intervals]

Model 1: wave adjusted; Model 2: Model 1 + age, gender, race/ethnicity, education, smoking, total caloric intake; Model 3: Model 2 + body mass index, prediabetes

^APercent bleeding on probing

^BDefined as AAP/CDC defined mild or no periodontitis vs. moderate/severe periodontitis

^Cp-value for type-3 sum of squares F-Test of effect across AHEI quartiles or p-value corresponds to a 3d.f. test for any difference in the odds of periodontitis across quartiles of AHEI

APDQS and Periodontal Disease

Increased adherence to APDQS was associated with decreased %BOP after multivariable adjustment (Table 3). Specifically, %BOP decreased from 37%±1 in quartile 1 to 34%±1 in quartile 4 (p-value=0.03, Table 3). However, there was no statistically significant association between periodontitis and APDQS and associated periodontal measures (Table 3 & Supplemental Tables 15 & 16). No statistically significant associations were observed between either positive foods or negative foods and MPD, MAL, %BOP, or periodontitis (Supplemental Tables 12 & 13).

Table 3:

Association between a priori diet quality score and periodontal disease.

	1st quartile APDQS (n= 231)	2nd quartile APDQS (n= 242)	3rd quartile APDQS (n= 215)	4th quartile APDQS (n= 235)	p-valueC
Mean Probing Depth
Model 1	2.03±0.02	1.97±0.02	2.01±0.02	1.96±0.02	0.08
Model 2	2.02±0.02	1.97±0.02	2.02±0.02	1.97±0.02	0.17
Model 3	2.02±0.02	1.97±0.02	2.02±0.02	1.97±0.02	0.13
Mean Attachment Loss
Model 1	0.87±0.04	0.78±0.04	0.77±0.04	0.73±0.04	0.08
Model 2	0.84±0.04	0.77±0.04	0.78±0.04	0.77±0.04	0.56
Model 3	0.84±0.04	0.77±0.04	0.79±0.04	0.77±0.04	0.57
%BOPA
Model 1	38±1	33±1	34±1	33±1	0.01
Model 2	37±1	32±1	34±1	34±1	0.04
Model 3	37±1	32±1	34±1	34±1	0.03
PeriodontitisB
Model 1	REF.	0.72[0.49,1.06]	0.64[0.43,0.95]	0.55[0.37,0.82]	0.02
Model 2	REF.	0.72[0.48,1.08]	0.69[0.45,1.08]	0.66[0.43,1.02]	0.22
Model 3	REF.	0.73[0.48,1.11]	0.70[0.45,1.08]	0.62[0.40,0.97]	0.18

Results for mean probing depth, mean attachment loss and %BOP are Mean±SE; results for periodontitis are reported as odds ratios[95% confidence intervals]

Model 1: wave adjusted; Model 2: Model 1 + age, gender, race/ethnicity, education, smoking, total caloric intake; Model 3: Model 2 + body mass index, prediabetes

^APercent bleeding on probing

^BDefined as AAP/CDC defined mild or no periodontitis vs. moderate/severe periodontitis

^Cp-value for type-3 sum of squares F-Test of effect across APDQS quartiles or p-value corresponds to a 3d.f. test for any difference in the odds of periodontitis across quartiles of APDQS

DISCUSSION

This study examined the cross-sectional associations between diet quality measures and periodontal disease in a sample of young adult, well-educated, employed and generally healthy participants. Groups defined by a stronger adherence to the diet scores (i.e. healthier diet) tended to have modestly better periodontal status although no findings for the composite diet scores were statistically significant. When considering sub-scores of the AHEI, nut consumption was associated with better periodontal health, while the opposite was true for consumption of red meat and trans-fats.

Nuts along with fruits, vegetables, and legumes contain flavonoids that have been shown in pre-clinical studies to have the ability to modulate T-cell differentiation, cytokines, and alter gut microbiota, potentially playing a part in human chronic diseases via anti-inflammatory mechanisms (Pei, Liu, & Bolling, 2020). The inverse association between nut consumption and periodontal disease may also be partially explained by the dietary fiber, fat content, calcium, and magnesium content of nuts, as all of these factors have been previously shown to be associated with periodontal disease (Ros, 2010). A recent analysis of the National Health and Nutrition Examination Survey (NHANES) data from 2009–2012 reported that dietary fiber was associated with better periodontal status among US adults aged 30 years or older (Nielsen et al., 2016). With respect to fat content, nuts are high in polyunsaturated fats, and previous studies have reported that improvement in periodontal disease can be accomplished by substituting saturated fats for polyunsaturated fats in the diet (Ros, 2010). Specifically, consumption of n-3 polyunsaturated fats, may alter the fatty acid composition of biological membranes lining dentoalveolar surfaces which in turn buffers against the detrimental effects of acidic cycles during food consumption (Varela-Lopez, Quiles, Cordero, Giampieri, & Bullon, 2015). Fatty acids also possess antioxidant and immunomodulatory effects (Varela-Lopez, Giampieri, Bullon, Battino, & Quiles, 2016a). Similarly, animal models with diets high in n-3/n-6 polyunsaturated fatty acid ratios have been shown to promote better periodontal outcomes (Varela-Lopez et al., 2015). These interpretations are tempered by our current observation that polyunsaturated fat consumption was not clearly associated with better periodontal health; we did not directly assess the association between fiber and periodontal disease.

In regard to red meat consumption, it has been postulated that because red meat provides a rich source of iron, it can reduce oxidative stress by increasing antioxidant enzymes and reduce the worsening of periodontal disease (Najeeb et al., 2016). However, various chronic disease processes are potentially catalyzed by iron initiated oxidative stress (Omaye & Omaye, 2019). Therefore, moderate consumption is recommended for prevention of chronic disease and this may be important in terms of reducing the risk of periodontal disease (Omaye & Omaye, 2019) since high levels of red meat consumption, in our study population, displayed higher markers for periodontal disease as compared to moderate and lower levels of consumption. This may suggest that moderate levels of consumption are satisfactory in maintaining periodontal health and that elevated levels of consumption can produce detrimental effects.

Similar to red meats, high consumption of trans-fatty acids has been shown to be associated with a pro-inflammatory profile, possibly contributing to chronic inflammatory diseases (Mazidi, Gao, & Kengne, 2017; Mozaffarian, 2006). For example, the association of trans-fatty acids with cardiovascular disease is increasingly becoming more understood and the association is suggestive of the underlying dietary inflammation caused by trans-fatty acids (Mazidi, Gao, Shivappa, et al., 2017). While the role of trans-fatty acids in the inflammatory process may not be completely understood, these inflammatory effects may be caused by incorporation of trans-fatty acids into endothelial cells, macrophages, or adipocyte cell membranes disrupting inflammatory signaling pathways and/or having ligand dependent effects on peroxisome-proliferator-activated receptor-γ (Mazidi, Gao, Shivappa, et al., 2017). If trans-fatty acids have a strong impact on systemic inflammation and contribute to chronic inflammatory diseases, it is possible that there is an association with periodontitis, a chronic inflammatory disease.

The strengths of this study lie in the large, multi-racial study population that increases the generalizability of the results. The use of full-mouth periodontal exams minimized bias in our assessment of periodontal disease when comparted to studies that utilize either partial recording methods or questionnaire-based definitions of periodontal disease. In addition, the food frequency questionnaire used to obtain diet data was validated by 2 previous studies, thus increasing the validity of the results (Institute, 2016). We also adjusted for important potential confounders, such as total caloric intake, age, gender, race/ethnicity, education, smoking, body mass index, and prediabetes that increases the internal validity of our study.

One of the caveats of this study was the use of a long dietary assessment instrument which might lead to participant fatigue, inaccurate reporting, and recall bias (Institute, 2016). However, this is a validated method of dietary assessment (Institute, 2016). In addition, because the diet data and dental exam data were collected cross-sectionally at baseline, temporality of associations cannot be established, and participants may have changed their diet after experiencing dental complications.

Oral hygiene practices may potentially confound the results of this study. Previous research has shown that frequency, duration, technique, and even dexterity impact the quality of brushing, which can change the impact brushing has on plaque removal (Moeintaghavi et al., 2017). If these factors are also related to dietary behaviors, confounding is possible, and we do not have detailed information on oral hygiene behaviors. Furthermore, despite a multiethnic cohort, the generalizability of our study results to other populations may be reduced since our study cohort consisted primarily of employed, well-educated women (73%) that were from the New York City area.

With all of these considerations, our findings suggest that general diet pattern is not a strong correlate of current periodontal status among generally healthy, employed young adults and that specific food consumption patterns might be of the most relevance in regard to the influence of diet on periodontal health. Although, moderate findings were observed for red meat, nut and trans-fatty acids consumption, the risk of false positive findings secondary to multiple comparisons precludes definitive statements about these findings. This emphasizes the need to focus future research on food subcategories in relation to periodontal disease development and progression. With a clearer understanding of these associations, dental practitioners would have a better ability to counsel their patients on appropriate diet practice to help prevent or manage periodontal disease. Additional, longitudinal and interventional research is necessary to build on the present findings and more clearly understand the role of dietary behaviors on periodontal health.

CLINICAL RELEVANCE.

Scientific rationale:

Few prior studies have examined the relationship between dietary patterns and periodontal disease. This study examined the cross-sectional association between diet quality and periodontal disease.

Principle findings:

Overall diet quality scores were not associated with periodontal status, while there was a modest association of the consumption of nuts, red meat and trans-fatty acids with periodontal status.

Practical implications:

These findings in conjunction with future findings can help guide clinical and public health recommendations for optimal periodontal health.