Disparities in Periodontitis Prevalence among Chronic Kidney Disease Patients

Abstract

Because of adverse effects of uremia in the innate and adaptive immune systems, we hypothesized that chronic kidney disease (CKD) patients would have higher prevalence of moderate periodontitis compared with individuals without CKD. We examined this hypothesis using the NHANES III dataset, including 12,081 adults stratified by Race-Ethnicity. We followed the American Academy of Periodontology/Centers for Disease Control and Prevention definition for moderate periodontitis. Estimated glomerular filtration rate (GFR) was calculated based on calibrated serum creatinine levels according to the Modification of Diet in Renal Disease Study formula. Analyses incorporated NHANES sampling weights. Overall, 14.6% of individuals with CKD were classified as having moderate periodontitis, compared with 8.7% in the non-CKD group (p = 0.001). A significant dose-response association (p = 0.001) was observed between prevalence of moderate periodontitis and CKD stages among non-Hispanic Blacks and Mexican-Americans, but not so for non-Hispanic Whites. Prevalence of periodontitis among participants with CKD was substantially higher among non-Hispanic Blacks (38.9%) and Mexican-Americans (37.3%) compared with non-Hispanic Whites (12.9%). Multivariate logistic regression models showed that Mexican-Americans and non-Hispanic Blacks with CKD were approximately 30% to 60% more likely to have moderate periodontitis compared with those without CKD, after adjustment for diabetes status and other potential confounders.

Introduction

Chronic kidney disease (CKD) is a worldwide public health problem, with adverse outcomes such as kidney failure, cardiovascular disease, and premature death (Levey et al., 2005). CKD is defined by glomerular filtration rate (GFR) < 60 mL/min/1.73 m² of body surface area with or without kidney damage (Levey et al., 2005). Based on this cut-off point, it has been estimated that eight million people in the US have CKD (Coresh et al., 2002b).

CKD patients are characterized by a uremic milieu, which has been associated with immunosuppression. Disturbances of the immune system in CKD involve both innate and adaptive immunity (Kato et al., 2008). There has been a wealth of evidence that disorders of both the innate and adaptive immune systems contribute to an increased rate of infections in the course of CKD. Functional abnormalities of monocytes, neutrophils, and dendritic cells are directly linked to infection risk in this patient population (Anding et al., 2003; Carracedo et al., 2006). Impaired maturation of Th lymphocytes also may lead to disabled immune response and infection susceptibility (Ando et al., 2005). As a result of these processes, high prevalence of chronic infection by Chlamydia pneumoniae has been observed in CKD patients (Stenvinkel et al., 1999).

It has been hypothesized that one of the major sequelae stemming from uremic-induced dysregulation of the immune system would be an increase in opportunistic infections, which, if confirmed, may necessitate increased surveillance among the CKD population. One such infectious state is periodontitis, a common, polymicrobial, and predominantly Gram-negative infection affecting many in the US adult population. However, the body of prior studies of CKD and periodontitis has produced mixed results, owing to wide variations in inclusion criteria, definitions of CKD and periodontitis, and control of confounders in analyses.

Thus, based on the adverse effects of uremia in the innate and adaptive immune systems, we hypothesized that CKD patients would have higher prevalence of chronic periodontitis compared with non-CKD individuals.

We examined this question in a large (n = 12,081) representative sample of the US population using the National Health and Nutrition Examination Survey-1988 to 1994 (NHANES III; Center of Disease Control and Prevention, 1996). Furthermore, given a much greater prevalence of periodontitis among different racial groups, we examined if the hypothesized relationship varied by race.

Materials & Methods

Study Population

NHANES, a periodic survey conducted by the Centers for Disease Control and Prevention (CDC, 1996), provides national estimates of health and nutritional status of the civilian, non-institutionalized population. Individuals participated in interviews conducted at home and in extensive physical examinations, performed at an examination center, which included blood and urine collection. Our study population (n = 12,081) consisted of persons who were 21 years of age and older, dentate (≥ 2 teeth, so that we could apply the selected definition for moderate periodontitis), and for whom there were no missing data on creatinine and other variables in the analyses. Age, sex, race-ethnicity, and smoking history were based on self-report during the interview. We constructed a binary age variable using a cut-off point of 60 years, based on risk for CKD (Levey et al., 2003). Smoking status was derived from the questions: “Have you smoked at least 100 cigarettes in your life?” and “Do you smoke cigarettes?” Diabetes status was based on the answer to: “Have you been told by your doctor that you have diabetes?” Glycated hemoglobin was used as an indicator of diabetes control. We constructed a variable for diabetes duration using a previously used cut-off point of 10 years (Fisher and Taylor, 2009). Hypertension and history of cardiovascular disease variables were extracted from interview data.

Education, health perception, and income variables were constructed based on previous reports (Borrell et al., 2002). Body mass index (BMI) was calculated as body weight (kg) divided by height (mm)² and stratified into 4 standard categories.

Moderate Periodontitis Definition

For NHANES III, the partial-mouth periodontal examination (PMPE) was conducted at 2 sites per tooth in 2 randomly chosen quadrants. All third molars were excluded. Missing teeth were calculated based on extractions due to periodontal disease, caries, or orthodontic reasons.

We followed the American Academy of Periodontology (AAP)/CDC definition for moderate periodontitis and used at least 2 interproximal sites with clinical attachment loss (CAL) ≥ 4 mm not on the same tooth or at least 2 sites with probing depth (PD) ≥ 5 mm not on the same tooth (Page and Eke, 2007).

Adjusted Prevalence of Moderate Periodontitis

Since PMPE as used in NHANES III has been shown to underestimate periodontitis prevalence (Susin et al., 2005), we applied the inflation factor of 2.4 (ratio of True Prevalence to Observed Prevalence), as calculated by Eke et al. (2010), for the AAP/CDC full-mouth periodontal examination adjusted prevalence.

Kidney Function

Estimated glomerular filtration rate (GFR) was calculated based on serum creatinine level, which was calibrated according to the equation −0.184 + 0.960 x NHANES III creatinine value (Coresh et al., 2002a; Selvin et al., 2007). GFR was estimated with the Modification of Diet in Renal Disease (MDRD) Study formula (Levey et al., 2007) that was recently re-expressed with standard creatinine (Levey et al., 2007), as follows: GFR = 175 x (standard serum creatinine)^−1.154 x (age)^−0.203 x 0.742 (if female) and x 1.212 (if African American). Estimated GFR is reported in mL/min/1.73 m². All GFR values above 200 mL/min/1.73 m² were truncated at that level (Coresh et al., 2007). All individuals with GFR < 15 mL/min/1.73 m² were excluded because of the likelihood of being on dialysis or seriously ill with low response rate. Cut-off points for CKD stages followed the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (Levey et al., 2005).

Statistical Analyses

Analyses were performed with PAWS Version 18 with Complex Sample Module incorporating sampling weights in the NHANES dataset (Mohadjer et al., 1996). The weight for the smallest analysis subpopulation was selected (WTPFHX6). Categorical variables were tested with the Pearson chi-square test, and continuous variables were tested with the t test for independent samples. We performed binary logistic regression to calculate crude and adjusted Odds Ratios (OR) and 95% Confidence Intervals (CI) for moderate periodontitis among CKD cases. Participants with GFR ≥ 60 mL/min/1.73 m² constituted the referent group. Three adjusted logistic regression analyses were applied: Model 1 included age and sex; Model 2 was based on 2 prior studies (Coresh et al., 2007; Fisher et al., 2010) and included age, sex, diabetes, diabetic control and duration, BMI, smoking, cardiovascular disease, and hypertension; and Model 3 included variables in Model 2 and factors related to socio-economic status (income, education, and health perception). The p values of < 0.05 were accepted as statistically significant in all analyses.

Results

Descriptive characteristics of the study sample stratified by race are presented in Table 1. Moderate periodontitis prevalence among non-Hispanic Whites was significantly higher in CKD compared with non-CKD cases (12.9% vs. 7.5%, respectively, p = 0.001). Among non-Hispanic Blacks, moderate periodontitis was also significantly more prevalent among CKD cases (38.9% vs. 14.6%, respectively, p = 0.001), as was observed among Mexican-Americans (37.3% vs. 9.0%, respectively, p = 0.001). CKD cases had significantly more missing teeth compared with non-CKD across all ethnic groups (Table 1). In the Other group, although CKD cases had more missing teeth, the difference did not reach statistical significance (data not shown). For the entire sample (data not shown), the prevalence of moderate periodontitis in the CKD cases was 14.7% compared with 8.7% in the non-CKD group (p = 0.001). Inflation-adjusted prevalence of moderate periodontitis among CKD cases was 35.28%.

Table 1.

Clinicopathological Characteristics of Study Population (n = 11,604)¹ Stratified by Race-Ethnicity

	Non-Hispanic Whites			Non-Hispanic Blacks			Mexican-Americans
	CKD N = 437	Non-CKD N = 4216	p-Value	CKDN = 88	Non-CKD N = 3182	p-Value	CKD N = 81	Non-CKD N = 3600	p-Value
Moderate Periodontitis	12.9 (1.8)2	7.5 (0.4)	0.001	38.9 (5.8)	14.6 (0.5)	0.001	37.3 (6.8)	9.0 (0.5)	0.001
Age
≥ 60 yrs	74.2 (3.1)	15.1 (0.3)	0.001	76.3 (4.9)	8.8 (0.5)	0.001	66.8 (6.3)	7.0 (0.4)	0.001
Sex
Female	62.7 (2.7)	50.9 (0.9)	0.001	61.3 (5.4)	54.5 (1.0)	0.001	39.6 (6.7)	47.1 (1.0)	0.282
Missing teeth
Mean ± SE	2.25 ± 0.17	1.71 ± 0.04	0.003	4.58 ± 0.65	2.77 ± 0.06	0.020	4.05 ± 0.62	1.57 ± 0.04	0.001
Education
< 12 yrs	28.0 (2.4)	18.2 (0.6)	0.001	63.8 (5.6)	26.5 (0.8)	0.001	58.6 (7.2)	56.3 (1.0)	0.750
Health Perception
Fair-Poor	27.0 (2.5)	11.3 (0.5)	0.001	60.1 (5.7)	18.4 (0.7)	0.001	48.7 (7.0)	30.4 (0.9)	0.001
Income
Low	38.6 (3.0)	21.7 (0.8)	0.936	71.4 (5.3)	51.3 (1.0)	0.001	68.1 (7.5)	51.7 (1.1)	0.127
Medium	38.4 (3.2)	47.0 (1.0)		17.3 (4.3)	38.7 (1.0)		21.1 (6.0)	38.1 (1.0)
High	23.0 (3.0)	31.3 (0.9)		11.3 (3.9)	10.0 (0.7)		10.8 (6.2)	10.1 (0.7)
Diabetes	10.3 (1.7)	3.7 (0.3)	0.001	21.7 (4.6)	5.7 (0.4)	0.001	22.0 (4.9)	6.0 (0.4)	0.001
Diabetic Control
HbA1c ≥ 7%	9.3 (1.7)	2.3 (0.3)	0.001	17.3 (4.2)	4.7 (0.4)	0.001	21.8 (5.0)	4.7 (0.4)	0.001
Diabetes Duration
≥ 10 yrs	49.1 (9.2)	30.4 (4.4)	0.149	68.1 (10.9)	33.6 (3.4)	0.004	68.1 (10.9)	33.6 (3.4)	0.004
Hypertension	53.2 (3.1)	20.0 (0.7)	0.001	53.2 (3.1)	20.0 (0.7)	0.001	53.7 (7.1)	15.4 (0.7)	0.001
Cardiovascular disease	12.6 (1.9)	2.2 (0.2)	0.149	13.4 (3.6)	1.8 (0.3)	0.003	12.2 (4.7)	1.1 (0.2)	0.001
Smoking Status
Current	11.0 (1.9)	27.8 (0.8)	0.001	19.1 (4.7)	33.2 (0.9)	0.001	15.2 (5.5)	22.7 (0.8)	0.001
Past	42.7 (3.0)	25.1 (0.7)		29.6 (5.1)	14.6 (0.7)		49.5 (7.1)	20.3 (0.8)
Never	46.3 (2.9)	47.1 (0.9)		51.3 (5.8)	52.2 (1.0)		35.3 (6.8)	57.0 (1.0)
Body Mass Index
Low < 18.5	1.7 (0.7)	2.2 (0.2)	0.001	1.5 (1.1)	2.1 (0.3)	0.019	0 (0)	(0.3)	0.469
Normal (18.5-24.9)	29.6 (2.6)	44.2 (0.9)		17.4 (4.2)	35.7 (1.0)		27.4 (6.7)	33.7 (0.9)
Overweight (25.0-29.9)	39.7 (2.9)	32.9 (0.8)		42.6 (5.8)	32.7 (0.9)		38.6 (6.6)	38.7 (1.0)
Obese (≥ 30.0)	29.0 (2.0)	20.7 (0.5)		38.5 (4.8)	29.4 (0.7)		27.5 (6.6)	18.4 (0.8)

¹n = 477 in Other category not presented in this Table.

²All data presented as % (SE) except for Missing Teeth (Mean ± SE).

As seen in Table 2, among the total sample, prevalence of moderate periodontitis increased stepwise across CKD stages, but this trend did not reach statistical significance (p = 0.320). In the non-Hispanic Blacks and Mexican-Americans, however, statistically significant (p = 0.001) dose-response effects across CKD stages were observed. In the Other group, the descriptive, dose-effect, and regression results were not significant, possibly due to the small sample of CKD cases (n = 17).

Table 2.

Dose-Response Association between Kidney Function and Moderate Periodontitis Prevalence (SE = Standard Error)

	Moderate Periodontitis Prevalence % (SE) [unweighted n]
Race-Ethnicity	CKD STAGE 1	CKD STAGE 2	CKD STAGE 3	CKD STAGE 4	p-Value
All Subjects	8.0 (0.5) [771]	9.5 (0.6) [676]	12.7 (1.8) [134]	13.7 (5.6) [9]	0.320
Non-Hispanic Whites	6.6 (0.6) [154]	8.6 (0.6) [278]	10.6 (1.9) [75]	11.5 (5.6) [3]	0.116
Non-Hispanic Blacks	13.1 (0.7) [342]	20.3 (1.6) [171]	35.5 (5.2) [29]	40.7 (2.1) [3]	0.001
Mexican-Americans	7.1 (0.5) [243]	16.4 (1.5) [213]	33.6 (1.5) [26]	100.0 (0.0) [3]	0.001
Other	11.6 (1.7) [32]	9.9 (3.3) [14]	10.7 (5.9) [4]	0.0 (0.0) [0]	0.791

CKD, chronic kidney disease.

Univariate logistic regression (Table 3) revealed that, among Mexican-Americans, CKD cases were almost 3-fold more likely than non-CKD cases to have moderate periodontitis (OR = 2.77, CI 95% 2.15-3.55). In multivariate models 1, 2, and 3, the association remained statistically significant but was somewhat attenuated.

Table 3.

Logistic Regression Analyses for the Association of Moderate Periodontitis Prevalence (dependent variable) and Chronic Kidney Disease (CKD) in Non-Hispanic Blacks and Mexican-Americans, Both in Relation to Non-CKD Individuals (referent group)

	Crude	Model 1	Model 2	Model 3
Non-Hispanic Blacks OR (95% CI)
CKD (GFR < 60)	1.85 (1.48-2.30)*	1.35 (1.07-1.70)*	1.30 (1.10-1.88)*	1.24 (0.95-1.62)
Mexican-Americans OR (95% CI)
CKD (GFR < 60)	2.77 (2.15-3.55)*	1.72 (1.28-2.38)*	1.70 (1.25-2.30)*	1.59 (1.14-2.13)*

Model 1 adjusted for sex and age. Model 2 adjusted for age, sex, smoking, diabetic status, diabetic control and duration, BMI, cardiovascular disease, and hypertension. Model 3 adjusted for all variables in Model 2, as well as health perception, income, and education.

Likewise, among non-Hispanic Blacks, those with CKD were almost 2-fold more likely to have moderate periodontitis (OR = 1.85, 95% CI 1.48-2.30) in the univariate analyses. Statistical significance held in models 1 and 2 but not so in model 3. Among non-Hispanic Whites (data not shown), although we observed a statistically significant crude OR of 1.40 (CI 95%, 1.10-1.78), multivariate models were not significant.

Discussion

We have produced evidence consistent with our hypothesis that prevalence of moderate periodontitis is greater among CKD cases compared with individuals without CKD in a representative sample of the dentate US adult population. We also found that the prevalence of periodontitis among participants with CKD was substantially greater among Mexican-Americans (37.3%) and non-Hispanic Blacks (38.9%) compared with non-Hispanic Whites (12.9%). Furthermore, our analysis demonstrated a statistically significant dose-response effect between stage of kidney dysfunction and moderate periodontitis prevalence among non-Hispanic Blacks and Mexican-Americans, but not so among non-Hispanic Whites. Although a multivariate logistic regression model showed that non-Hispanic Blacks with CKD were almost 2-fold more likely to have moderate periodontitis compared with non-Hispanic Blacks without CKD, after adjustment for potential confounders, particularly SES, the model did not remain statistically significant. The same multivariate model in the Mexican-Americans showed an OR of 1.59 (CI 95% 1.14-2.13) even after adjustment for confounders including SES. A differential risk factor for moderate periodontitis was not observed among non-Hispanic Whites and others.

Our findings agree with those of previous NHANES III reports that showed an almost double prevalence of periodontitis (19.6% vs. 10.7%) in non-Hispanic Blacks compared with non-Hispanic Whites in the general population (Borrell et al., 2002). Furthermore, our findings are consistent with reports of the higher prevalence of CKD in non-Hispanic Blacks and Mexican-American populations compared with non-Hispanic Whites (United States Renal Data System, 2009). The disparities in these populations have been attributed to high prevalence of type 2 diabetes, diabetic nephropathy, and lower SES (Bruce et al., 2010). Although there have been efforts to associate genetic variants for type 2 diabetes risk among different racial-ethnic groups, recent studies have found that SES differences could explain the increased prevalence of type 2 diabetes in particular populations (Waters et al., 2010). With respect to these findings, our study used multivariate regression models that incorporated SES status variables.

The main limitation of our study was the cross-sectional design that prevented the assessment of temporality. Additional long-term studies assessing the role of uremia in CKD populations are needed to confirm this mechanism and clarify the disparity by race.

Another limitation was the use of partial-mouth periodontal examination, which tends to underestimate disease prevalence (Eke et al., 2010). We estimated a true moderate periodontitis prevalence of 35.28% compared with 14.7% observed prevalence in CKD populations. Ours is the first report, to our knowledge, of a true prevalence study in CKD populations using NHANES III data.

A further limitation was the fact that the MDRD GFR equation has not been validated in Mexican-American populations (Coresh et al., 2007) and has been shown to underestimate GRF in individuals with normal kidney function (Levey et al., 2007).

Our study offers several important methodological advancements. First, compared with previous studies (Kshirsagar et al., 2005; Fisher et al., 2008), we used a more recent periodontitis case definition (Page and Eke, 2007), which has been adopted in official US public health publications on national prevalence of periodontitis (Dye et al., 2007). Our group has shown previously that the choice of periodontitis case definition could affect associations with renal function (Ioannidou et al., 2010). Furthermore, unlike prior studies (Fisher et al., 2008), we limited our analysis to dentate adults, since we were unable to determine the edentulism frequency attributed to periodontitis.

Second, we calibrated creatinine levels as recommended (Coresh et al., 2002a) for NHANES III, and we also used the re-expressed MDRD GFR equation as recommended (Levey et al., 2007). Therefore, our results were in compliance with the guidelines for GFR estimation using NHANES III data.

Several alternative explanations could support the statistically significant association between periodontitis and CKD. First, both diseases are characterized by a high prevalence of diabetes, which has been independently associated with periodontitis (Fisher et al., 2010) and, similarly, has been one of the major primary etiologies of CKD (Patel et al., 2005). Alternatively, smoking has been shown to be independently associated with CKD and periodontitis (Fisher et al., 2008).

Although there are several mechanisms that explain the high inflammatory status in the CKD population, non-traditional risk factors have been involved in a qualitatively and quantitatively different risk relationship with cardiovascular disease (CVD) in the CKD population (Sarnak, 2003). Among the novel factors, persistent periodontal inflammation, usually evidenced by elevated serum cytokine levels, may be the missing link that could possibly explain the tremendous burden of CVD in CKD and the direct impact of periodontal infection on CKD.

More recent sophisticated structural equation models have explored the bidirectional relationship of CKD and periodontitis using diabetes duration as a possible mediator of this relationship (Fisher et al., 2010). Based on these structural equation models, periodontal disease was found to have a direct effect on diabetes duration, which directly affects hypertensive status, leading to CKD after adjustment for the direct effects of the other potential risk factors. Therefore, this model suggested that individuals with moderate periodontitis were more likely to have CKD. Furthermore, additional structural models confirmed the direct effect of CKD on moderate periodontitis.

On the basis of this analysis, an estimated 5.8 ± 0.3 million US dentate adults with CKD stages 2 to 4 have moderate periodontitis, which, if untreated, may significantly affect the inflammatory status of this population, as has been shown in other populations (Ioannidou et al., 2006). Further interventional studies are needed to test the effect of periodontal treatment on inflammatory status and cardiovascular mortality in the CKD population.