Periodontal infection, impaired fasting glucose and impaired glucose tolerance: results from the Continuous National Health and Nutrition Examination Survey 2009-2010

Abstract

Aim: We investigated the relationship between periodontal disease, a clinical manifestation of periodontal infection, and pre-diabetes.

Methods: The National Health and Nutrition Examination Survey, 2009-2010 enrolled 1165 diabetes-free adults (51% female) aged 30-80 years (mean ± SD=50 ± 14) who received a full-mouth periodontal examination and an oral glucose tolerance test. Participants were classified as having none/mild, moderate or severe periodontitis and also according to mean probing depth ≥ 2.19 mm or attachment loss ≥ 1.78 mm, (respective 75th percentiles). Pre-diabetes was defined according to ADA criteria as either: (i) impaired fasting glucose (IFG) or impaired glucose tolerance (IGT). In multivariable logistic regression models, the odds of IFG and IGT were regressed on levels of periodontitis category.

Results: The odds ratios and 95% confidence intervals for having IGT among participants with moderate or severe periodontitis, relative to participants with none/mild periodontitis were 1.07 [0.50, 2.25] and 1.93 [1.18, 3.17], p = 0.02. The ORs for having IFG were 1.14 [0.74, 1.77] and 1.12 [0.58, 2.18], p = 0.84. PD ≥ 75 th percentile was related to a 105% increase in the odds of IGT: OR [95% CI] = 2.05 [1.24, 3.39], p = 0.005.

Conclusions: Periodontal infection was positively associated with prevalent impaired glucose tolerance in a cross-sectional study among a nationally representative sample.

Keywords: glucose metabolism; infection; periodontal disease; periodontitis.

Figure 1

A. Directed Acyclic Graph (DAG) representing one possible underlying causal structure of interrelationships among periodontal infection, prediabetes and several mediators or confounders of the association. This causal structure assumes potential socio-demographic confounders relate to periodontal infection and prediabetes through multiple mechanisms (e.g., confounding by socio-demographic variables can act through health behaviors and adiposity but also through other mechanisms not represented in our data); this causal structure would necessitate adjustment for all sociodemographic variables. A similar argument can be made for more proximal variables in the causal chain such as health behaviors and adiposity. Only statistical adjustment for all potential confounders will provide the least confounded estimate. B. Directed Acyclic Graph (DAG) representing one possible underlying causal structure of interrelationships among periodontal infection, prediabetes and several mediators or confounders of the association. This causal structure assumes that all confounding effects of sociodemographic variables act entirely through effects on either health behaviors or adiposity (two constructs which are measured in our data); this causal structure does not require adjustment for sociodemographic variables so long as health behaviors and adiposity adjustments are made.

Figure 1

A. Directed Acyclic Graph (DAG) representing one possible underlying causal structure of interrelationships among periodontal infection, prediabetes and several mediators or confounders of the association. This causal structure assumes potential socio-demographic confounders relate to periodontal infection and prediabetes through multiple mechanisms (e.g., confounding by socio-demographic variables can act through health behaviors and adiposity but also through other mechanisms not represented in our data); this causal structure would necessitate adjustment for all sociodemographic variables. A similar argument can be made for more proximal variables in the causal chain such as health behaviors and adiposity. Only statistical adjustment for all potential confounders will provide the least confounded estimate. B. Directed Acyclic Graph (DAG) representing one possible underlying causal structure of interrelationships among periodontal infection, prediabetes and several mediators or confounders of the association. This causal structure assumes that all confounding effects of sociodemographic variables act entirely through effects on either health behaviors or adiposity (two constructs which are measured in our data); this causal structure does not require adjustment for sociodemographic variables so long as health behaviors and adiposity adjustments are made.