Quantifying the Association Between Self-Reported Diabetes and Periodontitis in the New Zealand Population.

BACKGROUND

This study aims to investigate the association between diabetes and periodontitis in the New Zealand (NZ) adult population.

METHODS

Data from two NZ national surveys (N = 2,048) were analyzed to compare estimates of the strength of the association between diabetes and periodontitis using two multivariate epidemiologic approaches (cohort and matched case-control studies). This was possible because the first survey provided participants for the second survey. Periodontitis cases were identified using 14 case definitions, including six severe definitions. The cohort study identified those with diabetes in 2006/07 and those with periodontitis in 2009 to determine the patients with diabetes odds of having periodontitis, using logistic regression modeling (adjusting for smoking status, sociodemographic, and dental characteristics). The matched case-control study identified cases of severe periodontitis in 2009 and compared their 2006/07 diabetic status with that of controls (individually matched on age group, sex, and socioeconomic status). Conditional logistic regression modeling was used for the case-control study, adjusting for ethnicity, smoking status, and dental characteristics.

RESULTS

Overall, 3.6% of the periodontally examined adults reported having diabetes. There was no sex difference in diabetes prevalence, but it was greater in older age groups. Depending on the definition of periodontitis used, there were different estimates of risk for periodontitis, with odds ratios ranging from 1.91 (P = 0.01) to 3.51 (P = 0.22) using the cohort study approach. Diabetes was associated with a greater risk of having periodontitis using only two of the 14 periodontitis case definitions. No association was observed using the matched case-control study.

CONCLUSIONS

The diabetes-periodontitis association in the NZ population remains unclear. This study demonstrates that the determination of the strength of a putative association is method dependent.