Assessment of Self‐Report Measures for Predicting Population Prevalence of Periodontitis

Abstract

Self-report measures have been used successfully for the surveillance of chronic diseases in adult populations. This pilot study assessed the use of self-report oral health measures for predicting the population prevalence of periodontitis in United States adults. Data were collected from 456 subjects participating in a 2007 study conducted by the Centers for Disease Control and Prevention. Each subject answered eight predetermined oral health self-report questions obtained from in-person interviews and were given a full-mouth periodontal examination using the National Health and Nutrition Examination Survey protocol. The predictiveness of measures from these self-report questions was assessed by multivariable logistic regression modeling measuring receiver operating characteristic (ROC) statistics, sensitivity, and specificity. Multivariable modeling incorporating self-report measures on gum disease, loose teeth, and tooth appearance alone were most useful in predicting the prevalence of severe periodontitis and improved with the addition of demographic and risk factor variables, yielding an ROC value of 0.93, sensitivity of 54.6%, and specificity of 98% at the observed 4.8% prevalence of disease. Scaling and root planing treatments, loose teeth, and the use of mouthwash, combined with demographic and risk factor covariates, were moderately useful in predicting total periodontitis. Multivariable modeling of specific self-report oral health measures is promising for predicting the population prevalence of severe periodontitis, confirming earlier assessments from a national survey. These results justify further assessments of self-report oral health measures for use in the surveillance of periodontitis in the adult United States population.