Background: Evidence is accumulating to support poor oral health as a risk factor for systemic conditions, including cardiovascular diseases, diabetes control, adverse pregnancy outcomes, and pneumonia. Prohibitive costs for clinical assessment of periodontal disease limit information to assess the prevalence and trends of periodontal diseases in the United States population. However, self-report is used widely to assess economically the population-based prevalence of various medical conditions and health-related behaviors and characteristics.
Methods: The goal of this secondary data analysis was to identify self-report items sufficiently correlated with clinical periodontal disease for use via face-to-face or telephone interviews. Data for analysis were collected for a project focused on oral health that included face-to-face interview items regarding oral health-related self-care, professional care, and barriers; knowledge, beliefs, and attitudes; risk behavior; impact on quality of life; and demographic characteristics. Also, participants had complete oral examinations.
Results: Logistic regression analyses identified self-reported items contributing to two sets of models predicting moderate or severe periodontal disease (MODSEV) and severe periodontal disease (SEV). Age, gender, race/ethnicity, smoking, and periodontal health-related self-report items constituted predictive models with maximum sensitivity and specificity of 71% and 83%, respectively, with area under the receiver operating characteristic curve (AUC) of 0.85 (as a measure of accuracy) for MODSEV. For SEV, predictive models' maximum sensitivity and specificity were 92% and 53%, respectively, with a maximum AUC of 0.92.
Conclusion: These analyses suggest that self-report may be valid for surveillance of periodontal disease burden and trends in the American population, in lieu of more costly clinical periodontal examinations.