Background: The aim of this study is to evaluate the validity of screening methods in predicting periodontitis in people with disabilities using the objective salivary hemoglobin level, a subjective self-report questionnaire, and a combined model of the two methods with demographic characteristics.
Methods: The participants were 195 patients with disabilities aged >18 years who were examined using the community periodontal index (CPI), salivary hemoglobin level, and answers to 10 self-report questions (n = 192). Multivariable logistic regression and receiver operating characteristic (ROC) curve analysis were performed to evaluate the validity of the methods and the combined model in predicting the prevalence of ≥CPI 3 (probing depth [PD] ≥4 mm) or CPI 4 (PD ≥6 mm).
Results: Overall, 75.9% of the study group (148 of 195) were diagnosed with ≥CPI 3, and 38.5% of the study group (75 of 195) were diagnosed with CPI 4. The areas under the ROC curve (AUCs) of the salivary hemoglobin level were 0.578 (sensitivity of 41% and specificity of 77%) and 0.662 (sensitivity of 53% and specificity of 75%) for predicting the prevalence of ≥CPI 3 and CPI 4, respectively. Multivariable modeling incorporating four different questions for predicting ≥CPI 3 or CPI 4 indicated higher AUCs of 0.710 and 0.732, respectively, yielding higher sensitivity (55% for ≥CPI 3 and 69% for CPI 4) than that of salivary hemoglobin level. The most useful prediction models for ≥CPI 3 or CPI 4 were combined models, which yielded AUCs of 0.773 and 0.807, respectively, with sensitivity values of 70% and specificity values >75%.
Conclusion: The salivary hemoglobin level, self-report questionnaire, and the combined method demonstrated screening potential that could predict the population prevalence of ≥CPI 3 or CPI 4.
Keywords: Disabled persons; ROC curve; hemoglobins; periodontal index; periodontitis; self-report..