Predictive Power of Attachment Loss Measures: Severity of Periodontal Diseases

Objective: To create and evaluate the predictive power of fifteen different measures of loss of attachment (LOA) regarding clinically determined periodontal needs. Methods: Data from National Health and Nutrition Examination Survey (1999-2002) was for analysis. The clinical norm used to evaluate the different LOA measures is the clinical determined outcome measure of “whether the subject has periodontal needs”. The 15 LOA measures were created based on factors, such as the worst scenario, the average, or the cumulative LOA across 14 teeth (half mouth). Each LOA was measured at two sites of mid-facial and mesial. The covariates controlled in the analysis are the demographic characteristics (e.g., age, gender, ethnicity, and education), smoking status, and diabetes status. Weighted logistic regression models were used for each of the LOA measures and the goodness of fit of these models were evaluated by Akaike's information criterion (AIC), a penalized likelihood and others. Results: 28.7% of the population had clinically determined periodontal needs. All of the nine continuous LOA measures and six binary LOA measures were statistically significant for predicting periodontal needs. According to the models, the three most predictive LOA measures for periodontal needs are (1) the highest sum of the mid-facial and mesial measures, which are the worst case for both measurement sites; (2) the highest of the mesial measure, the worst case of mesial measure; (3) the sum of the mid-facial and mesial measures, the cumulative of both sites; Conclusion: This analysis is among the first to evaluate the predictive power of different severity of periodontal disease measures based on LOA. The highest of the sum of the worst case at mid-facial and mesial has the strongest predictive power, followed by the highest measured for mesial and then the sum of mid-facial and mesial measures. Supported by NIDCR/NIH grant R03DE018767.