Clinical parameters as predictors of destructive periodontal disease activity.

Abstract The purpose of the present investigation was to evaluate the usefulness of clinical measurements of periodontal disease in predicting destructive periodontal disease activity. Periodontal status was monitored at 3414 sites in a total of 22 subjects. Repeat attachment level measurements recorded at 2-month intervals were analyzed by the tolerance method to detect destructive periodontal disease activity. The number of sites that showed or did not show activity and the absence or presence of a clinical parameter before and after the monitoring period were computed. The diagnostic sensitivity of a clinical parameter in predicting disease was expressed as the proportion of sites showing attachment loss which exhibited that parameter. Diagnostic specificity was expressed as the proportion of sites not exhibiting the clinical parameter and not showing attachment loss. In addition, the probability of false positive and false negative diagnoses were computed, using the assumption that the destructive periodontal disease activity rate of sites at risk was 3%. The sensitivity of clinical measurements of gingival redness, plaque, suppuration and bleeding on probing ranged from 0.03 (suppuration) to 0.42 (plaque). Specificity of these measurements was better, ranging from 0.71 for plaque to 0,97 for-suppuration. Disease activity was most often. associated with shallow pockets, but shallow pockets by far dominated she-sites at risk-. Thus; pocket-depth: of <4mm was a sensitive diagnostic test for disease activity (0.69), but the measured specificity of 0.25 indicated that it would be a poor predictor of disease activity. Molar teeth and interproximal surfaces were more likely sites of disease activity than other teeth or surfaces with sensitivity values of 0.52 and 0.83, and specificity values of 0.2S and 0.34, respectively. The probability of detecting false positives was high using any of the clinical parameters ranging from 0.95–0.97. Because no clinical parameter demonstrated high sensitivity and high specificity values, none of the clinical parameters used individually or in combination were found useful in predicting disease activity at individual sites.