Clinical risk indicators for periodontal attachment loss

Abstract

The purpose of the present investigation was to evaluate the association of baseline clinical parameters of periodontal disease with disease progression in the following year. 271 randomly selected subjects from Ushiku Japan were monitored for overt gingivitis and plaque accumulation at 4 sites per tooth and probing pocket depth, probing attachment level and bleeding on probing at 6 sites per tooth for all teeth at baseline and 1 year. A subject was considered to exhibit additional attachment loss if one or more sites increased 3 mm or more in a probing attachment level measurement in one year. The clinical variables included age, sex, number of missing teeth, mean pocket depth and attachment level. In addition, the % of sites which exhibited overt gingivitis, visible plaque, pocket depths, attachment levels or gingival recession over certain mm thresholds or bled on probing were determined. X² analysis was used to seek significant associations between the baseline clinical variables and subsequent attachment loss in a subject. Only 74 of the 271 subjects (27.3%) exhibited additional attachment loss of 3 mm or more at 1 or more sites after 1 year. Older subjects had a greater risk of disease progression than younger subjects. There were no significant differences in % of individuals exhibiting additional attachment loss when divided on the basis of sex, number of missing teeth or % of sites with overt gingivitis. However, the greater the % of sites with visible plaque or which bled on probing, the greater was the likelihood of subsequent attachment loss. Increasing mean levels of pocket depth or attachment level or increasing %s of sites exhibiting prior attachment loss were strongly related to the proportion of individuals with subsequent attachment loss. Gingival recession exhibited similar but weaker relationships. Log-linear analysis suggested that the association between bleeding on probing, age, or plaque levels with additional attachment loss may be explained by the association of these variables with baseline attachment loss. The analyses were repeated with a positive subject defined as having only 1 active site or 2 or more active sites. 37 subjects fit the 1st criterion and the remaining 37 the 2nd criterion. The associations observed were almost identical to those found when subjects were considered positive on the basis of 1 or more changing sites. Discriminant analysis was used to classify subjects as active or inactive using up to 11 predictor variables. 217 of 271 subjects (80%) were correctly classified by discriminant analysis using all 11 predictor variables with a sensitivity of 65% and specificity of 86%. Logistic regression analysis employing 4 predictor variables produced an overall agreement, sensitivity and specificity of 78%, 38% and 93%, respectively.