Radiographic quantification of alveolar bone level changes. Predictors of longitudinal bone loss

The purpose of the present study was to determine the relative importance of a set of predictors for the radiographic change in alveolar bone level over 2 years in a group of 180 subjects and to assess the fit of the applied multivariate model. The included predictors were age, sex, initial bone level, type of initial lesion, tooth type, proximal tooth surface, calculus, and the presence of metal crowns or proximal fillings. The multivariate method of choice was multiple classification analysis. The results indicated that in descending order tooth type, initial bone level, and age were the most important predictors. Sex, calculus, and crown and filling margins also showed significant direct effects (beta) on longitudinal bone loss (p less than 0.05). However, their relative importance was far less than that of tooth type, initial bone level, and age. Proximal side showed no significant direct effect. Using seven predictors, the applied multivariate model explained 20% of the variance in longitudinal bone loss.