Distributions of periodontal attachment levels. Mathematical models and implications

Distributions of periodontal attachment levels at probing sites within patients have traditionally been used in clinical diagnosis and treatment planning. Progression from mild to moderate to severe disease is generally associated with increasing magnitudes of attachment loss at greater percentages of sites. Recent analyses of distributions of periodontal attachment levels have suggested three general patterns of loss defined by: (1) loss at less than about one third of all sites, (2) more widespread disease with multiple peaks and (3) normally distributed loss with virtually all sites being affected. In attempting to stimulate these three patterns using a model based on the burst theory of periodontal attachment loss, divergent assumptions about burst magnitude, frequency, and possible local immunity were required. These findings were used to support the hypothesis that distinctly different disease processes are associated with the different patterns of attachment level. In the present investigation an alternative model was developed which was theoretically consistent with the view that the three patterns reflect arbitrary stages in a continuous disease "aging" process. Simple assumptions concerning attachment loss probabilities and rates enabled the generation of attachment level distributions that matched all three patterns previously attributed to separate disease processes, depending only upon the duration of the process.