A cluster analysis model for caries risk assessment.

Cluster analysis was applied to determine, the natural grouping of individuals, among sixty 8-10-year-old children, and to identify the most significant set of markers for risk assessment. The risk clusters were obtained with initial clinical and bacteriological measurements including dmf + DMFS, active caries, mutans streptococci and lactobacilli counts in plaque or saliva on two media, and Snyder's test results. The morbidity clusters were constructed with the final clinical indexes and incidence after 18 months (dependent variables). A risk cluster was identified that included the following significant initial variables; dmf + DMFS, active caries, counts of mutans streptococci from plaque on TSY20B and lactobacilli in saliva, and Snyder's test results. This set of markers identified 86% of the children at high risk who developed high morbidity, as well as 94% of children in the low-risk cluster who developed low or no caries. The results of this investigation provide the basis to develop a system for caries risk assessment.