Classification of dental caries patterns in the primary dentition: a multidimensional scaling analysis

Clinical patterns of early childhood caries (ECC) encompassing specific teeth or surfaces have been previously proposed on an a priori basis and have been used as case definitions. The underlying assumption is that the patterns result from different host response and environmental conditions. Identifying caries patterns has utility in refining case definitions of ECC. Well-defined caries patterns should enhance the ability of an analysis to identify meaningful associations between suspected risk factors and ECC. The purpose of this project was to identify patterns of caries in the dentition of preschool children using multidimensional scaling without a priori pattern delineation.

Between February 1994 and September 1995, five examiners visually examined 5171 Arizona preschool children aged 5-59 months old. Multidimensional scaling (MDS) was used as a classification/taxonomy technique to identify any underlying structure of the caries data. MDS provided a classification scheme for individual tooth surfaces based on the dissimilarity measures of squared Euclidean distance and of variance using an alscal analysis. Both two- and three-dimensional solutions were pursued; s-stress, stress, R-square and residual patterns were assessed in determining the best dimensional model, with the resulting quadrant positions of the tooth surfaces suggesting potential caries patterns.

All models demonstrated excellent fit. Two- and three-dimensional solutions suggested four caries patterns: (i) any maxillary incisor surfaces, (ii) first molar occlusal surfaces, (iii) second molar pit and fissure surfaces, and (iv) any smooth surfaces, excluding the maxillary incisor surfaces.

This is the first delineation of primary dentition caries patterns produced by a classification analysis without a priori pattern definitions. The identified caries patterns may arise from specific risk factors and/or be a function of the timing of various risk factor exposures. Use of these patterns as case definitions should enhance the ability to identify associations between suspected risk factors and ECC.