Modelling the bilateral symmetry of caries incidence

• Published in: Caries research Vol. 42; no. 4; p. 291
• Main Authors: Burnside, G, Pine, C M, Williamson, P R
• Format: Journal Article
• Language: English
• Published: Switzerland 01-01-2008
• Subjects: Adolescent; Algorithms; Child; Dental Caries - epidemiology; Dental Caries - prevention & control; Functional Laterality; Humans; Incidence; Longitudinal Studies; Models, Statistical; Pit and Fissure Sealants; Predictive Value of Tests; Probability; Scotland - epidemiology; Scotland
• ISSN: 1421-976X
• DOI: 10.1159/000148161

Past caries experience has been shown to be the best predictor of the development of caries in the future, and clinical observations suggest that caries develops symmetrically in similar teeth on each side of the mouth. This study investigates whether caries on a given surface can be used as a predictor of future caries on the corresponding surface on the other side of the mouth.
The data come from a 3-year trial examining the caries-preventive efficacy of chlorhexidine varnish on adolescents. A logistic multilevel model was fitted with 3 levels; participant, tooth and surface. The outcome variable was the development of caries into enamel or dentine, after 3 years. Covariates were tooth position, the caries status of the contralateral surface at baseline, the caries status of the corresponding surface in the opposing jaw at baseline, the caries status of adjacent teeth and the total number of decayed, missing, filled surfaces at baseline.
The effect of caries at baseline on the contralateral surface was highly significant (odds ratio = 4.80, 95% CI = 4.38-5.38). The effect of caries at baseline on the corresponding surface in the opposing jaw was also significant, but smaller in magnitude (odds ratio = 1.66, 95% CI = 1.49-1.83).
Multilevel modelling provides a clinically useful method of estimating the probability of a surface developing caries over a period of time, based on the caries status of the contralateral surface and the corresponding surface in the opposing jaw, while controlling for the natural clustering in tooth surface data.