Modelling Childhood Caries Using Parametric Competing Risks Survival Analysis Methods for Clustered Data

• Published in: Caries research Vol. 44; no. 1; pp. 69 - 80
• Main Authors: Stephenson, J., Chadwick, B.L., Playle, R.A., Treasure, E.T.
• Format: Journal Article
• Language: English
• Published: Basel, Switzerland S. Karger AG 01.01.2010
• Subjects: Child; Child, Preschool; Cluster Analysis; Cohort Studies; Dental Caries - epidemiology; Dental Caries Susceptibility; Dentistry; Female; Fluoridation; Follow-Up Studies; Forecasting; Humans; Male; Models, Biological; Models, Statistical; Molar - pathology; Original Paper; Proportional Hazards Models; Regression Analysis; Risk Factors; Social Class; Survival Analysis; Tooth Exfoliation; Tooth, Deciduous - pathology; United Kingdom - epidemiology; Survival analysis; Primary teeth
• ISSN: 0008-6568; 1421-976X
• DOI: 10.1159/000279326

Caries in primary teeth is an ongoing issue in children’s dental health. Its quantification is affected by clustering of data within children and the concurrent risk of exfoliation of primary teeth. This analysis of caries data of 103,776 primary molar tooth surfaces from a cohort study of 2,654 British children aged 4–5 years at baseline applied multilevel competing risks survival analysis methodology to identify factors significantly associated with caries occurrence in primary tooth surfaces in the presence of the concurrent risk of exfoliation, and assessed the effect of exfoliation on caries development. Multivariate multilevel parametric survival models were applied at surface level to the analysis of the sound-carious and sound-exfoliation transitions to which primary tooth surfaces are subject. Socio-economic class, fluoridation status and surface type were found to be the strongest predictors of primary caries, with the highest rates of occurrence and lowest median survival times associated with occlusal surfaces of children from poor socio-economic class living in non-fluoridated areas. The concurrent risk of exfoliation was shown to reduce the distinction in survival experience between different types of surfaces, and between surfaces of teeth from children of different socio-economic class or fluoridation status. Clustering of data had little effect on inferences of parameter significance.