C Källestål1, H Stenlund. 1. Department of Epidemiology, Public Health and Clinical Medicine, Umeå University, Umeå, Sweden. carina.kallestal@epiph.umu.se
Abstract
AIM: To use different analytical approaches to data from a longitudinal, quasi-experimental study evaluating caries-preventive measures in order to estimate the differences in results due to the analytical tool used. METHODS: 1,165 patients at high risk for caries underwent yearly examinations in 1996, 1997 and 1998 and randomly received one out of four tested preventive measures. Two analytical methods were used: (1) increment was calculated using the indices DMFS(a) and D(e)MFS(a), (2) incidence density was calculated for each individual using new approximal dentin lesions or dentin and enamel lesions in the numerator and the risk time for the surfaces in the denumerator. The Poisson regression with overdispersion was used for multivariate modelling of the outcomes. RESULTS: In all analyses of approximal dentine as well as dentine and enamel caries bad oral health at baseline increased the risk of new approximal dentine caries. In the incidence analysis poor oral health behaviour also increased the risk. CONCLUSION: Since the difference between the two analytical methods are small for young individuals the method of preference would be increment analysis. Copyright 2003 S. Karger AG, Basel
AIM: To use different analytical approaches to data from a longitudinal, quasi-experimental study evaluating caries-preventive measures in order to estimate the differences in results due to the analytical tool used. METHODS: 1,165 patients at high risk for caries underwent yearly examinations in 1996, 1997 and 1998 and randomly received one out of four tested preventive measures. Two analytical methods were used: (1) increment was calculated using the indices DMFS(a) and D(e)MFS(a), (2) incidence density was calculated for each individual using new approximal dentin lesions or dentin and enamel lesions in the numerator and the risk time for the surfaces in the denumerator. The Poisson regression with overdispersion was used for multivariate modelling of the outcomes. RESULTS: In all analyses of approximal dentine as well as dentine and enamel caries bad oral health at baseline increased the risk of new approximal dentine caries. In the incidence analysis poor oral health behaviour also increased the risk. CONCLUSION: Since the difference between the two analytical methods are small for young individuals the method of preference would be increment analysis. Copyright 2003 S. Karger AG, Basel