Shihoko Sakuma1, Muneto Nakamura, Hideo Miyazaki. 1. Division of Preventive Dentistry, Department of Oral Health Science, Graduate School of Medical and Dental Science, Niigata University, Japan. shihopre@dent.niigata-u.ac.jp
Abstract
OBJECTIVES: The purpose of this study was to construct prediction models for 1.5-year-old children who were at risk for the development of caries. METHODS: The participants were 5,107 Japanese children who resided in 21 municipalities in the Shizuoka prefecture and received oral health examinations at both 1.5 and 3 years old. General practitioners in the respective municipalities conducted the examination and the evaluation of plaque deposit. At the examination at 1.5 years old, the guardians were asked to complete a questionnaire concerning the child care environment and the health habits of the children and themselves. They were divided into four groups based on the population of the concerned municipality. Approximately 30 percent of the subjects in each group were found to be in the high-risk category (deltadft > or = 1 during both examinations). A stepwise multivariate logistic regression analysis in each group was performed using high-risk subjects as the dependent variable and 28 independent variables from the examination at 1.5 years old and the questionnaire. RESULTS: A significant model could be constructed for each group using eight to 14 independent variables, respectively. The variables common to more than three groups are related to the child's characteristics, especially sugar consumption, feeding, and caries experiences. The predictive ability of the models had a sensitivity of 58.0 to 64.0 percent and a specificity of 64.1 to 70.5 percent. Because the negative predictive value was 77.4 to 80.5 percent, these models might be considered to be in the permissible range. CONCLUSIONS: The models could not predict caries risk but may be able to predict those cases who will not develop caries.
OBJECTIVES: The purpose of this study was to construct prediction models for 1.5-year-old children who were at risk for the development of caries. METHODS: The participants were 5,107 Japanese children who resided in 21 municipalities in the Shizuoka prefecture and received oral health examinations at both 1.5 and 3 years old. General practitioners in the respective municipalities conducted the examination and the evaluation of plaque deposit. At the examination at 1.5 years old, the guardians were asked to complete a questionnaire concerning the child care environment and the health habits of the children and themselves. They were divided into four groups based on the population of the concerned municipality. Approximately 30 percent of the subjects in each group were found to be in the high-risk category (deltadft > or = 1 during both examinations). A stepwise multivariate logistic regression analysis in each group was performed using high-risk subjects as the dependent variable and 28 independent variables from the examination at 1.5 years old and the questionnaire. RESULTS: A significant model could be constructed for each group using eight to 14 independent variables, respectively. The variables common to more than three groups are related to the child's characteristics, especially sugar consumption, feeding, and caries experiences. The predictive ability of the models had a sensitivity of 58.0 to 64.0 percent and a specificity of 64.1 to 70.5 percent. Because the negative predictive value was 77.4 to 80.5 percent, these models might be considered to be in the permissible range. CONCLUSIONS: The models could not predict caries risk but may be able to predict those cases who will not develop caries.
Authors: Tariq Ghazal; Steven M Levy; Noel K Childers; Barbara Broffitt; Gary Cutter; Howard W Wiener; Mirjam Kempf; John Warren; Joseph Cavanaugh Journal: J Public Health Dent Date: 2014-09-11 Impact factor: 1.821