Abstract
The purpose of this study was to derive a method for predicting children at high risk of caries at the 1.5 year-old dental examination in order that preventive measures might be taken.
The subjects were 286 children who received dental examinations when they were 1.5 years and 3.5 years old.
The 36 independent variables used to predict children at high risk were comprised of variables obtained from a baseline questionnaire (21 items) and variables obtained from a baseline dental examination (15 items). The variables based on the questionnaire included sociodemographic variables, parental characteristics, nutritional characteristics, oral health practice, etc.
Discriminant analysis was used to determine the discriminant function by which those variables as a whole discriminated between the true high and the non-high risk groups as defined. True high risk children were difined in two ways: the highest 25% of children based on the dmft scored at 3.5 years of age (dmft≥7) and the new dmft scores from 1.5 years to 3.5 years of age (new dmft≥6). Using the dmft score as an indicater of true high risk there were nine baseline variables, which were ‘Classification of OABC’, ‘Period of infant feeding’, ‘Sucrose intake from foods which have a long clearance time’, ‘Mother's DMFT score’, ‘Tooth brushing habits’, ‘Order of birth’, ‘BL-dmfs’, ‘Age at start of weaning’, and ‘Weight at birth’, obtained from each child for use in prediction. Then a discriminant function comprised of the nine variables was made. Children predicted to be at high risk were determined by discriminant scores which were higher than the discriminant boundary score -0.434.
The result of the discriminant analysis showed that sensitivity was 71.2% and specificity was 71.8%. It was clear that this screening model comprised of nine variables showed higher validity in prediction, and higher sensitivity than the model comprised of only the one variable, ‘Classification of OABC’ (sensitivity: 22.7%, specificity: 97.3%). The results indicate that in predicting the prevalence of dental caries, which is a disease associated with a multiplicity of facters, these collective variables increase the validity in prediction.
Using the new dmft score as an indicator of true high risk there were eight variables obtained from each child for use in prediction. A discriminant function comprised of eight variables was made. Children predicted to be at high risk were determined by discriminants scores which were higher than the discriminant boundary score -0.187. The result showed that sensitivity and specificity was 70.7% and 59.7%, respectively. It was found that the screening model using the dmft score as an indicator of true high risk was more effective than that using the new dmft model.
