1. Utility technology in the assessment of the cut-off between a negative and a positive test in a caries prediction model
- Author
-
C. van Loveren, Ewald M. Bronkhorst, W.H. van Palenstein Helderman, M.A. van 't Hof, and Cariologie/EPT (OUD, ACTA)
- Subjects
Caries prediction ,Caries incidence ,Dental Caries ,Risk Assessment ,Sensitivity and Specificity ,Statistics ,Effective Primary Care and Public Health [EBP 3] ,Econometrics ,Dental Caries Activity Tests ,Humans ,False Positive Reactions ,Positive test ,Longitudinal Studies ,Child ,General Dentistry ,False Negative Reactions ,Mathematics ,Netherlands ,Models, Statistical ,Incidence ,Regression analysis ,Effective primary care and public health [NCEBP 7] ,Explained variation ,Test (assessment) ,Regression Analysis ,Cut-off ,Caries experience - Abstract
Item does not contain fulltext The methodology for the assessment of a negative or positive test in caries prediction models has not received much attention. This study aims to explain how utility technology could be applied in a caries prediction model for the assessment of the cut-off between a negative and a positive test. In this study loss of utilities was assigned to false outcomes of the prediction test. A regression equation with past caries experience variables of 11-year-old children and caries increment in the 11- to 15-year forecast period as outcome was computed. The adjusted explained variance for caries increment was 0.45. Formulas were constructed for the loss of utilities for false-negative tests whereas false-positive tests were corrected with a fixed loss of utilities. Each 11-year-old child of the 252 children was screened at various points of the regression equation. Loss of utilities was calculated for each child on the basis of the validation criterion, the outcome of the test and the actual caries increment of the child. The point on the regression equation with the least loss of utilities for the whole group was taken as the cut-off between a negative and a positive test. If the validation criterion for the prediction period was set on no caries, the prediction model resulted in a sensitivity of 84% and a specificity of 73%. This prediction model has potential when caries incidence is low and resources limited.
- Published
- 2007
- Full Text
- View/download PDF