The use of logistic discrimination and receiver operating characteristics (ROC) analysis in dentistry

Logistic regression is a statistical method which allows an investigator to 'explain' or 'predict' a binary response variable from a set of independent variables. In particular, it may be used to classify persons for example, as diseased or healthy, high risk or low risk etc. (logistic discrimination). During recent years this method has been of increasing interest and importance in dentistry. Since this demanding statistical method may not be easily accessible to dentists, a description is provided of its basic characteristics in an introductory and condensed form. A worked example, 'identification of children with high caries risk', is presented in order to demonstrate the application and use of the method. Following the presentation of the logistic model, evaluation of the performance of a classification model by means of 'receiver operating characteristic (ROC) analysis' is demonstrated. The presentation of these statistical tools here, puts more weight on verbal explanation and on graphical representations than on mathematical details. This statistical tools here, puts more weight on verbal explanation and on graphical representations than on mathematical details. This may help to make the methods accessible to readers who have insufficient time to study a more comprehensive discourse