Descriptive models of restorative treatment decisions

Area of Science:

• Dental Public Health
• Clinical Decision Making
• Biostatistics

Background:

• Understanding dentists' decision-making processes for restorative treatments is crucial for various applications.
• Descriptive models can aid in personnel planning and assessing dental treatment program effectiveness.

Purpose of the Study:

• To develop descriptive models predicting dentists' restorative treatment decisions for individual teeth.
• To assess the utility of these models in estimating treatment probabilities and normative treatment needs.

Main Methods:

• Logistic regression was employed to model the probability of restorative treatment for molar, premolar, and anterior teeth.
• Models utilized data from oral examinations, questionnaires, and dentists' treatment plans.
• Model accuracy was evaluated against actual treatment decisions in a separate community sample.

Main Results:

• Models achieved kappa values of 0.60-0.65 on the original dataset, indicating reasonable accuracy.
• Accuracy decreased significantly when predicting treatment in a community sample, with kappa values ranging from 0.10-0.20.
• The models demonstrated limitations in predicting dichotomous treatment interventions by individual dentists.

Conclusions:

• Models based on clinical and nonclinical data can estimate individual tooth treatment probability with moderate accuracy.
• The approach shows potential for developing measures of normative dental treatment needs.
• Inaccuracies in prediction may stem from sample differences and individual dentist variations in decision-making.