Cost comparison of prediction-based decision-making for periodontally affected molars.

Aim: Tooth loss prediction could improve decision-making for periodontally affected molars. We analysed the costs of removing all, none or only those molars predicted to be at-risk for extraction during supportive periodontal therapy (SPT).

Methods: This is a model-based study using a German private payer perspective. Building mainly on cost and effectiveness data from 2039 molars (301 patients) which received long-term SPT, five (exemplary) prediction/decision strategies were compared: Retaining all molars, removing molars with furcation involvement (FI) ≥ I/ ≥ II/ III, or removing all molars. Each strategy came with different proportions of molars correctly or erroneously (false positively) removed. Retaining as well as removing molars had long-term cost implications (via SPT or tooth replacement). Using Monte-Carlo microsimulations, we estimated the lifetime costs of an average population in our cohort and evaluated what accuracy a prediction method needed to have to save costs in different risk groups.

Results: Removing only molars with FI III (1188 Euro) and removing no molars (1195 Euro) were significantly less costly than removing all molars (1454 Euro). Prediction methods needed to be highly specific in most populations to avoid unnecessary tooth loss and the associated high costs.

Conclusions: Removing molars prior to SPT should be decided cautiously.