Validation of model predicting furcation involvement in newly crowned teeth-A 5-year retrospective follow-up

Background: This study aimed to perform a prediction model validation for furcation involvement (FI) risk in molars receiving a new fixed prosthesis (FP) using a unique cohort assessed at three time points. Methods: Following the Oral Health Statistical (OHStat) reporting guidelines, this cohort study examined 181 patients (203 molars) from 2018-2023. Teeth without FI were followed longitudinally post-crown placement at 1- (T1), 3- (T2), and 5-years (T3). A logistic regression model was built in order to predict FI and the related performance was assessed through metrics like AUC, sensitivity, specificity, and calibration. Results: FI was observed in 4.43% of teeth at 1 year, increasing to 21.67% at 3 years and 28.57% at 5 years. Univariate analysis revealed significant predictors at 3-5 years: a history of periodontitis was associated with higher FI risk at 5 years (RR = 3.56, p = 0.024), with advanced stages also increasing risk-stage III: RR = 2.59 at 3 years and RR = 3.32 at 5 years; stage IV: RR = 3.76 at 3 years and RR = 3.75 at 5 years. Short root trunks significantly increased FI risk across all intervals (1 year: RR = 3.96; 3 years: RR = 6.08; 5 years: RR = 4.75). Medium trunks did not differ significantly from long trunks. The predictive model performed best at 3 years (AUC = 0.81, sensitivity = 0.79, specificity = 0.87) and remained robust at 5 years (AUC = 0.76, sensitivity = 0.69, specificity = 0.90). Conclusion: The predictive model demonstrated high accuracy with a substantial ability to identify FI cases over time. Clinicians should consider such an assessment before crown or bridge restoration, with particular caution in patients with periodontitis.