AIM: To develop and internally validate a nomogram built on a multivariate prediction model including parameters from the new classification of periodontal diseases, able to predict, at baseline, the occurrence of tooth loss due to periodontal reason (TLP). MATERIALS AND METHODS: A total of 315 individuals diagnosed with periodontal disease and receiving a minimum of one annual supportive periodontal therapy visit were included in the study. Patients were staged and graded based upon baseline data. The population was divided into a development (254 patients) and a validation (61 patients) cohort to allow subsequent temporal validation of the model. According to the TLP at the 10-year follow-up, patients were categorized as "low tooth loss" (≤ 1 TLP) or "high tooth loss" (≥ 2 TLP). Bootstrap internal validation was performed on the whole data set to calculate an optimism-corrected estimate of performance. RESULTS: The generated nomogram showed a strong predictive capability (AUC = 0.81) and good calibration with an intercept = 0 and slope = 1. These findings were confirmed by internal validation using bootstrapping (average bootstrap AUC = 0.83). CONCLUSIONS: The clinical implementation of the present nomogram guides the prediction of patients with high risk of disease progression and subsequent tooth loss for personalized care.
AIM: To develop and internally validate a nomogram built on a multivariate prediction model including parameters from the new classification of periodontal diseases, able to predict, at baseline, the occurrence of tooth loss due to periodontal reason (TLP). MATERIALS AND METHODS: A total of 315 individuals diagnosed with periodontal disease and receiving a minimum of one annual supportive periodontal therapy visit were included in the study. Patients were staged and graded based upon baseline data. The population was divided into a development (254 patients) and a validation (61 patients) cohort to allow subsequent temporal validation of the model. According to the TLP at the 10-year follow-up, patients were categorized as "low tooth loss" (≤ 1 TLP) or "high tooth loss" (≥ 2 TLP). Bootstrap internal validation was performed on the whole data set to calculate an optimism-corrected estimate of performance. RESULTS: The generated nomogram showed a strong predictive capability (AUC = 0.81) and good calibration with an intercept = 0 and slope = 1. These findings were confirmed by internal validation using bootstrapping (average bootstrap AUC = 0.83). CONCLUSIONS: The clinical implementation of the present nomogram guides the prediction of patients with high risk of disease progression and subsequent tooth loss for personalized care.