RESUMEN
AIM: To assess the ability of two-way interactions between baseline stage, grade and extent to predict tooth loss due to periodontitis (TLP) over a long-term follow-up period. MATERIALS AND METHODS: Patients treated for periodontal disease with a complete medical history, baseline periodontal chart, full mouth radiographs and a minimum of ≥10 years follow-up were recruited. Supportive periodontal therapy (SPT) visits were recorded during the entire follow-up period. Patients were categorized according to their stage, grade and extent. The absolute survival at 10-, 20-, and 30-year follow-up was calculated for TLP. Kaplan-Meier survival curves were plotted at the tooth-level and multilevel Cox regression frailty models were constructed in order to assess the association among predictive variables and TLP by taking into account the hierarchical patient-teeth structure. RESULTS: 442 patients (11,125 teeth) with a mean follow-up of 23 years met the inclusion criteria and were included in this study. The most prevalent diagnosis at baseline was stage III grade B (30.3%), followed by stage II grade B (23.5%). Among the parameters analysed, stage and grade were found to be the best predictors of TLP. Statistically significant differences were observed for extent only in patients with severe disease (stage IV or grade C). The multilevel Cox regression analysis demonstrated that patients with higher concomitant baseline staging and grading developed greater TLP over the follow-up period. CONCLUSIONS: Higher concomitant staging and grading corresponded to greater risk for TLP and generalized extent only became a significant predictor in patients with stage IV or grade C disease.