RESUMEN
AIM: The goal of the present investigation was to identify site-level factors that might allow prognostication of individual implants in partially dentate patients with multiple non-splinted restorations. METHODS: We analysed clinical and radiographic characteristics of 222 non-splinted single implants in function for at least 5 years in 86 partially dentate individuals at the time of functional loading and at follow-up, with the outcome variable being peri-implantitis. Principal component analysis identified factors contributing to greatest variability and linear discriminant analysis coupled with Random Forest Classifier used to identify risk predictors. RESULTS: After controlling for patient-level factors, the following characteristics were associated with significantly increased risk for peri-implantitis: Periodontal disease on adjoining teeth at the time of restoration (Odds Ratio (OR): 8.0), implant placement at a depth of 6 mm or more in relation to the CEJ of adjacent tooth (OR: 8.5), asymmetric prosthesis (OR: 4.3), history of tooth loss due to periodontitis (OR: 2.4) and a mean baseline plaque index of 1.6 or more (OR: 7.9). CONCLUSIONS: Our findings suggest that a system that incorporates both subject level and implant-level factors is required to effectively prognosticate the success of individual implants.