Predictive Factors of Dental Implant Failure: A Retrospective Study Using Decision Tree Regression

Introduction: Dental implants are routinely used to replace missing teeth. Therefore, the primary aim of the present study was to assess the single-unit implant failure rate over a period of seven years from 2015 to 2021, with a minimum of two years post-implant follow-up. The secondary aim was to identify the risk factors associated with implant failure using machine learning decision tree regression and Kaplan-Meier survival analyses.

Materials and methods: An eight-year retrospective study was conducted using the clinical records of 224 patients who received single-unit dental implants between January 2014 and December 2021, where risk factors for early (EIF) and late implant failure (LIF) were identified. The patients' clinical case records and radiographs were used to assess implant failure.

Results: Smoking and peri-implantitis were principal contributors to failure (p=0.001). Implant failure was more common in males, the maxillary jaw, and posterior teeth, although these factors were not significantly associated with implant failure (p>0.05). The duration of failure was 16.87±4.6 months for LIF, in contrast to 5.71±1.38 months in EIF. Bruxism and peri-implantitis were correlated with diminished survival duration, especially when compounded by additional risk factors such as diabetes mellitus. Isolated peri-implantitis yielded an average failure duration of approximately 13.4 months, whereas bruxism intensified the failure interval to approximately 13.8 months. Kaplan-Meier survival analysis revealed that among the identified causes of failure, peri-implantitis and smoking were the predominant factors, followed by bruxism, diabetes, and complications related to osseointegration.

Conclusion: Age, sex, type of surgical procedure, sinus lift, and grafting procedures were not significantly associated with dental implant failure, whereas bruxism, peri-implantitis, lack of osseointegration, smoking, and type 2 diabetes mellitus were significant predictors.