A longitudinal retrospective study of the analysis of the risk factors of implant failure by the application of generalized estimating equations

Purpose: Many studies have identified risk factors for dental implant failure, although few have investigated the correlation among implant fixtures within single patients. A better analytical method may include repeated measures analysis including generalized estimating equations (GEE). This retrospective cohort study aimed to (1) identify the risk factors for failure of dental implantation and (2) evaluate an analytical method using GEE analysis. Methods: We analyzed data on early and late implant failures in 296 patients providing 721 rough surface dental implants (2.44 implants per patient). Potential predictors of implant failure included age, gender, smoking, location of implant, use of bone augmentation, number of remaining teeth, opposing tooth condition, fixture length, fixture diameter and type of suprastructure (fixed or removable partial denture). The likelihood of early and late implant failure was estimated by GEE. Results: The early failure rate was 1.5% (11/721 implants, 7/296 patients) and the 10-year cumulative survival rate was 94.0% (7/710 implants, 5/293 patients). The GEE analysis revealed that a significant risk factor for early implant failure was smoking (p < 0.01), whereas significant risk factors for late failure were maxillary implant (p = 0.02), posterior implant (p < 0.01), number of remaining teeth (>= 20) (p<0.01), opposing unit being a removable partial denture or nothing (p = 0.04) and having a removable type suprastructure (p<0.01). Conclusions: GEE analysis showed that smoking was a risk factor for early implant failure, and several risk factors were identified for late implant failure. (C) 2015 Japan Prosthodontic Society. Published by Elsevier Ireland. All rights reserved.