nearly significantp < 0.05
We developed a multivariate logistic regression model that included the following candidate variables that were identified as being significant or nearly significant in the bivariate analyses ( p < 0.05): age, diabetes, implant system, and reasons for implant removal.