Factors associated with failure of surface-modified implants up to four years of function

Objectives: The relative impact of innovative treatment concepts on the failure of surface-modified implants is not well-understood. This retrospective study aimed to explore this using data obtained in a university postgraduate training centre. Material & methods: Patients treated with implants for a variety of indications over a 3-year period were included. All implants had been at least one year in function. Clinical records were evaluated for implant failure and in reference to implant length/diameter/location, time from tooth loss to implant placement, bone condition (native/grafted), surgical protocol (two-/one-stage), loading protocol (delayed/early/immediate), type of prosthesis (removable/fixed), surgeon’s experience level (resident/trainee) and specialty (periodontist/oral surgeon). The impact of each covariate on failure was tested using the Fisher’s exact test. Kaplan-Meier survival functions were constructed and Mantel-Cox logrank tests were used to compare survival functions. To correct for possible interaction Cox proportional Hazards regression was adopted. Results: 41/1180 (3.5%) implants were lost in 34/461 (7.4%) patients (245 ♀, 216 ♂; mean age 51, range 18-90). Factors showing significant impact on failure on the basis of univariate analyses were implant location (p=0.015), surgical protocol (p=0.002), loading protocol (p=0.002), surgeon’s experience level (p=0.035) and specialty (p=0.001). When controlling for other covariates, only the loading protocol had a significant influence (p=0.049) with early loading more prone to failure (p=0.014) when compared to delayed loading. Immediate loading and delayed loading showed comparable implant survival (p=0.311). Conclusions: Implant therapy may be highly successful in a training centre where inexperienced clinicians are strictly monitored and personally guided. Implant specific variables do not affect implant survival but early loading is a risk indicator for implant failure, whereas immediate loading is not