Meta-Analysis of Three Implant Experiments in the Sheep Mandibular Model

Methods: Animals were sacrificed after 3 months unloaded healing. Computer-assisted and stereological HMA techniques were used. Implant surfaces were described semi-quantitatively by SEM. Implant types were ranked by percent bone-implant contact (%BIC). Variables including experiment year, implant surface, site/position and measurement technique were analysed using random effects GLS regression. Outcome measures were tested for significance after adjusting for all other variables, with machined-surface implants as the reference.

Results: Of 132 implants placed in 57 sheep in 3 experiments over 3 years, 90.2% survived. Rank-order of implant surfaces by %BIC, from least to most: machined Conclusions: Regression analysis suggested that observed differences in bone-implant contact between unloaded implants in this model, grouped either by implant types or by surface treatment, represents real differences between these implants. %BIC in this animal model appears to be relatively independent of other variables. The rank-order for implants with machined, roughened or HA surfaces in sheep was similar to published results for human patients and may be useful for distinguishing between unloaded implants with different surface treatments.

Supported by: Dentsply (USA); SteriOss (NZ); Noble Biocare Australia; Southern Implants, South Africa; ITI Foundation; Otago Research Committee grant; Professor Lang, University of Berne; Professor Persson, University of Washington.

warwick.duncan@dent.otago.ac.nz