Reported studies verifying the effects of offset placement include ones where implant bodies were embedded in rectangular experimental models [11–14], ones where rectangular bone models were constructed with FEA models [15, 16], and ones where FEA models were constructed from CT data on human mandibles [17, 18]. The artificial mandible models used in the present study were type II in the Lekholm and Zarb classification [24] and created using clinically valid estimates of bone quality, bone structure, and bone morphology. If the effects of implant placement alone are being verified, then models that are simple in form can be used. However, artificial mandibles were used in the present study in order to take into account the constraints on offset amounts due to anatomical morphology, the differences in stress distribution in the surrounding bone due to bone morphology, and other aspects of clinical practice.

Analysis was done with FEA models, in which the cancellous bone was simplified as being a homogeneous body. FEA models have been reported to be useful for verifying the behavior of implants when loaded [22], and so, the effects of different placements were verified with FEA models and compared with the results from the experimental models. In a previous study, Omori et al. compared the compressed displacement between experimental models and FEA models to verify the validity of the FEA models [22]. They reported that the constraint conditions in the artificial mandible bottom and differences in the actual Young’s modulus were two reasons why the two models produced different absolute values for the compressed displacement. In order to apply the actual Young’s modulus to the FEA models in the present study, FEA models with a known Young’s modulus were created and the resulting values of compressed displacement were compared with the results from the experimental models; after the appropriate Young’s modulus was found, the FEA models were created again for further analysis. In previous studies verifying the usefulness of offset placement, one set of FEA models was created and analyzed by changing the conditions or settings. Few studies used different FEA models with the same placement models [15–17]. In the present study, we carried out the same experiments with both the models to verify the validity of each analysis.