Abstract The initial fixation of an anatomical cementless glenoid component, provided by different numbers and types of screws, and the risk of bone fracture were evaluated by estimating the bone¿implant interface micromotions and the principal strains around the prosthesis. Four different fixation configurations using locking or compression screws were tested. Estimation of the micromotions at the bone¿implant interface was performed both experimentally, using an in vitro model, and computationally, using a numerical model. Principal bone strains were estimated using the numerical model. Subject variability was included by modelling two different bone qualities (healthy and rheumatoid bone). For the fixation configurations that used two screws, experimental and modelling results found that the micromotions at the bone¿implant interface did not change with screw type. However, screw type had a significant effect on fixation when only one screw was used; in this case, a locking screw resulted in less micromotion at the bone¿implant interface compared with the compression screw. Bone strains were predicted by the numerical model, and strains were found to be independent of the screw type; however, the predicted strain levels calculated in rheumatoid bone were larger than the strain levels that may cause bone damage for most considered arm positions. Predicted bone strain in healthy bone did not reach this level. While proper initial component fixation that allows biological fixation can be achieved by using additional screws, the risk of bone failure around the screws must be considered, especially in cases of weak bone. Keywords Shoulder, arthroplasty, glenoid, cementless, screws, micromotion, failure@en

21 elbows in 18 patients with rheumatoid arthritis were treated with a Souter-Strathclyde total elbow prosthesis. 18 elbows were included in a radiostereometry (RSA) study. The aim of this clinical RSA study was to assess the three-dimensional micromotion pattern of the Souter-Strathclyde prosthesis, and thereby gain insight in the aseptic loosening process of this prosthesis. Implants were defined as at risk of aseptic loosening when the translation rate during the second postoperative year was more than 0.4 mm along one or more coordinate axes and/or the rate of rotation was more than 1° about one or more coordinate axes. Clinical examination revealed an increase in the range of motion and a marked reduction in pain. The RSA showed that 8 of 18 humeral components were at risk of aseptic loosening, although no signs of such loosening - defined as a complete radiolucent line of 2 mm or more - were found on the plain radiographs. In 7 humeral components, an anterior tilt about the transverse axis was seen that resulted in an anterior translation of the proximal tip and a posterior translation of the component's trochlea. Long-term studies of the Souter- Strathclyde prosthesis, have shown that this rotation is a specific pattern of failure in some implants. None of the ulnar components was at risk for aseptic loosening. Improvements in fixation of the Souter-Strathclyde total elbow arthroplasty should focus on the humeral component. At present, the lateral flange of the implant is enlarged to improve rotational stability about the transverse and longitudinal axes. The effect of this change in design on micromotion of the Souter-Strathclyde total elbow prosthesis will be studied in a randomized RSA study comparing the new design to the existing one.@en