The short-term clinical failures of cementless stems, such as proximal bone resorption and thigh pain, have limited their acceptance despite the potential long-term benefits of biological fixation for younger and more active patients. The clinical problems are believed to be related to failure at the bone-stem interface. By optimizing the design of the prosthesis to prevent or minimize failure at the interface, it should be possible to reduce the clinical problems. In analytical and experimental studies, it has been shown that surface coating distribution is an important design parameter which can affect interface mechanics. The main objective of this study was to experimentally optimize the surface coating distribution, where the optimal distribution was a reasonable compromise between two of the mechanisms which are responsible for interface failure, namely proximal stress shielding and distal relative motion.

An unique experimental model was developed to study the effect of varying the surface coating distribution on stress shielding and relative motion. The experimental model consisted of a synthetic femur, a titanium alloy prosthesis and cyanoacrylate gel to simulate bone-stem bonding. Validation tests were conducted to determine the moduli of the synthetic femur components, the friction between the stem and the bone, the bonding strength of the glue and the controllability of the bonding distribution. The results of these tests confirmed that the experimental model was adequate for the purposes of this study. However, the moduli of the synthetic femur components are low compared to human bone and, therefore, the model should only be used for intercomparison studies. With this prosthesis, bonding of the distal 20%-30% of the prosthesis was not possible, because of inadequate stem-bone contact in that region.

Three synthetic femurs were tested with six bonding distributions per bone. Each bone was subjected to simulated loads of single leg stance and stair climb. Surface strains and stem-bone relative motion were measured to indicate stress shielding and relative motion, respectively. The experimental results were compared with the results of a finite element analysis.

The surface strains at the proximal and mid-stem levels of the femur varied with the bonding distribution. The maximal axial surface strains tended to decrease as the extent of the bonding was increased to 65% of the stem length. In other words, the stress shielding increased as the bond was extended distally. With bonding beyond 65%, the strains remained the same or decreased slightly.

Relative motion of the targets at the proximal level (where the stem and bone were bonded) were typically less than 20 ýtm. At the distal tip of the stem, the motion decreased with more distal bonding to a minimum with 81% bonding. With full coating, some relatively large motions were observed which is consistent with the absence of distal bonding. Calculated interface motions were 1.5 to 3.4 less than target motions.

There was generally good agreement between the trends observed experimentally and those predicted by the finite element analysis. As well, there was good agreement in the magnitudes of the surface strains. The most obvious discrepancies between the FEA and the experiment were the differences in strain and motion with full coating. Thus the FEA further supports the experimental observation that full bonding could not occur.

Although general trends were observed and reported, the small sample size limited the statistical power of the results. More samples are required before statistically significant conclusions can be made. Nonetheless, the trends in the data suggest that the optimal distribution (which minimizes the combination of stress shielding and relative motion) is to coat the proximal 65% to 81% of the stem length. More generally, this study experimentally demonstrated the complicated interaction between relative motion and surface strains with variation in bonding distribution. The conflicting effects of varying the surface coating distribution on these two parameters confirm that an optimization approach is necessary to determine the most suitable coating distribution, where the optimal solution corresponds to a compromise between the parameters.