Introduction

Friction between head and cup is a primary factor for survival of total hip joint replacement (THR) and its gliding surfaces. In up to 40% of all revisions, the cup or inlay must be replaced as result of friction-induced wear [1]. Aim of the study was to measure the friction-induced temperature increase in vivo in THR and to identify possible individual parameters of influence.

Musculoskeletal loading plays an important role in the primary stability of THA. There are about 210,000 primary THA interventions p.a. in Germany. Consideration of biomechanical aspects during computer-assisted orthopaedic surgery is recommendable in order to obtain satisfactory long-term results. For this purpose simulation of the pre- and post-operative magnitude of the resultant hip joint force R and its orientation is of interest. By means of simple 2D-models (Pauwels, Debrunner, Blumentritt) or more complex 3D-models (Iglič), the magnitude and orientation of R can be computed patient-individually depending on their geometrical and anthropometrical parameters. In the context of developing a planning module for computer-assisted THA, the objective of this study was to evaluate the mathematical models. Therefore, mathematical model computations were directly compared to in-vivo measurements obtained from instrumented hip implants.

With patient-specific parameters the magnitude and orientation of R were model-based computed for three patients (EBL, HSR, KWR) of the OrthoLoad-database. Their patient-specific parameters were acquired from the original patient X-rays. Subsequently, the computational results were compared with the corresponding in-vivo telemetric measurements published in the OrthoLoad-database. To obtain the maximum hip joint load, the static single-leg-stance was considered. A reference value for each patient for the maximum hip load under static conditions was calculated from OrthoLoad-data and related to the respective body weights (BW).

On average there are large deviations of the results for the magnitude (Ø=147%) and orientation (Ø=14.35° too low) of R obtained by using Blumentritt's model from the in-vivo results/measurements. The differences might be partly explained by the supplemental load of 20% BW within Blumentritt's model which is added to the input parameter BW in order to consider dynamic gait influences. Such a dynamic supplemental load is not applied within the other static single-leg-stance models. Blumentritt's model assumptions have to be carefully reviewed due to the deviations from the in-vivo measurement data.

Iglič's 3D-model calculates the magnitude (Ø17%) and the orientation (Ø49%) of R slightly too low. For the magnitude one explanation could be that his model considers nine individual 3D-sets of muscle origins and insertion points taken from literature. This is different from other mathematical models. The patient-individual muscle origin and insertion points should be used.

Pauwels and Debrunner's models showed the best results. They are in the same range compared to in-vivo data. Pauwels's model calculates the magnitude (Ø5%) and the orientation (Ø28%) of R slightly higher. Debrunner's model calculates the magnitude (Ø1%) and the orientation (Ø14%) of R slightly lower.

In conclusion, for the orientation of R, all the computational results showed variations which tend to depend on the used model.

There are limitations coming along with our study: as our previous studies showed, an unambiguous identification of most landmarks in an X-ray (2D) image is hardly possible. Among the study limitations there is the fact that the OrthoLoad-database currently offers only three datasets for direct comparison of static single leg stance with in-vivo measurement data of the same patient. Our ongoing work is focusing on further validation of the different mathematical models.