Towards unbiased and accurate simulations of screw-bone constructs with homogenized FE models

Abstract

Homogenized finite element (hFE) models are increasingly used to investigate internal fixation of complex bone fractures with plates and screws. However, using hFE models for this purpose has also been criticized, especially if the screws are anchored in trabecular bone [1]. That is due to the many inherent simplifications, such as modelling trabecular bone as a continuum, and simplifying the screw thread geometry. Validating hFE models of screw-bone constructs based on experiments is challenging, as a lot of error sources are mixed and can hardly be separated (e.g. error due to homogenization, screw insertion damage, screw-bone contact). This might lead to biased models and could explain why some studies report good agreement with experimental results [2], while others do not [1]. In this study, we assessed hFE model accuracy using μFE models rather than experiments as a reference to delineate and control different error sources. As a first step, this study was limited to osseointegrated screws to exclude the effect of screw insertion damage and contact. hFE models were built based on “best evidence” without parameter tuning to avoid model bias.