Correlation of the experimental and numerical results for the holding power of dental, traumatic, and spinal screws

The holding power of the bone–screw interfaces is one of the key factors in the clinical performance of screw design. The value of the holding power can be experimentally measured by pullout tests. Historically, some researchers have used the finite-element method to simulate the holding power of the different screws. Among them, however, the assumed displacement of the screw withdrawal is unreasonably small (about 0.005–1.0 mm). In addition, the chosen numerical indices are quite different, including maximum stress, strain energy, and reaction force. This study systematically uses dental, traumatic, and spinal screws to experimentally measure and numerically simulate their bone-purchasing ability within the synthetic bone. The testing results (pullout displacement and holding power) and numerical indices (maximum stress, total strain energy, and reaction forces) are chosen to calculate their correlation coefficients. The pullout displacement is divided into five regions from initial to final withdrawal. The experimental results demonstrate that the pullout displacement consistently occurs at the final region (0.6–1.6 mm) and is significantly higher than the assumed value of the literature studies. For all screw groups, the measured holding power within the initial region is not highly or even negatively correlated with the experimental and numerical results within the final region. The observation from the simulative results shows the maximum stress only reflects the loads concentrated at some local site(s) and is the least correlated to the measured holding power. Comparatively, both energy and force are more global indices to correlate with the gross failure at the bone–screw interfaces. However, the energy index is not suitable for the screw groups with rather tiny threads compared with the other specifications. In conclusion, the underestimated displacement leads to erroneous results in the screw–pullout simulation. Among three numerical indices the reaction-force is the optimal index for the screw–pullout problem.