Finite element method-based study of pedicle screw–bone connection in pullout test and physiological spinal loads

Finite element (FE)method has been widely used to study the screw–bone connections. Screw threads are often excluded from the FE spine model to reduce computational cost. However, no study has been conducted to compare the effect of such simplification in the screw models on the predicting accuracy of the model. The effects of different screw–bone connection types on the overall spine biomechanics are also unknown. In this study, three different types of screw–bone connections were compared using FE simulations in this study: (1)screw and bone are not fully bonded (contact connection); (2)screw is rigidly bonded with the bone (bonded connection); and (3)simplified-geometry-rigid (SGR)connection. Screw pullout test and physiological spinal loading test were simulated for the screws in this study: (1)pullout test where the pedicle screws were inserted in polyurethane foam; and (2)physiological spinal loading test (flexion, extension, lateral bending, and axial rotation)where the screws were fused into previously-validated FE lumbar spine model. The FE spine model used in this study included L ₁ –L ₅ spine levels and simulated major ligaments and resultant muscle forces. This study indicated that the holding capability in the screw–bone interaction is smaller and the bone and implants are subjected to larger von Mises stress (up to 44.88%)in the contact connection than those in the bonded connection. Among the four spinal loading cases tested in this study, flexion produced the highest von Mises stress in both the bone and the implants. Considerable differences were observed between simplified and non-simplified screw FE models in the von Mises stress at screw–bone contact region within spinal loading environment and the ultimate screw pullout strength in pullout test. This study concluded that both the spinal implants and the bone are subjected to higher stress immediately after the pedicle-screw-instrumented surgery and before the screw and bone are fully bonded. The screw–bone interface is less likely to fail after the screw and bone are fully bonded. SGR screw model is able to predict screw force and rod stress that are consistent with those predicted by non-simplified screw models.