Effects of motion segment simulation and joint positioning on spinal loads in trunk musculoskeletal models

Musculoskeletal models represent spinal motion segments by spherical joints/beams with linear/nonlinear properties placed at various locations. We investigated the fidelity of these simplified models (i.e., spherical joints with/without rotational springs and beams considering nonlinear/linear properties) in predicting kinematics of the ligamentous spine in comparison with a detailed finite element (FE) model while considering various anterior-posterior joint placements. Using the simplified models with different joint offsets in a subject-specific musculoskeletal model, we computed local spinal forces during forward flexion and compared results with intradiscal pressure measurements. In comparison to the detailed FE model, linearized beam and spherical joint models failed to reproduce kinematics whereas the nonlinear beam model with joint offsets at -2 to +4mm range (+: posterior) showed satisfactory performance. In the musculoskeletal models without a hand-load, removing rotational springs, linearizing passive properties and offsetting the joints posteriorly (by 4mm) increased compression (∼32%, 17% and 11%) and shear (∼63%, 26% and 15%) forces. Posterior shift in beam and spherical joints increased extensor muscle active forces but dropped their passive force components resulting in delayed flexion relaxation and lower antagonistic activity in abdominal muscles. Overall and in sagittally symmetric tasks, shear deformable beams with nonlinear properties performed best followed by the spherical joints with nonlinear rotational springs. Using linear rotational springs or beams is valid only in small flexion angles (<30°) and under small external loads. Joints should be placed at the mid-disc height within -2 to +4mm anterior-posterior range of the disc geometric center and passive properties (joint stiffnesses) should not be overlooked.