The mechanical responses of the human cadaver cervical spine to torsional loading have been investigated using a dynamic test system. The kinematic conditions have been proscribed to recreate the in vivo loading environment. Load free changes in axial length with rotation, and the determination of an appropriate axis of rotation were required to generate meaningful data.

Eight channels of force, moment, and angular deflection have been quantified. Relaxation testing was used to develop the parameters of a quasi-linear viscoelastic constitutive relation with a continuous relaxation spectrum. While this model showed high predictive ability there were some limitations caused by the inability to perform an infinite rate ramp and hold rotation (a requirement of a relaxation test). The detrimental effects of the finite rate ramp and hold rotation were significantly reduced by an extrapolation deconvolution technique. In this technique, the relaxation function was extrapolated to true zero time, and the measured pseudoelastic torsional modulus was deconvolved of intrinsic relaxation in the Laplace domain. The resulting model showed significantly greater predictive ability and an improved representation of viscous effects.

The role of torsional loading in cervical injury was also investigated. Torsional loads applied to the base of the skull produced atlantoaxial rotary facet dislocation. Torsional loads applied directly to the lower cervical spine resulted in unilateral facet dislocation;​ however, the torque required to produce this injury was significantly greater than the torque to failure in the atlantoaxial joint (24%, p < 0.05). It was therefore concluded that torsional loads applied directly to the base of the skull do not mediate lower cervical ligamentous injuries or dislocations.