Three-dimensional simulation of Harrington distraction instrumentation for surgical correction of scoliosis.

Harrington distraction rod surgery on six female patients with idiopathic scoliosis was simulated in three-dimensional osseoligamentous finite element models with individual geometry taken from preoperative stereo roentgenographic reconstructions of the spine and ribcage and compared with the measured outcome. Boundary conditions at the ends of the spine were used to maintain pelvis and head alignment. Published material and flexibility properties were used. The amount of hook distraction was calculated from measured changes in the distance between the hook sites (range, 13-27 mm). Initial simulations underestimated the Cobb angle correction by an average 6%. They underestimated the spinal elongation by 36% and predicted an average 12 degrees increase in kyphosis angle compared with an actual 10 degrees average decrease. Agreement for sagittal plane changes improved in five cases when the beams representing the motion segments were displaced posteriorly. In the sixth case (with the rod applied over a lordotic spinal region), agreement was improved with the motion segment beams displaced anteriorly. The amount of the beam displacement that gave the best agreement was variable, and we were not able to predict it for each individual. Both measured and simulated changes in vertebral transverse plane rotations and in rib angulations were small. The greatest source of errors in these simulations appeared to be inadequate representation of in vivo motion segment behavior by in vitro measured stiffness properties.