Introduction and Aims: To determine differences in metal memory, at physiological temperatures, between 6mm stainless steel and titanium rods as a function of construct failure in scoliosis surgery.

Method: Different length Ti and SST rods were contoured at room temperature with a radius of curvature of 30cm and incubated at 37±2°C. Rods were photographed biweekly over graph paper with a digital camera. The images were processed using Jandel Sigma Scan. A best-fit regression polynomial was used to calculate the average curvature. After 36 weeks, the curvature of each rod was normalised against the initial curvature and plotted against time, with a linear regression performed to assess changes in curvature, expressed as percent of change per year.

Results: Changes in both SST rods and pre-bent Ti rod were within measurement error (0.52% increase for long SST, 0.26% decrease for short SST). In contrast, both manually bent Ti rods changed markedly (decreases of 6.76% and 5.2% for long and short Ti rods respectively).

Conclusion: Continuous physiologic heat conduction may contribute to a loss of curvature in Ti rods due to memory properties. The ideal implant should retain the intended contour. Ti rods, subjected to physiologic heat, lost correction. The use of custom contoured Ti rods for the surgical correction of spinal deformities should be questioned.