Supra-condylar humerus fractures (SCHF) are amongst the most common fractures requiring surgical stabilisation in the pediatric age group (1). Closed reduction and percutaneous fixation with Kirschner wires (KW) is currently the standard of care (2). The number of KW used and their configuration has been the subject of much research (3, 4). The failure modes leading to loss of fracture reduction are not clear and have not been quantified. The aim of this study is to compare the mechanical stability of the opt-used configurations for various loading modes and contact interactions at the KW/bone interface.

A Gartland type-III SCHF was introduced to a fourth generation composite saw bone (Sawbones®, Vashon, Washington, USA). The model was CT scanned with a slice spacing of 0.5mm and pixel size 0.3×0.3mm. The CT data set was imported into AmiraDev (AmiraDev 5.2 Visage Imaging, Inc). A uniaxial mechanical test was conducted in order to measure the KW pullout forces from the distal humerus.

A model of the fractured humerus was constructed with the following steps: 1) manual segmentation; 2) surface generation of each fragment, and; 3) automatic volumetric grid generation for each fragment. The fracture was then virtually reduced and KWs were placed at the desired configurations (Fig 1a-b). For each configuration, a separate model was generated. Material properties were assigned to the bone-model elements according to the manufacturer's data sheet; Young's modulus E = 16GPa and E = 150MPa for the cortical and cancellous bone respectively. The KW were assigned a Young's modulus of 200GPa. Each of the models created in Amira was imported to a finite element application (Abaqus 6.9, DS-Simula) for structural analysis. For each of KW configuration four different torque forces load types were simulated (Fig 1c left): 1) a clockwise and counterclockwise torque with a magnitude of 1.5 NM (Newton/Meters); 2) a translational force with a magnitude of 30 N (Newtons) in the direction of the humerus shaft, and; 3) a shear force with a magnitude of 30 N in the direction parallel to the fracture plane. The results were normalised such that the maximum displacement for the crossed pin configuration with a coefficient of friction equal to zero (μ = 0) was used as unity for each load configuration. Similarly, for each of KW configuration four different translational forces load types were simulated (Fig 1c right): 1) a clockwise and counter clock-wise torque with a magnitude of 1.5 NM (Newton/Meters); 2) a translational force with a magnitude of 30N in the direction of the humerus shaft, and; 3) a shear force with a magnitude of 30N in the direction parallel to the fracture plane. The results were normalised as described above.