This paper summarizes a series of investigations to determine whether biomechanical simulations can be used to plan scoliosis surgery by predicting the outcome as a function of choices available to the surgeon. In order to test this idea, simulations have initially been performed with prior knowledge of the real outcome (from postoperative radiographs). In addition, intra-operative measurements of the spine shape and displacement of instrumentation have been used to improve documentation of the influence of different steps in the surgical procedure. Several difficulties have emerged which have given new understanding of both the mode of action of surgical procedures, as well as the biomechanical modeling. 1. Important reduction of the Cobb angle occurs in the anesthetized patient compared to pre-operative (standing) radiographs. 2. For an elastic analysis the stiffness of the spine and rib cage should be known for each anatomical level of each patient. 3. Shape changes in the non-instrumented part of the spine are mainly due to the patient and their muscles, not the instrumentation directly, and are difficult to analyze biomechanically. 4. The large difference in stiffness between the stiffest and the least stiff parts of the model, as well as the large displacements which occur in surgical procedures, create technical difficulties in simulations using finite element modeling. 5. Segmental instrumentation offers surgeons many variables and multistep maneuvers to adapt to individual patients’ needs, but conversely create many unknown inputs for the biomechanical analyses, and difficulties in validation of model predictions.

It is concluded diat in order to be a reliable tool to assist with pre-operative planning, deterministic modeling of scoliosis surgery will require more information to formulate the models, better specification of inputs, and improved analysis tools.