Inaccuracy in selection of massive bone allograft using template comparison method

The use of massive bone allografts is increasing year by year and selection method remains unchanged. Superposition of patient's radiograph over allograft image and comparison of distances is the gold standard. Experiment was led to test selection procedure of a major european tissue bank. Four observers were asked to select an allograft for 10 fictive recipients. Nine allografts were provided. To simulate a perfect allograft, recipient himself was inserted in the pool of allografts (trap graft). The 10 potential bone transplants were classified in four categories (from adequate to unacceptable). In addition, observers were asked to choose the three best grafts for a given recipient. Quadratic kappa measuring agreement on classification between two observers ranged between 0.74 (substantial) and 0.47 (moderate). Trap graft was quoted by observers as adequate four times (10%) and was cited eight times (20%) among the three best matching allografts. None of the observers discovered that recipient was among allograft panel. This study demonstrates that current selection method is inaccurate for hemipelvic allograft selection. New methods should be developed and tested to assist tissue banks in bone allograft selection.