Interobserver variation using the AO/ASIF classification of long bone fractures.

A comprehensive classification system has been proposed by the AO/ASIF Foundation for the classification of long bone fractures. After an explanatory talk and with the aid of an illustrated pamphlet, 18 orthopaedic surgeons were asked to classify 10 long bone fractures according to the AO system. Three of the participating surgeons had previous experience of the classification system. After individual classification, a consensus classification was derived and the results of the individual and consensus codings were compared. Only 32 per cent of all codings agreed with the final consensus. There was no difference between the surgeons with previous experience of the system (66 per cent) and novice coders (69 per cent) in the number of inaccurate codes when compared with the consensus codes. Reasons for error in coding are discussed. It is recommended that if the AO system is used for the purposes of research and computer-based audit, a consensus of opinion is used as the basis of classification.