Reliability evaluation of classifying radial head fractures by the system of Mason

Inter-observer and intra-observer reliability for classifying radial head fractures by the system of Mason was analyzed. Twenty-three cases of isolated radial head fractures and twenty-five sets of corresponding AP and lateral radiographs representing these fractures were assembled. The cases were reviewed and assessed independently according to the system of Mason by twenty practicing orthopedic surgeons. On two occasions, the inter-observer and intra-observer variation was analyzed by standard unweighted Kappa statistics. In both observations, complete agreement was seen in only 16% of the cases. Kappa statistic values indicated that 69% of the cases at first observation and 45% of the cases at second observation suggest moderate to poor agreement. Intra-observer agreement between the first and second observation was graded fair to poor in 60% of the cases. Individual observer consistency was, on average, only 78% (range 60% to 92%). The demonstrated wide degree of variation suggests that the Mason classification is unreliable.