Abstract
By means of the Pettersson scoring system, the degree of severity of haemophilic arthropathy in 128 knee, elbow and ankle joints was estimated by 5 radiologists. Their scores varied on average by approximately 3 points. Minimal and maximal joint involvement were scored consistently; exclusion of haemophilic arthropathy was possible with the highest degree of certainty. With regard to changes of a moderate degree agreement was seldom reached. With precisely defined characteristics, a high degree of consistency between observers was evident. Interobserver variation in the scoring was more marked in the ankle joint than in the other joints. Scoring was more consistent with the more experienced observers.