Evaluation of a hock radiograph reading system for screening purposes in young horses

"\"To evaluate the validity of a pre‐established Radiographic Reading System (RRS) for detection of. radiographic signs of Developmental Orthopaedic Diseases (DOD) in the equine hock.. Out of 350 stallions screened for admission at the Belgian Studbook, 135 hocks (4 standard views) were. selected so that the sample contained the most common DOD lesions. Of the selected hocks, 22 were DOD. free and 113 had radiographic changes compatible with DOD. Radiographs were read in blind by 8 readers. (one of them defined as the expert as routinely involved in radiographic screening plans) using a dedicated. RRS in the form of a grid (Table 1). An official report by the official Studbook commission was available for. each hock. Data were analysed statistically using a Z‐test (P<0,05) in order to compare:. ‐ the official report versus the reading of the 4 views by the 8 readers;. ‐ the reading of the LM view alone and of the 4 views by the expert versus the reading of the LM view and. of the 4 views by the other 7 readers.. Agreement between readers was determined by the percentage of total agreement (Po) and by the multirater free‐marginal kappa (kfree).. Comparison between the official report and the reading of the 8 readers in the 4 views showed significant. differences only for medial malleolus osteochondrosis and for degenerative changes of the dorsal tarsometatarsal joint margins.. Comparison between the reading of the LM view between the expert and the other 7 readers was significantly different only for detection of subchondral bone cyst‐like lesions and for degenerative changes. in the dorsal margins of distal tarsal joints. The expert detected cyst‐like lesions more frequently, whereas. less experienced readers detected more distal tarsal joints changes. Difference in reporting degenerative. distal tarsal joints changes was also significant when comparing the readings of the 4 views.. Agreement between readers in classifying the hock as osteochondrosis positive or negative resulted optimal for the reading in the LM view alone (Po=0.93; Kfree=0.86), substantial for the reading in the 4 views (Po=0.82; Kfree=0.65).. Minor differences were seen between the official report and the reading by the 8 readers suggesting that a. RRS helps less experienced readers in identification of radiographic changes. Differences in detection of. cyst‐like lesions shows that identification of low contrast abnormalities in bone requires a higher degree of. experience if compared with identification of abnormalities altering profile and shape of the bone. Less. experienced readers tend to overestimate degenerative changes in distal tarsal joints. The optimal agreement between readers strongly supports the use of a RRS in screening protocol of sires (3).\""