Inter-rater agreement in radiographic canine hip dysplasia evaluation

The objective was to determine the agreement between scrutineers assessing canine hip dysplasia (CHD) within and between the Fédération Cynologique Internationale (FCI) and the Swiss (CH) grading schemes, and between first-opinion and second-opinion scrutineers using the CH scheme. 62 scrutineers, participating in a quality assessment programme, were asked to evaluate 12 sets of hip radiographs. All radiographs had been previously evaluated by a first-opinion and 7 by both a first-opinion and second-opinion scrutineers using the CH scheme. Inter-rater agreement (κ) and intraclass correlation (ICC) were evaluated for the Norberg angle and dysplasia grades, and ICC for the numerical CH scores. The κ and ICC were 0.290 and 0.471 for Norberg angles, 0.372 and 0.577 for FCI grades, and 0.417 and 0.567 for CH grades, respectively. The ICC for CH scores was 0.614. Total agreement for grades was found in a single case. For radiographs evaluated using both schemes, raters gave the same grade in 441 cases, a lower FCI than CH grade in 78 cases and a lower CH than FCI grade in 32 cases. Intrarater agreement between grades of the two schemes was 0.786. Agreement between the first-opinion and second-opinion scrutineers was 0.250 for Norberg angles and -0.286 for CH grades. Although agreement between the two schemes was good, agreement between scrutineers using either scheme was only fair to moderate, and no agreement was found between first-opinion and second-opinion scrutineers. Application of the FCI and CH schemes may impact the incidence of hip dysplasia in populations with both severe and moderate dysplasia, but is unlikely to impact populations with only mild dysplasia. The CH system of first-opinion and second-opinion examinations should be re-evaluated.