Background and purpose: Perthes' disease leads to radiographic changes in both the femoral head and the acetabulum. We investigated the inter-observer agreement and reliability of 4 radiographic measurements assessing the acetabular changes.
Patients and methods: We included 123 children with unilateral involvement, femoral head necrosis of more than 50%, and age at diagnosis of 6 years or older. Radiographs were taken at onset, and 1 year and 5 years after diagnosis. Sharp's angle, acetabular depth-width ratio (ADR), lateral acetabular inclination (LAI), and acetabular retroversion (ischial spine sign, ISS) were measured by 3 observers. Before measuring, 2 of the observers had a consensus meeting.
Results: We found good agreement and moderate to excellent reliability for Sharp's angle for all observers (intra-class correlation coefficient (ICC) > 0.80 with consensus, ICC = 0.46-0.57 without consensus). There was good agreement and substantial reliability for ADR between the observers who had had a consensus meeting (ICC = 0.62-0.89). Low levels of agreement and poor reliability were found for observers who had not had a consensus meeting. LAI showed fair agreement throughout the course of the disease (kappa = 0.28-0.52). The agreement between observations for ISS ranged from fair to good (kappa = 0.20-0.76).
Interpretation: Sharp's angle showed the highest reliability and agreement throughout the course of the disease. ADR was only reliable and showed good agreement between the observers when landmarks were clarified before measuring the radiographs. Thus, we recommend both parameters in clinical practice, provided a consensus is established for ADR. The observations for LAI had only fair agreement and ISS showed inconclusive agreement in our study. Thus, LAI and ISS can hardly be recommended in clinical practice.