Reproducibility and sensitivity to change of a new method of computer measurement of joint space width in hip osteoarthritis. Performance of three radiographic views obtained at a 3-year interval

Osteoarthritis Cartilage. 2009 Jul;17(7):864-70. doi: 10.1016/j.joca.2008.12.003. Epub 2008 Dec 24.


Background: Measurement of radiographic joint space width (JSW) and of joint space narrowing (JSN) is the currently recommended method for assessment of anatomical severity and structural progression of osteoarthritis (OA), respectively. A standard radiographic view of the pelvis is commonly used for measurement of hip OA but other views are available.

Objectives: To evaluate the inter-intra reader reproducibility and the sensitivity to change of a new automated method of measurement of the hip JSW and to assess which radiographic view [pelvis anteroposterior (AP) view, hip AP view, hip oblique view] provides the greatest accuracy for JSW and JSN measurements.

Material and methods: An AP pelvis radiograph, an AP radiograph centered on the target hip (AP hip) and an oblique view were performed at baseline (M0) and 3 years later (M36) in 50 hip OA patients. Two readers, blinded to each other's results and time sequence, measured twice, at a minimum 15 day interval, the six radiographs of each patient, using a novel version of a previously validated software program whose edge-based algorithm automatically detects the joint space contours. Inter-observer cross-sectional (M0+M36) and longitudinal (M0-M36) reproducibility of JSW measurement was assessed by the intra-class correlation coefficient (ICC) and the Bland-Altman method. Sensitivity to change was estimated by the standardized response mean (SRM). An ANOVA was used to analyze differences related to the observer and the view.

Results: Intra-observer reproducibility: For JSW measurement, the ICC value, for observers 1 and 2 respectively, were 0.92 and 0.83 for the pelvic view, 0.96 and 0.88 for the hip AP view, and 0.90 and 0.86 for the oblique view. For JSN, ICC was 0.94 and 0.82 for the pelvic view, 0.97 and 0.78 for the hip AP view, and 0.95 and 0.86 for the oblique view. Inter-observer reproducibility: For JSW measurement, ICC was 0.87 for the pelvic view, 0.98 for the hip AP view, and 0.87 for the oblique view. The mean inter-observer difference (SD) was 0.0 (0.31), -0.01 (0.15) and -0.04 (0.4)mm for pelvic, AP and oblique views respectively. For JSN, ICC was 0.91 for the pelvic view, 0.93 for the hip AP view, and 0.90 for the oblique view. Sensitivity to change: SRM values were 0.61 (observer 1) and 0.65 (observer 2) for the pelvic view, 0.68 and 0.75, respectively, for the hip AP view, 0.64 and 0.66, respectively, for the oblique view. JSN did not vary significantly with the observer and the view. In 27% of cases intervention by the observer was necessary to correct the computer's identification of the acetabular edge in the area of interest.

Conclusion: Computer measurement of the radiographic hip joint space provided good intra- and inter-observer reproducibility and good sensitivity to change. However, it was necessary for the observer to intervene frequently to select the area of interest and adjust detection of the bone edge. The hip AP view performed better than the pelvis and oblique views, but not significantly so.

Publication types

  • Multicenter Study
  • Randomized Controlled Trial

MeSH terms

  • Analysis of Variance
  • Double-Blind Method
  • Female
  • Hip Joint / diagnostic imaging*
  • Hip Joint / pathology
  • Humans
  • Male
  • Middle Aged
  • Observer Variation
  • Osteoarthritis, Hip / diagnostic imaging*
  • Osteoarthritis, Hip / pathology
  • Radiographic Image Interpretation, Computer-Assisted / standards*
  • Reproducibility of Results
  • Sensitivity and Specificity
  • Software / standards