Inadequacy of plain radiographs in the diagnosis of early osteoarthritis of the hip.

Aim: The study was undertaken to determine the value of standard radiographs in the early stages of osteoarthritis. Methods: Standard radiographs and arthro-MRI's from thirty hips operated on for early arthrosis (age 25-57 years) were independently analyzed by two orthopaedic surgeons and one radiologist blinded from the intra-operative findings. The radiographs were read on two occasions two months apart. The radiographic findings were then compared to the intra-operative findings. Results: Intra-operatively, all cases had a labral lesion and, in all but three of the cases, there was a major acetabular cartilage lesion. Each investigator diagnosed all of the labral and/or cartilage lesions on the arthro-MRI. However, on average, the investigators judged 20% (10-35%) of the hips to be normal on the standard radiographs. The probability of detecting an abnormal hip joint was statistically significantly better with arthro-MRI in four of six readings (p < 0.05) and there was a trend in favor of the arthro-MRI in the other two readings (p < 0.1). Intra-observer agreement when using the Tonnis classification of arthrosis on standard radiographs was 0.26 (-0.1-0.62), 0.69 (0.42-0.96) and 0.83 (0.53-1) [kappa-statistic, (95% confidence interval)]. The interobserver agreement was 0.24 (- 0.07 -0.55). Conclusion: Plain radiographs in the early stages of osteoarthrosis of the hip are neither reliable nor valid to diagnose the onset of disease. Therefore, in the case of a normal radiograph and clinical suspicion of arthrosis, a "normal" radiograph does not exclude the diagnosis and an artho-MRI should be obtained for further evaluation.