The clinical utility of imaging in osteoarthritis and its importance in future prediction of total hip replacement; a nested case-control study within the AGES-Reykjavik cohort

Current guidelines are split on the role that imaging has in the clinical assessment of osteoarthritis, yet clinical computed tomography (CT) imaging has now revealed how a 3D approach can improve prediction of total hip replacement (THR) over 2D measures alone. We applied 2D grading and measurement along with 3D cortical bone mapping to ordinary clinical CT imaging of the pelvis in a cohort of healthy older people, aiming to discover which of these features had clinical utility in predicting total hip replacement (THR) within 8 years and which were related to baseline hip pain. Using a nested case-control design in the AGES-Reykjavik study, 74 future THR cases were age and sex-matched with 184 controls from the cohort (age 74±5yrs). Baseline assessment involved a validated hip pain questionnaire and pelvic CT. The following were performance-tested using ROC analysis and Clinical Utility Index: (i) hip pain; (ii) Kellgren and Lawrence grade (K&L grade), (iii) minimum joint space width (mJSW); and (iv) 3D cortical bone thickness (CTh). The clinical utility index for prediction of future THR from baseline pain was poor at 0.28, with the inclusion of imaging improving this to 0.79 (K&L grade) and 0.82 (3D CTh). Self-reported hip pain at baseline was also a poor-to-marginal predictor of THR (AUC=0.63), but 3D cortical thickening at the femoral head was predictive of future THR (0.81). Having radiographic osteoarthritis strongly predicted THR irrespective of hip pain (0.85). Combining hip pain, K&L grade and 3D cortical thickness gave optimal prediction (0.88). Ascertainment bias may have occurred if primary care physicians requested their own radiographs of their patients' hips. Imaging features from standard clinical CT identifies patients at high risk of progression to surgery for osteoarthritis, regardless of baseline pain.