Hip Osteoarthritis MRI Scoring System (HOAMS): reliability and associations with radiographic and clinical findings

To develop a semiquantitative MRI-based scoring system (HOAMS) of hip osteoarthritis (OA) and test its reliability and validity.

Fifty-two patients with chronic hip pain were included. 1.5T magnetic resonance imaging (MRI) was performed on all patients. Pelvic radiographs were scored according to the Kellgren-Lawrence (KL) system. Clinical outcomes were assessed by the hip osteoarthritis outcome score (HOOS). MRIs were analyzed using a novel whole-joint MRI score that incorporated 13 articular features. Reliability was determined on a random subset of 15 cases. Weighted-kappa statistics and overall agreement were used as a measure of intra- and inter-observer reliability. Associations between MRI features and radiographic OA severity were calculated using Cochran-Armitage test for trend. Ordinal logistic regression was used to assess associations between MRI features and severity of pain and functional limitation.

Distribution of radiographic grading was: KL 0=12 (27%), KL 1=11 (25%), KL 2=14 (32%), KL 3=5 (11%) and KL 4=2 (5%). Intra-reader reliability for the different features ranged from 0.18 (cysts) to 0.85 (cartilage). Inter-reader reliability ranged between 0.15 (cysts) and 0.85 (BMLs). Low kappas were due to low frequencies of some features as overall percent agreement was good to excellent (83.8% and 83.1%). There was a strong association between MRI-detected lesions and radiographic severity (P=0.002). Non-significant trends were observed between MRI features and clinical outcomes.

MRI-based semiquantitative assessment of the hip shows adequate reliability. Presence of more severe MRI-detected intraarticular pathology shows a strong association with radiographic OA. The results suggest possible associations between MRI-detected pathology and clinical symptoms.