Introduction
To determine the inter-observer reliability of ultrasound measurements and grading of rotator cuff and shoulder joint disease in patients with inflammatory arthritis.
Methods
Both shoulders of 50 patients (56 ± 18 years, 39F:11M) with inflammatory arthritis, being treated with biologics, were scanned by two experienced radiologists. Rotator cuff and articular surface features in the shoulder joint were measured and graded using previously published ultrasound scoring methods.
The inter-observer agreements for rotator cuff tendon, glenohumeral joint (GHJ) and subacromial subdeltoid bursal (SSB) measurements were obtained with intraclass correlation coefficients (ICC) for continuous data. Those for the grade of rotator cuff tendon tear, humeral head (HH) and acromio-clavicular joint (ACJ) appearances were analysed with kappa (K) statistics for categorical data.
Results
Moderate inter-observer agreement (ICC = 0.58, P ≤ 0.01) was shown for the supraspinatus tendon thickness. Poor agreement (ICC < 0.49) was found for the thickness of the subscapularis (P ≤ 0.01), infraspinatus (P ≤ 0.05) and biceps tendons. There was excellent agreement for the SSB thickness (ICC = 0.88, P ≤ 0.01) and poor agreement for GHJ distance.
Moderate agreement (K = 0.57, P ≤ 0.01) for grading of a supraspinatus tendon tear was found while excellent agreement for the subscapularis (K = 0.81, P ≤ 0.01), infraspinatus and biceps (both K = 1, P ≤ 0.01) tendons was noted. The most common agreement occurred for no tear or a complete tear. Poor agreement (K < 0.49) was demonstrated for grading of HH erosions and ACJ appearances.
Conclusion
Ultrasound measurement of the subacromial subdeltoid bursa was reliable between observers, as was detection of the presence or absence of a complete rotator cuff tendon tear. Reliability was limited for measurements of rotator cuff tendon thickness, grading partial tendon tears or identification of subtle shoulder joint pathology.