Conservative therapy is usually sufficient for soft tissue injury to the shoulder, but if shoulder pain is caused by a rotator cuff tear, more intervention, and even surgery, maybe needed. Diagnosis of a rotator cuff tear is important and can be made by exploration of the shoulder (either open or arthroscopically) or double-contrast arthrography. These are painful and invasive procedures. Ultrasonography and magnetic resonance imaging are other methods of diagnosing rotator cuff tear, but these methods are either operator-dependent or expensive. Litaker and colleagues attempted to determine what predictors in the physical examination raise clinical suspicion of rotator cuff tear.

Included were all clinic patients between January 1990 and December 1994 who had arthrography to evaluate suspected rotator cuff tears. Exclusion criteria included suspicion of adhesive capsulitis, recent fracture of the shoulder, recent shoulder surgery and osteoarthritis of the glenohumeral joint. Specific elements of the history and physical examination were recorded, and one surgeon then predicted the presence of a rotator cuff tear (without knowledge of the arthrography result). A probability score was calculated for each patient, with low probability defined as zero to one point, medium as two to three points and high as four to five points.

A total of 448 patients were enrolled in the study. Two thirds (67.2 percent) of the patients had arthrographic evidence of at least a partial rotator cuff tear. The physical examination findings most likely to be associated with rotator cuff tears were evidence of infraspinatus or supraspinatus atrophy, weakness with elevation or external rotation of the arm, the “arc of pain” sign and the impingement sign. The combination of factors most likely to be associated with rotator cuff tear was weakness on external rotation, age of at least 65 years and the presence of night pain. Weakness with external rotation was tested by having the patient sit or stand with the arms next to the body. The elbows were then flexed to 90 degrees with thumbs up, and the shoulders were rotated 20 degrees internally. The examiner then applied inward pressure while instructing the patient to resist such pressure.

Night pain was defined as shoulder pain that prevented the patient from staying asleep. These factors were then used to determine the probability score. Weakness and age of at least 65 years were assigned two points and presence of night pain was assigned one point. The likelihood ratio of having a rotator cuff tear in those with a low probability score was 0.23; the likelihood ratio for a moderate probability score was 1.36 and for a high probability score, 9.84. In 91 percent of patients with a high probability score, the arthrogram was abnormal.

The best sensitivity, at more than 97 percent, was found for the following signs: night pain, impingement sign (positive if there is an increase in pain when the fully elevated arm has the hand supinated and the arm adducted against the ear and then internally rotated) and painful arc (positive when pain is minimal at full elevation of the arm, and then increases as the arm is actively lowered in the abduction plane), yet these signs had a specificity of less than 10 percent.

The authors conclude that confirmatory diagnostic testing should be limited to patients who have a moderate or high probability of rotator cuff tear. Conservative therapy can be initiated in patients with low probability scores. By using this scoring system, diagnostic tests and consultant referral can be used more appropriately.