Skip to content

Assessing the Diagnostic Accuracy of Special Tests for Rotator Cuff Tears

Key findings

  • For detection of supraspinatus tears, the Jobe test and the "full can test" performed best
  • For detection of infraspinatus tears, the Hornblower sign performed better than lag signs
  • For detection of subscapularis tears, the belly-press and bear hug tests had the highest sensitivities

Rotator cuff tears are difficult to diagnose based on clinical examination alone. More than a dozen “special tests” have been described, but few data are available on their accuracy, and clinicians often resort to expensive MRI.

To provide guidance, Jon J.P. Warner, MD, chief of the Shoulder Service at Massachusetts General Hospital, and colleagues compared the diagnostic accuracy of 15 special tests. In the American Journal of Physical Medicine and Rehabilitation, they report that the Jobe test and "full can test" performed well for diagnosis of supraspinatus tears, and the Hornblower sign for infraspinatus tears.

The researchers, who are affiliated with sports medicine and shoulder clinics, are conducting a prospective longitudinal cohort study called the Rotator Cuff Outcomes Workgroup. Between February 2011 and December 2012, they recruited 208 patients older than 45 years who had experienced shoulder pain and/or loss of range of motion for at least four weeks.

Of these, 69 patients had a rotator cuff tear diagnosed by clinical impression and confirmed by MRI, and had special tests performed:

  • Five tests were used for detecting subscapularis tears (lift-off test, passive lift-off test, belly-press test, belly-off sign and bear hug)
  • Three tests for infraspinatus tears (external rotation lag sign at 0 degrees, external rotation lag sign at 90 degrees and Hornblower sign)
  • Five tests for supraspinatus tears (Jobe test, full can test, drop-arm test, Neer sign and Hawkins sign)
  • Two tests for proximal biceps tendon pathology (speed test and bicipital groove tenderness)

Among tests for supraspinatus tears, the Jobe test had sensitivity of 88% and specificity of 62%, and a likelihood ratio of 2.30, which represents the likelihood that the test would be positive in a patient with a rotator cuff tear as opposed to one without a tear. The full can test had lower sensitivity of 70%, but higher specificity of 81% and a higher likelihood ratio of 3.75. The drop-arm test had a very high likelihood ratio of 6.45, but the researchers consider it unreliable because it resulted from very high specificity and very low sensitivity.

For detection of infraspinatus tears, the external rotation lag signs had high specificity but very low sensitivity. The researchers recommend using the Hornblower sign. However, because it also had high specificity (96%) and low sensitivity (17%), its utility is limited when the test is negative. A positive test indicates a high likelihood of rotator cuff tear.

The tests for subscapularis tears also generally had high specificity but low sensitivity. The belly-press and bear hug tests had the highest sensitivities. The tests for biceps pathology had relatively low sensitivities and specificities, and their likelihood ratios were similar.

The researchers conclude that their data should assist clinicians in deciding which special tests to rely on and help reduce the use of MRI.

Learn more about the Shoulder Service

Refer a patient to the Shoulder Service

Related topics


Three-dimensional planning software may improve outcomes for shoulder arthroplasty. A new technology allows surgeons to install the prosthetic joint with greater foresight and precision.


In some cases of posterior labral tears of the shoulder with concomitant posterior shoulder instability, MRI shows increased posterior humeral subluxation. Mass General surgeons thought MRI might help with the notoriously challenging diagnosis of posterior shoulder instability.