Supraspinatus tears: predictability of magnetic resonance imaging findings based on clinical examination.

Background: The shoulder physical examination is frequently performed to diagnose rotator cuff tears in clinical practice; however, there is insufficient evidence on which physical tests are efficient for shoulder impingements in primary care settings. The purposes of this study were (1) to analyze the accuracy of 8 clinical tests to diagnose supraspinatus tendon lesions and investigate whether the combination of these tests can improve the diagnostic values; (2) to assess which are the best positivity criteria for shoulder maneuvers-pain and/or weakness; and (3) to investigate the ability of these tests to distinguish between partial- and full-thickness tears.

Methods: A total of 733 consecutive patients were prospectively evaluated by 4 shoulder surgeons in this multicenter diagnostic study from May 2017 to December 2018, and 8 clinical tests (empty-can, full-can test, drop arm, painful arc, Neer sign, Hawkins test, Patte test, and resisted external rotation) were compared with magnetic resonance imaging performed by blinded radiologists. We assessed the sensitivity, specificity, accuracy, positive and negative predictive values, and diagnostic odds ratio (DOR) for all tests.

Results: For overall supraspinatus tears, the empty-can test showed the highest sensitivity (0.81), the Patte test (positive for pain and weakness) and the drop-arm test showed the highest specificity (0.99 and 0.98, respectively), and the best combination was the Neer sign and the drop-arm test (DOR, 12.92). The positivity criteria for pain associated with weakness showed the highest performance, with DORs of 16.94 for the Patte test and 10.45 for the empty-can test. The Patte test and resisted external rotation, positive for pain and weakness, showed the highest ability to distinguish between full- and partial-thickness tears (DOR, 5.69 and 5.35, respectively). The shoulder maneuvers showed low negative predictive values; the highest value (0.58) was found with the empty-can test.

Conclusions: (1) The clinical tests demonstrated excellent diagnostic values; the empty-can test had the highest sensitivity, whereas the drop-arm test and the Patte test had the highest specificity; and the best combination for detecting supraspinatus tears was the Neer sign and the drop-arm test. On the other hand, the physical examination findings showed limited values to rule out tears. (2) The best positivity criterion for shoulder maneuvers was pain associated with weakness. (3) The Patte test and resisted external rotation showed the highest ability to distinguish between full- and partial-thickness tears.