Thirteen studies (n=1,990, range 42 to 522) were included in the review. Ten studies enrolled consecutive patients, all studies used a valid reference standard, two studies reported that the index test and reference standard were measured independently and all studies reported that the decision to perform any intervention was independent of the index test. Eight studies reported that the decision to perform the reference standard was independent of the index test, in the remaining studies only those with a positive test underwent surgery.
Accuracy of clinical examination was generally poor. Out of a total of 89 sets of 2x2 data (studies reported data for multiple tests and thresholds), only one evaluation showed good diagnostic performance (positive likelihood ratio >10 and negative likelihood ratio <0.1). This test involved palpation for diagnosing rupture of the supraspinatus tendon; however, a second study reported poorer diagnostic performance.
Six of the evaluations reported positive likelihood ratio >10. Tests evaluated by these studies were positive palpation, combined Hawkins/painful arc/infraspinatus test, Napoleon test, lift-off test, belly-press test and drop-arm test. Two evaluations reported negative likelihood ratio <0.1 when they evaluated the empty-can test and Hawkins-Kennedy test; other studies of these tests showed poorer diagnostic performance.