Importance: Rotator cuff disease (RCD) is the most common cause of shoulder pain seen by physicians.
Objective: To perform a meta-analysis to identify the most accurate clinical examination findings for RCD.
Data sources: Structured search in MEDLINE, EMBASE, and CINAHL from their inception through May 2013.
Study selection: For inclusion, a study must have met the following criteria: (1) description of history taking, physical examination, or clinical tests concerning RCD; (2) detailing of sensitivity and specificity; (3) use of a reference standard with diagnostic criteria prespecified; (4) presentation of original data, or original data could be obtained from the authors; and (5) publication in a language mastered by one of the authors (Danish, Dutch, English, French, German, Norwegian, Spanish, Swedish).
Main outcomes and measures: Likelihood ratios (LRs) of symptoms and signs of RCD or of a tear, compared with an acceptable reference standard; quality scores assigned using the Rational Clinical Examination score and bias evaluated with the Quality Assessment of Diagnostic Accuracy Studies tool.
Results: Twenty-eight studies assessed the examination of referred patients by specialists. Only 5 studies reached Rational Clinical Examination quality scores of level 1-2. The studies with quality scores of level 1-2 included 30 to 203 shoulders with the prevalence of RCD ranging from 33% to 81%. Among pain provocation tests, a positive painful arc test result was the only finding with a positive LR greater than 2.0 for RCD (3.7 [95% CI, 1.9-7.0]), and a normal painful arc test result had the lowest negative LR (0.36 [95% CI, 0.23-0.54]). Among strength tests, a positive external rotation lag test (LR, 7.2 [95% CI, 1.7-31]) and internal rotation lag test (LR, 5.6 [95% CI, 2.6-12]) were the most accurate findings for full-thickness tears. A positive drop arm test result (LR, 3.3 [95% CI, 1.0-11]) might help identify patients with RCD. A normal internal rotation lag test result was most accurate for identifying patients without a full-thickness tear (LR, 0.04 [95% CI, 0.0-0.58]).
Conclusions and relevance: Because specialists performed all the clinical maneuvers for RCD in each of the included studies with no finding evaluated in more than 3 studies, the generalizability of the results to a nonreferred population is unknown. A positive painful arc test result and a positive external rotation resistance test result were the most accurate findings for detecting RCD, whereas the presence of a positive lag test (external or internal rotation) result was most accurate for diagnosis of a full-thickness rotator cuff tear.