A set of three clinical tests can detect the presence or absence of rotator cuff tears.

Question: What is the diagnostic accuracy of the physical examination in predicting arthroscopy findings in patients with suspected rotator cuff tear? Design: Cross-sectional study comparing the results of the clinical examination with presence of partial or complete rotator cuff tear on arthroscopy. Setting: Australian orthopaedic surgeon’s practice. Patients: Patients (400) with shoulder injuries that warranted arthroscopic examination. Description of tests and diagnostic standard: Each subject was examined with 23 commonly-used shoulder tests and then underwent arthroscopy. Subjects were examined for wasting, tenderness, active and passive range of motion, shoulder strength and the presence of three signs: the drop arm sign, O’Brien’s sign and impingement sign. Main outcome measures: Prevalence of rotator cuff tear per 10 year age group and positive and negative likelihood ratios (‘+LR’ and ‘–LR’ respectively)(a). Main results: Prevalence of rotator cuff tear increased with age from less than 5% for subjects aged in their twenties to more than 80%(b) for subjects in their seventies. Most clinical shoulder tests, when used in isolation, could not distinguish between patients with and without rotator cuff tears (individual results were not provided). The three exceptions were the presence of supraspinatus weakness, weakness in external rotation and the impingement sign. If all three features were present the +LR and -LR were: 48.0 and 0.76; if at least two were present: 7.6 and 0.42; at least one present: 1.9 and 0.01. Conclusion: Tears of the rotator cuff are common in patients with shoulder injuries referred for arthroscopy and prevalence increases markedly with age. Individual physical examination tests are inaccurate in diagnosing rotator cuff tears. The presence of supraspinatus weakness, weakness in external rotation and the impingement sign are associated with greatly increased likelihood of rotator cuff tear and the absence of all three is associated with greatly reduced likelihood of rotator cuff tear.