Objectives: To determine the value of elements of the bedside history and physical examination in predicting arthrography results in older patients with suspected rotator cuff tear (RCT).
Design: Retrospective chart review
Setting: Orthopedic practice limited to disorders of the shoulder
Participants: 448 consecutive patients with suspected RCT referred for arthrography over a 4-year period
Main outcome measure: Presence of partial or complete RCT on arthrogram
Results: 301 patients (67.2%) had evidence of complete or partial RCT. Clinical findings in the univariate analysis most closely associated with rotator cuff tear included infra- and supraspinatus atrophy (P < .001), weakness with either elevation (P < .001) or external rotation (P < .001), arc of pain (P = .004), and impingement sign (P = .01). Stepwise logistic regression based on a derivation dataset (n = 191) showed that weakness with external rotation (Adjusted Odds Ratio (AOR) 6.96 (3.09, 13.03)), age > or = 65 (AOR 4.05(2.47, 16.07)), and night pain (AOR 2.61 (1.004, 7.39)) best predicted the presence of RCT. A five-point scoring system developed from this model was applied in the remaining patient sample (n = 216) to test validity. No significant differences in performance were noted using ROC curve comparison. Using likelihood ratios, a clinical score = 4 was superior in predicting RCT to the diagnostic prediction of an expert clinician. This score had specificity equivalent to magnetic resonance imaging or ultrasonography in diagnosis of RCT.
Conclusions: The presence of three simple features in the history and physical examination of the shoulder can identify RCT efficiently. This approach offers a valuable strategy to diagnosis at the bedside without compromising sensitivity or specificity.