Accuracy and reliability of MRI-reports to determine which shoulder is symptomatic for workers compensation patients with unilateral symptoms

Introduction: Shoulder abnormalities on imaging are increasingly common with advancing age. We tested the difference in the accuracy of diagnosing the injured shoulder between surgeons who reviewed MRI reports and who did and did not receive information about the patients. Methods: Thirty people aged 40 or older that experienced new shoulder symptoms after an event at work had MRI of both shoulders. Members of the Science of Variation Group (SOVG) were invited to review the radiologist's reports and diagnose the symptomatic side. Ninety-seven surgeons participated. Results: Surgeon observers provided with patient information were not more accurate in diagnosing the symptomatic shoulder (Odds Ratio: 0.90; 95% CI: 0.75 to 1.1; P = 0.29). There was slight agreement between surgeons (kappa = 0.10). The sensitivity and specificity for diagnosing the symptomatic side were 51% (CI: 48%–54%) and 67% (CI: 64%–70%) respectively. A binomial test showed that surgeons indicated the injured shoulder slightly more frequently than expected by random chance (59%; P < 0.001). Conclusions: Surgeons were only slightly better at indicating the symptomatic side than random chance. Shoulder pain in people aged 40 or older should probably be considered age-related unless there is good objective evidence of acute traumatic pathology. Level of evidence: Level II, diagnostic study.