Missed Diagnosis In Rotator Cuff Injury On MRI And Ultrasonography: A Systematic Review And Meta-Analysis Of Studies Published In The Last 10 Years

Abstract

Missed diagnoses of rotator cuff tears on non-invasive imaging remain a clinically important problem. Recent data have shown that a substantial proportion of tears — particularly subscapularis tears and partial-thickness lesions — are overlooked on MRI and ultrasonography (US). Whether one modality systematically misses more tears than the other in contemporary practice is unclear. The objective of this study was to evaluate MRI and US in terms of missed diagnoses (false-negative rate) and overall diagnostic accuracy for rotator cuff injury, through a systematic review and meta-analysis of studies published in the last 10 years (January 2015 – December 2025). A comprehensive literature search was performed in PubMed/MEDLINE, Scopus, Web of Science, and the Cochrane Library on the main concepts of rotator cuff tear, MRI (including conventional MRI and MR arthrography), US, and diagnostic accuracy. Inclusion criteria consisted of original research studies that assessed the diagnostic accuracy of MRI and/or US (index tests) for the diagnosis of rotator cuff tear using surgical findings (arthroscopy or open surgery) as the reference standard. The QUADAS-2 tool was used to assess methodologic quality. Meta-analyses were performed to compare MRI and US in the diagnosis of any rotator cuff tear, full-thickness tears, partial-thickness tears, and subscapularis tears. Between-study variation was analyzed using the Cochran Q test and I² statistic. Twelve studies satisfied inclusion and exclusion criteria (MRI, n = 7; US, n = 6; both modalities in the same patients, n = 3), consisting of 2,253 total patients and 2,417 total shoulders. For any rotator cuff tear, MRI showed a pooled sensitivity of 0.91 (95% CI, 0.87–0.94) and specificity of 0.89 (95% CI, 0.85–0.92); US showed a pooled sensitivity of 0.87 (95% CI, 0.82–0.91) and specificity of 0.91 (95% CI, 0.86–0.94). For full-thickness tears, sensitivity was 0.93 (95% CI, 0.90–0.95) for MRI and 0.91 (95% CI, 0.88–0.94) for US (p = 0.42). Diagnostic performance dropped for partial-thickness tears (sensitivity: MRI 0.70 [95% CI, 0.62–0.78]; US 0.65 [95% CI, 0.54–0.75]) and for subscapularis tears (sensitivity: MRI 0.56 [95% CI, 0.44–0.68]; US 0.30 [95% CI, 0.18–0.46]). Heterogeneity across studies was moderate to high (I² = 48%–75%). Corresponding miss rates were: full-thickness tear 7% (MRI) vs 9% (US); partial-thickness tear 30% vs 35%; subscapularis tear 44% vs 70%. MRI and US show comparable accuracy for full-thickness rotator cuff tears in studies published in the last 10 years, and either modality can be considered a first-line imaging option in that clinical setting. Both modalities, however, miss a clinically meaningful proportion of partial-thickness and subscapularis tears, with higher miss rates on US. When the subscapularis or a small partial tear is clinically suspected, MRI (or MR arthrography) may reduce missed diagnoses. This review quantifies the contemporary risk of missed diagnoses on MRI and US, supports shared decision-making between clinicians and patients when selecting an initial imaging modality, and identifies the subscapularis tendon and partial-thickness articular-surface tears as priorities for reader training and protocol optimization.