Diagnostic Accuracy and Cost-Effectiveness of Magnetic Resonance Imaging Versus Ultrasonography in Rotator Cuff Tears: A Critical Narrative Review

Abstract

Rotator cuff tears are among the most frequent structural causes of shoulder pain and disability in adults, and imaging is used both to confirm the presence of a tear and to characterise it sufficiently to guide treatment. Magnetic resonance imaging and ultrasonography are the two modalities most often deployed for this purpose, yet the choice between them continues to be governed as much by local capacity, operator expertise and reimbursement arrangements as by measured diagnostic performance. This critical narrative review evaluates the strength, consistency and methodological quality of the evidence comparing the two modalities, and examines how that evidence interacts with the sparse health-economic literature. Literature was identified through structured searching of biomedical and multidisciplinary bibliographic sources, supplemented by citation tracking and by examination of professional guideline documents, with a final search date of 18 June 2026. Evidence was appraised against established criteria for diagnostic accuracy research and synthesised thematically rather than pooled. Repeated quantitative syntheses converge on the conclusion that the two modalities perform comparably for full-thickness tears of the posterosuperior cuff, with pooled sensitivity and specificity typically above 0.90, and that both perform substantially less well for partial-thickness tears, where sensitivity commonly falls between 0.51 and 0.84. This apparent equivalence rests largely on indirect comparison across heterogeneous, methodologically limited studies rather than on adequately powered head-to-head evaluation. Tendon-specific analyses reveal a divergent and unresolved pattern for the subscapularis, where several direct comparisons favour ultrasonography while pooled magnetic resonance data indicate limited sensitivity. Economic evaluations are few, are dominated by decision-analytic models constructed from United States price data, and reach conclusions that are highly sensitive to modelling assumptions and to the availability of intraoperative assessment. The most defensible conclusion is that modality choice for tear detection is a system-level and workforce question rather than a purely diagnostic one, while magnetic resonance imaging retains an advantage for reproducible characterisation of large tears and for detection of concomitant intra-articular pathology.