• Thiago de Angelis Guerra Dotta, Jorge Henrique Assunção, Eduardo Baptista, Fernando Brandão Andrade E Silva, GracitelliMauro Emilio ConfortoMECInstituto de Ortopedia e Traumatologia, Hospital das Clinicas, Faculdade de Medicina, Universidade de Sao Paulo-IOTHCFMUSP, Rua Dr. Ovídio Pires de Campos 333 3 Floor, Cerqueira Cesar, São Paulo, SP, 05403-010, Brazil., NetoArnaldo Amado FerreiraAAFInstituto de Ortopedia e Traumatologia, Hospital das Clinicas, Faculdade de Medicina, Universidade de Sao Paulo-IOTHCFMUSP, Rua Dr. Ovídio Pires de Campos 333 3 Floor, Cerqueira Cesar, São Paulo, SP, 05403-010, Brazil., and Eduardo Angeli Malavolta.
• Instituto de Ortopedia e Traumatologia, Hospital das Clinicas, Faculdade de Medicina, Universidade de Sao Paulo-IOTHCFMUSP, Rua Dr. Ovídio Pires de Campos 333 3 Floor, Cerqueira Cesar, São Paulo, SP, 05403-010, Brazil. thiagogdotta@gmail.com.
• Arch Orthop Trauma Surg. 2024 Mar 1; 144 (3): 114911591149-1159.

IntroductionDespite being the most used exam today, few studies have evaluated the accuracy of findings on non-contrast magnetic resonance imaging (MRI). The primary objective of the study was to evaluate the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of non-contrast MRI findings in frozen shoulder, isolated and in combination. The secondary objectives were to define the interobserver and intraobserver agreement of the assessments and the odds ratio for frozen shoulder because of the various findings of MRI.MethodsA retrospective diagnostic accuracy study comparing non-contrast MRI findings between the frozen shoulder group and the control group. Sensitivity, specificity, positive and negative predictive value, accuracy, odds ratio, interobserver and intraobserver agreement were calculated for each finding and their possible associations.ResultsThe hyperintensity on capsule in the axillary recess presented 84% sensitivity, 94% specificity, and 89% accuracy. The obliteration of the subcoracoid fat triangle in the rotator interval had sensitivity 34%, specificity 82% and accuracy 58%. For coracohumeral ligament thickness ≥ 2 mm had specificity 66%, 48% specificity and 57% accuracy. Capsule thickness in the axillary recess ≥ 4 mm resulted in 54% sensitivity, 82% specificity, and 68% accuracy. Regarding interobserver agreement, only the posteroinferior and posterosuperior quadrants showed moderate results, and all the others showed strong reliability. The odds ratio for hyperintensity in the axillary recess was 82.3 for frozen shoulder. The association of these findings increased specificity (95%).ConclusionThe accuracy of non-contrast magnetic resonance imaging is high for diagnosing frozen shoulder, especially when evaluating the hyperintensity of the axillary recess. The exam has high reliability and reproducibility. The presence of an association of signs increases the specificity of the test.Level Of EvidenceLevel III, study of diagnostic test.© 2024. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

17 keywords...

hide…