Background
Rotator cuff tear represents one of the most common cause of shoulder pain and motor disability, accounting for about 70% of all shoulder dysfunctions. Different imaging modalities are commonly used for its preoperative evaluation. Magnetic resonance imaging (MRI) allows the surgeon to identify several tear characteristics, such as size, muscle atrophy and fatty infiltration that have been suggested to be key indicators of outcome after surgery. Several classifications have been described to fully investigate and standardized main tear characteristics. However, there is still confusion when it comes to reliability of those classifications. The purpose of this systematic review was threefold: to identify all the MRI classifications available in the literature for the preoperative assessment of rotator cuff tears, to summarize available data on the reliability of identified classifications and to assess the methodological quality of reliability studies.
Methods
This systematic review was conducted following the PRISMA guidelines (Preferred Reporting Items for Systematic Reviews and Meta-Analysis). Search was performed using major databases (Pubmed and Ovid Medline). All studies reporting MRI assessment in patients with a postero-superior rotator cuff tear were included. After the identification of the available MRI criteria, reliability studies were further analyzed. Descriptive statistics were used to summarize findings. When the data were enough, a meta-analysis was carried out. Pooled estimates for each meta-analysis were obtained using a random-effects model. The heterogeneity was investigated by calculating the I2 index. Methodological quality was assessed using the Quality Appraisal of Reliability studies (QAREL) checklist.
Results
Ninety-one studies were included in this review. Fifty-five different classifications were identified among seven categories of outcomes. Fifty-three studies reported inter-observer reliability, whereas twenty-six studies reported intra-observer reliability of some of the identified classifications. Structural integrity was the most investigated outcome, with twenty-two different classifications described. According to the meta-analysis, inter-observer reliability showed a correlation coefficient (COR) of 0.79 (CI 95%: 0.75-0-82). Tear size evaluation also showed a good inter-observer reliability after data pooling: COR 0.79 (CI 95%: 0.60-0.90). A wide range of reliability from poor to excellent agreement was found for the other outcomes. Heterogeneity was always substantial. Methodological quality was high for twenty studies, moderate for fourteen studies, and low for twenty-one studies.
Conclusion
Preoperative magnetic resonance is a reliable imaging modality to identify the presence of a full-thickness tear and its width. For all other classification commonly used to assess preoperatively tear characteristics there is still not enough evidence in the literature to consider them accurate and reproducible.