Reliability assessment of a mechanism-based approach to post-injury knee magnetic resonance imaging interpretation by general radiologists

• Main Authors: James Stutterheim, Matthew D. Goodier
• Format: Article
• Language: English
• Published: AOSIS 2018-06-01
• Series: South African Journal of Radiology
• Subjects: magnetic resonance imaging; mechanism-based approach; injury mechanism; pattern of injury
• Online Access: https://sajr.org.za/index.php/sajr/article/view/1253
• ISSN: 1027-202X; 2078-6778

Background: A mechanism-based approach to post-injury knee magnetic resonance imaging (MRI) interpretation, following acute complex knee injury, is cited by several authors to provide increased reporting accuracy and efficiency, by allowing accurate prediction of injury to at-risk structures. This remains to our knowledge untested in a developing world setting and is of interest to us as South African general radiologists.   Objective: To assess the reliability of a mechanism-based approach to complex post-trauma knee MRI interpretation when implemented by general radiologists in a South African setting, and compare our results with the findings of North American authors who compiled and assessed the same classification. To measure the agreement between the observers.   Methods: A quantitative, observational, investigative, retrospective study was performed using a sample of 50 post-trauma knee MRI studies conducted at Grey’s Hospital, Pietermaritzburg. Two investigators independently applied the consolidated mechanism-based approach compiled by Hayes et al. as a research tool to interpret the knee MRI studies, blinded to each other’s findings.   Results: Injury mechanism was assigned in 32% of cases by the principle investigator and in 20% of cases by the supervisor, with fair agreement between the observers (k = 0.39). The investigators agreed that 62% of cases were not classifiable by mechanism, 26% because of highly complex injury and 26% because of non-specific findings.   Conclusion: Our findings indicate that the Hayes et al. classification is a non-ideal tool when used by general radiologists in our setting, as the pure injury mechanisms described in the classification were rare in our study group. Patient epidemiology and investigator experience are highlighted as potential limiting factors in this study. Despite this, we advocate that the concept of a mechanism-based approach for the interpretation of acute post-trauma knee MRI holds value for general radiologists, particularly in patients imaged before resolution of bone bruising (within 12–16 weeks of injury), and those injured in sporting and similar athletic activities.