CT grading scales for splenic injury: Why can't we agree?

Introduction: Computed tomography (CT) has been proven to be able to accurately diagnose splenic injury. Many have published CT splenic injury grading scales to quantify the extent of injury. However, these scales have failed at predicting clinical outcomes and therefore cannot be used to accurately predict the need for intervention. We hypothesised that low interrater reliability is the reason why these scales have failed at predicting clinical outcomes. Methods and Materials: This is a retrospective study of patients who were admitted to the Royal Perth Hospital with blunt splenic injury as coded in the trauma registry. The abdominal CT images of these patients were reviewed by three consultant radiologists and were graded using the six different splenic injury grading scales. We assessed interrater reliability between each of the scales using generalised kappa and proportion of agreement calculations. Results: The images of 64 patients were reviewed. The interrater reliability yielded a generalised kappa score of 0.32-0.60 and proportion of agreement ranging from 34.4% to 65.5%. Conclusion: The six studied CT splenic injury grading scales did not have a high enough interrater reliability to be adequate for clinical use. The poor interrater reliability is likely to contribute to the failure of the scales at predicting clinical outcomes. Further research to improve the interrater reliability is recommended.