Inter- and intrarater reliability in computed axial tomographic grading of splenic injury: Why so many grading scales?

Objective. After splenic trauma, critical decisions regarding operative intervention are often made with the aid of computed axial tomographic (CT) scan findings. No CT scan-based grading scale has been demonstrated to predict accurately which patients require operative or radiologic intervention for their splenic injuries. We hypothesized that use of the most common grading scale, the American Association for the Surgery of Trauma scale, would be associated with low intra- and interreliability scores. We assessed the ability of experi- enced trauma radiologists to differentiate grade III from grade IV splenic injuries. Methods. The films of patients who had undergone abdominal CT scanning before splenectomy for grade III or IV injuries were serially evaluated by four trauma radiology faculty weekly for 3 weeks. We assessed intra- and interrater reliability for grading and for presence of contrast blush. Results: Intrarater reproducibility yielded a weighted kappa score of 0.15 to 0.77. Interrater reliability weighted kappa scores ranged from 0 to 0.84, with a mean value of 0.23. Conclusion. CT imaging is not reliable for identifying grades III and IV splenic injury, as experienced radiologists often underestimate the magnitude of injury. Interrater reliability is poor. Factors other than the CT grade of splenic injury should determine whether patients require operative or angiographic therapy.