Presentation Authors: Sorena Keihani*, Bryn Putbrese, Douglas Rogers, Gregory Stoddard, Salt Lake City, UT, Kaushik Mukherjee, Loma Linda, CA, Sarah Majercik, Murray, UT, Joshua Piotrowski, Christopher Dodgion, Milwaukee, WI, Brenton Sherwood, Bradley Erickson, Iowa City, IA, Ian Schwartz, Sean Elliott, Minneapolis, MN, Erik DeSoucy, Scott Zakaluzny, Sacramento, CA, Nima Baradaran, Benjamin Breyer, San Francisco, CA, Brian Smith, Philadelphia, PA, Brandi Miller, Richard Santucci, Detroit, MI, Matthew Carrick, Plano, TX, Jurek Kocik, Tyler, TX, Timothy Hewitt, Frank Burks, Royal Oak, MI, Marta Heilbrun, Atlanta, GA, Raminder Nirula, James Hotaling, Jeremy Myers, Salt Lake City, UT
Introduction: The American Association for the Surgery of Trauma (AAST) injury scale is the most widely used grading system for renal trauma. However, reproducibility of the AAST grading is not well studied. We aimed to assess the agreement for AAST renal trauma grading between radiologists and the grades entered in a multi-center database of high-grade renal trauma.
Methods: Data on high-grade renal trauma (AAST grades III-V) was collected from 14 Level-1 trauma centers from 2014-2017. Patients with initial CT scans were included. 2 radiologists, blinded to the submitted AAST grades and outcomes, reviewed the scans and re-graded the injuries according to the 1989 original AAST grading (O-AAST). After measuring the inter-radiologist agreement, both radiologists were asked to reach consensus on the discrepancies. The reproducibility of AAST grading was evaluated using weighted Kappa analysis for ordinal variables. Inter-radiologists agreement as well as final agreement between re-graded readings (O-AAST) and the injury grades submitted by the centers (C-AAST) were measured. Agreement was interpreted based on the kappa coefficient as slight (0-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), and almost perfect (0.81-1.00).
Results: 322 patients met the inclusion criteria. Injuries were submitted by centers as grade III (59.8%), grade IV (33.1%), and grade V (7.1%). Upon re-grading the injuries, inter-radiologist agreement was substantial for O-AAST (Kappa: 0.76, 95% CI: 0.66-0.84). After resolving the discrepancies, injuries were re-graded using the O-AAST as non-gradable (1.2%), grade I (0.9%), grade II (4.6%), grade III (71.5%), grade IV (16.6%), and grade V (5.2%). Comparing the O-AAST re-grading and C-AAST, the agreement was moderate (Kappa: 0.50, 95% CI: 0.40-0.60). Overall, 27% of injuries were downgraded, 5% were upgraded, and 68% remained the same.
Conclusions: The agreement between center-reported renal injury grades and re-grading of injuries by radiologists blinded to the outcomes was moderate and many injuries were downgraded. These findings have implications for using AAST grading for both predicting the need for interventions, as well as using administrative data, such as the National Trauma Databank for research purposes when AAST grade is an important variable.