Subspecialty radiology services at tertiary care referral centers are often requested to reinterpret exams performed on patients transferred from outside institutions. These reinterpretations, often called “overreads”, become part of the patient’s permanent medical record and are often requested to help direct patient management in the setting of oncologic, surgical, and trauma care [5, 10].

Previous studies have noted a discrepancy rate up to nearly 20% between the original report and the overread by emergency radiologists in the setting of trauma [5,6,7,8,9,10]. Other radiology subspecialties have made similar observations but with even higher discrepancy rates [3]. A systematic review by Wu et al., published in 2014, analyzed 46 studies focused on double reading of exams. They found that the pooled total discrepancy rate was 7.7% and the major discrepancy rate affecting patient management was 2.4%, though this varied by modality and body region [10]. A meta-analysis by Geijer & Geijer published in 2018 found a very wide discrepancy rate of 0.4 to 22%, though this varied by modality, body region, and subspecialty. They found that the pooled discrepancy rate was 7.7% and a major discrepancy rate of 2.4%. They also found that major discrepancy rates for head CT and spine CT (0.8% and 0.7%, respectively) were lower than those for chest CT and abdominal CT (2.8% and 2.7%, respectively). Interestingly, they found that the lack of blinding of the initial report was associated with a lower major discrepancy rate (2.0% vs. 12.1% for blind reading), suggesting that access to the original report can insert interpreter bias [6].

In our study, we found that nearly a quarter (24.9%) of imaging exams affecting nearly half (49.6%) of trauma patients had either a major or minor discrepancy in the original report, with a major discrepancy rate of 18.5%, which is at the higher end of the range that has been previously reported [3, 5,6,7,8,9,10]. The most commonly encountered discrepancies in our study were fractures or injuries to the bowel/mesentery or solid organs. While the most common change in management was obtaining further imaging studies, over one-quarter of major discrepancies (7 out of 25 [28%], or 2.3% of all imaging exam re-interpretations) resulted in either surgery or an interventional radiology procedure. Motor vehicle collisions, which accounted for 34% of all exams, accounted for 57% of major discrepancies (p = 0.0228 compared to non-MVC mechanism of injury). The greater discrepancy rate in MVC’s may reflect a greater overall complexity of the exams due to the presence of multiple simultaneous injuries and/or the presence of more subtle though clinically relevant findings (such as mesenteric or bowel injuries). Falls, which represented a greater percentage of all exams (42%), represented a lower percentage (19%) of all major discrepancies.

Our findings support the added value of emergency radiologist overreads for transferred trauma patients. This correlates with metanalyses that have shown that the overread is the more accurate interpretation approximately 90.5% of the time [3]. Our data adds to the growing body of literature supporting that discrepancies are greater in emergent patients than non-emergent patients, with rates of discrepancy reported as high as 19.7% in trauma populations in systematic reviews and as low as 2.4% in non-traumatic population systematic reviews [8, 9]. Despite increasing reports like ours, there is still a relative paucity of data regarding trauma patients transferred to level 1 trauma centers. This may be due to the institutional differences in overread procedures [5] or due to difficulties with reimbursement [6].

This study is subject to several limitations, as it is a one-year, retrospective review of imaging exam reinterpretation involving trauma patients transferred to a single level 1 trauma center. Access to the original outside report may introduce bias, as noted by Geijer & Geijer (Geijer 2014). Additionally, requesting a second opinion from a re-interpreting radiologist could also contribute to bias, since patients transferred to this institution are often more severely injured or require a higher level of care.