Discordant.Reads

Discordance in CT interpretations between pediatric radiology overreads and community and adult radiologists reads following traumatic injury Jack H. Scaife a , Davian J. Martinez a , Christopher E. Clinker a , Sommer L. Glasgow a, Anastasia M. Kahan a , Christopher M. Roach b , Katie W. Russell a,b , Robert A. Swendiman a,b, * a Department of Surgery, University of Utah School of Medicine, Salt Lake City, UT, USA b Primary Children's Hospital, Intermountain Health, Salt Lake City, UT, USA a r t i c l e i n f o Article history: Received 2 October 2025 Accepted 19 October 2025 Keywords: Overread Trauma Overtriage a b s t r a c t Introduction: For pediatric trauma patients, imaging interpretations from transferring institutions may differ from those of pediatric radiologists. At our Level I pediatric trauma center, all outside CT scans are reinterpreted by pediatric radiologists. This study evaluates the discordance rate between initial outside CT interpretations and pediatric overreads, examining whether continued secondary review is needed. Methods: We retrospectively compared CT interpretations of pediatric radiologists and transferring radiologists in pediatric trauma patients transferred to our center from 2021 to 2023. Included patients had a documented injury, an outside interpretation, and a pediatric overread. CT scans reviewed included the head, cervical/thoracic/lumbar spine, chest, and abdomen & pelvis (A&P). Results: A total of 500 patients with 1,192 CT scans were included. Overall discordance was 16 % (189/ 1,192). Patients with discordant scans had higher injury severity scores (ISS 14 vs. 10, p = 0.006). CT A&P had the highest discordance rate at 25 % (101 scans, p < 0.001), and the second highest rate of major treatment changes (38 %, p < 0.001). CT C-spine had the highest rate of discordant scans resulting in major treatment change (63 %, p < 0.001). Only 13 discordant cases would have changed transfer de- cisions; eleven involved CT head scans. Conclusion: Significant discordance exists between outside and pediatric radiologist CT interpretations in pediatric trauma, particularly for CT A&P. Although head CT overreads were the only scans that might have prevented transfer, the overall discordance rate justifies the need for pediatric radiologist review at Level I pediatric trauma centers. © 2025 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). 1. Introduction Traumatic injuries are a leading cause of emergency depart- ment (ED) visits and death in pediatric patients [1,2]. However, access to pediatric-specific care is limited, with only 65 % of American children living within 60 min of pediatric trauma cen- ters, resulting in more than 80 % of pediatric emergency care occurring in non-children's hospitals where specialized pediatric radiology expertise may be limited [3,4]. In the United States, only 3 % of radiologists are subspecialized in pediatrics and are pre- dominantly concentrated in academic centers and metropolitan areas [5]. As a result, children with trauma-related injuries are frequently transferred to pediatric trauma centers after undergo- ing initial imaging interpreted by general or adult radiologists. To address potential variability, it is common practice at many pedi- atric centers to secondarily review outside imaging by fellowship- trained pediatric radiologists. At our Level I pediatric trauma center, all outside CT scans are reinterpreted by pediatric radiol- ogists. The time spent waiting for this scan can be less than 10 min or longer than an hour. It should be noted that only stable patients will wait for transfer until scans have been overread. Unstable patients are transferred immediately. Previous studies have assessed the disagreement rate in over- reads, but none have looked at predictors of discordant reads. Discordance rates in all overreads vary from 12 to 41 %, while the discordance rate for abdominal CTs ranges from 27 to 54 % [6—8]. * Corresponding author. 100 North Mario Capecchi Drive, Suite 3800, Salt Lake City, UT 84113, USA. E-mail address: Robert.swendiman@hsc.utah.edu (R.A. Swendiman). Contents lists available at ScienceDirect Journal of Pediatric Surgery journal homepage: www.sciencedi rect.com/journal/ journal-of-pediatric-s urgery https://doi.org/10.1016/j.jpedsurg.2025.162765 0022-3468/© 2025 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). Journal of Pediatric Surgery 61 (2026) 162765

Additionally, many of these discordant reads have been noted to have a significant influence on the care of these patients. No large studies have assessed the rate of disagreement in pediatric trauma patients, and none have assessed patient factors that may contribute to higher rates of discordant reads. In this study, our aim was to determine the rates of discordance between initial CT imaging interpretation and reinterpretation by an attending pediatric radiologist at a pediatric level I trauma center for transferred patients. Additionally, we sought to establish the impact that discordance had on clinical care. We stratified the clinical impact of overreads into major and minor categories. Major impacts included findings that necessitated immediate subspe- cialty referral, a change in admission status, additional urgent im- aging, or instances where the overread would have prevented patient transfer. Minor impacts referred to changes such as nonurgent treatment, outpatient referral, or any care modifications that could safely be delayed up to 24 h without adverse conse- quences. Finally, we stratified discordance rates and clinical impact according to body system scanned. We hypothesized that a sig- nificant proportion of the reads would be discordant, but the impact on clinical care of these discordant reads would be relatively low. 2. Methods and materials The Institutional Review Board approved this study. The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. 2.1. Study design and data source We retrospectively analyzed the trauma registry at an Amer- ican College of Surgeons (ACS) verified Level I pediatric trauma center. The data set includes all children transferred to the Level I pediatric trauma center from another facility from September 2021 through October 2023 for traumatic injuries. Eligible pa- tients had undergone outside hospital CT imaging with both initial interpretation and corresponding pediatric radiologist overread available in the electronic health record (EHR) (see Fig. 1). The cross-sectional imaging modalities included were CT head, CT cervical spine, CT thoracic spine, CT lumbar spine, CT chest, and CT abdomen/pelvis (A&P). Patients with qualifying studies on multiple dates were included for each unique encounter. We excluded cases that only had imaging performed at the Level I pediatric trauma center, lacked a documented outside interpretation, or involved other imaging modalities (e.g., CT maxillofacial, extremity CTs, MRI, ultrasound, or radiographs). Patient age, race, ethnicity, and ISS were collected from the trauma registry. Age in years was categorized into the age groups of less than 10 years old, 11—17 years old, and older than 17 years old. ISS was categorized into mild if the ISS was less than 9, moderate for ISS between 9 and 15, severe for 15—25, and major for ISS greater than 25. 2.2. Primary outcome Each outside interpretation was compared to the pediatric ra- diologist's overread. The pediatric radiologist's interpretation was considered the reference standard. Discordance was defined using the following criteria: 1. A change in interpretation of injury severity; 2. Omission of a finding identified in the overread; 3. Inclusion of a finding not identified in the overread; 4. An explicit disagreement or reinterpretation as stated by the pediatric radi- ologist. In cases of ambiguous interpretations, patients were initially designated as discordant. Studies not meeting these criteria were classified as concordant. All discordant encounters identified were re-evaluated and verified by one of two board- certified pediatric trauma surgeons (RAS and KWR). 2.3. Secondary outcomes To determine the clinical significance of discordant reads, we categorized discordance as being clinically insignificant or signifi- cant, as adapted from previously published methodologies (see Fig. 1. Consort diagram outlining categorization of scans into discordance, level of significance, and impact on treatment. Table 1 This table demonstrates examples of each discordance category used in this study. Cases are categorized as No Discordance (no difference between outside hospital [OSH] and pediatric over-read), Minor Discordance (findings of limited clinical relevance), Major Discordance (missed or overcalled findings with potential clinical impact), and Transfer-Altering discordance (discordance that would have changed the need for transfer). OSH = outside hospital; ED = emergency department; ICU = intensive care unit. Type OSH Diagnosis Over-read Diagnosis Key Difference Clinical Outcome No discordance Subdural hematoma Subdural hematoma No change Admitted to the floor Minor discordance Skull fracture Vascular groove Overcalled skull fracture Discharged from ED Major discordance Normal Skull fracture and extra-axial hematoma Missed skull fracture and hemorrhage Admitted to ICU, repeat imaging Transfer-Altering Midline shift No midline shift Overcalled the critical finding Discharged from ED J.H. Scaife, D.J. Martinez, C.E. Clinker et al. / Journal of Pediatric Surgery 61 (2026) 1627652

Table 1 for examples) [8,9]. Clinically insignificant changes were defined as additional findings with no potential to change man- agement or incidental findings not requiring follow-up. Clinically significant reads were defined as differences with the potential to alter clinical management or affect patient outcomes. Clinically significant reads were then sub-categorized into major and minor changes in treatment. Major changes included changes that would require immediate sub-specialty referral, a change in admission status, additional urgent imaging, or cases where the overread would have prevented transfer. Minor changes included nonurgent treatment, outpatient referral, or any change in care that would not be adversely affected by a 24-h delay. All categorizations were independently reviewed by two board-certified pediatric trauma surgeons. Additionally, we sought to understand the effect of discordant reads on unnecessary transfers of patients. Therefore, in all pa- tients whose read led to a major change in treatment, it was assessed whether the discordant aspect of the read would have resulted in a change in the decision to transfer the child. These cases were categorized as a change in transfer pattern. Addition- ally, among these patients, we determined whether the discor- dance resulted in an unnecessary transfer and thus categorized these reads as discordant, resulting in unnecessary transfer. 2.4. Statistical analysis All statistical analyses were performed using R (Version 4.3.1) [10]. Hypotheses were tested using a two-sided approach, with p < 0.05 considered statistically significant. Univariate analyses were performed using Wilcoxon rank sum tests for continuous variables and chi-squared tests for categorical variables. Patients with missing ISS were excluded from ISS univariate analysis. A multivariable regression was performed to measure the odds of discrepant reads. We performed a generalized linear model, which included imaging type, age group, and ISS group. These variables were chosen a priori and based on univariate significance. Patients with missing ISS were excluded from the multivariable model. 3. Results Out of 2,116 pediatric trauma patients transferred to our facility during the study period, 500 met the inclusion criteria. These patients had 1,192 imaging studies with an outside hospital interpretation and a pediatric subspecialist overread (see Table 2). CT head was the most common imaging study (33.8 %). The ma- jority of our patients were white (77 %), non-Hispanic (72 %), and male (63 %). The median patient age was 11.7 years (IQR 5—15), and the median Injury Severity Score (ISS) was 10 (IQR 5—14). In total, 189 of 1,192 studies (15.8 %) demonstrated discordance between the outside and pediatric radiologist interpretations. CT A&P had the highest rate of discordant reads (26 %), while CT cer- vical spines had the lowest rate of discordant reads (6.2 %) (Table 3). The differences in the rate of discordant reads were statistically significant. Additionally, patients with discordant reads had a significantly higher ISS than those with concordant reads (14 vs 10, p < 0.001). Neither race nor age was significantly associated with discordant reads. On multivariate analysis, CT A&P had significantly higher odds of discordant reads in comparison to every other im- aging type, other than CT chest, which had no significant difference Table 2 This table demonstrates the demographic and imaging characteristics of the concordant and discordant CT Read groups. Continuous variables are presented as median (interquartile range). Categorical variables are presented as a number (percentage). ISS = Injury Severity Score CT = Computed Tomography. Characteristic Concordant Reads, N = 978 a Discordant Reads, N = 191 a p-value b Age (years) 12.9 (7.3, 15.3) 12.1 (6.7, 15.1) 0.2 Age groups 0.086 <10 years 326 (81 %) 77 (19 %) 10—16 years 598 (86 %) 99 (14 %) 17 years 56 (81 %) 13 (19 %) ISS 10 (6, 17) 14 (9, 17) <0.001 ISS groups Mild (ISS<9) 269 (89 %) 32 (11 %) Moderate (9—15) 436 (85 %) 78 (15 %) Severe (16—25) 160 (77 %) 49 (23 %) Major (>25) 436 (85 %) 78 (15 %) Unknown 17 8 Race 0.6 Asian 7 (88 %) 1 (13 %) Black 20 (71 %) 8 (29 %) Native American 46 (87 %) 7 (13 %) Unavailable 137 (84 %) 26 (16 %) Polynesian 10 (83 %) 2 (17 %) White 758 (84 %) 147 (16 %) Ethnicity >0.9 Hispanic 106 (84 %) 20 (16 %) Non-Hispanic 703 (84 %) 136 (16 %) Unavailable 169 (83 %) 35 (17 %) Imaging type <0.001 CT abdomen & pelvis 159 (74 %) 56 (26 %) CT cervical spine 243 (94 %) 16 (6.2 %) CT chest 136 (81 %) 31 (19 %) CT head 330 (82 %) 74 (18 %) CT lumbar spine 57 (90 %) 6 (9.5 %) CT thoracic spine 53 (87 %) 8 (13 %) a n (%); Median (IQR). b Pearson's Chi-squared test; Wilcoxon rank sum test; Fisher's exact test. Table 3 This table compares discordance grade, treatment changes, and transfer outcomes across different CT imaging types among transferred pediatric trauma patients. Values are presented as a number (percentage). CT = Computed Tomography. Characteristic CT Abdomen & Pelvis N = 56 a CT Chest, N = 31 a CT Head, N = 74 a CT C-Spine, N = 16 CT T-Spine, N = 7 CT L-Spine, N = 5 p-value b Discordance grade <0.001 Insignificant 26 (46 %) 24 (77 %) 45 (61 %) 4 (25 %) 3 (43 %) 3 (60 %) Significant 30 (54 %) 7 (23 %) 29 (39 %) 12 (75 %) 4 (57 %) 2 (40 %) Treatment change 0.001 Major 21 (38 %) 2 (6.5 %) 17 (23 %) 10 (63 %) 0 (0 %) 0 (0 %) Minor 9 (16 %) 5 (16 %) 13 (18 %) 2 (13 %) 4 (57 %) 2 (40 %) None 26 (46 %) 24 (77 %) 44 (59 %) 4 (25 %) 3 (43 %) 3 (60 %) Isolated injury transfer change <0.001 No 46 (82 %) 30 (97 %) 60 (81 %) 7 (44 %) 7 (100 %) 5 (100 %) Yes 10 (18 %) 1 (3.2 %) 14 (19 %) 9 (56 %) 0 (0 %) 0 (0 %) Actual transfer change 0 (0 %) 0 (0 %) 11 (42 %) 2 (12.5 %) 0 (0 %) 0 (0 %) <0.001 a n (%). b Pearson's Chi-squared test; Fisher's exact test. J.H. Scaife, D.J. Martinez, C.E. Clinker et al. / Journal of Pediatric Surgery 61 (2026) 162765 3

(see Table 4). Patients aged 10—16 had lower odds of having in comparison to children younger than 10 years old (OR [CI 95 %]: 0.68 [0.48,0.97]; p = 0034). Also, children with severe and major ISS had higher odds of discordant reads in comparison to children with minor ISS (severe OR [CI 95 %]: 2.51 [1.53, 4.18], p < 0.001; major OR [CI 95 %]: 2.04 [1.09, 3.76], p = 0.023). Among the discordant reads, discordant CT C-spine reads had the highest rates of significance (75 %), followed by CT A&P (54 %) (p = 0.002). CT C-spine and CT A&P also had higher frequencies of overreads resulting in major treatment changes compared to CT head (p = 0.001). However, no discordant reads for CT A&P or CT Chest would have led to differences in the actual transfer decision. In comparison, 13 discordant scans would have resulted in a change of transfer. Eleven were CT heads, while 2 were CT C-spine (Table 5). Of these discordant CT head reads, 10 were overcalled by the outside radiologist, which led to an unnecessary transfer, while only one overread found injuries that warranted transfer in a pa- tient with a previously normal scan. Of the CT C-spine reads, for patient #12, the outside read had concern for a ligamentous injury, while the overread was normal. However, an MRI confirmed the outside read in this case, and the patient was taken for neuro- surgical intervention. For patient #13, the outside read had concern for the atlantooccipital ligamentous injury, and the pa- tient was transferred; however, the overread was read as normal, and the patient was discharged from the ED. 4. Discussion This study found a discordance rate of 15.8 % between outside radiologists and pediatric subspecialists in CT interpretations for transferred pediatric trauma patients. CT abdomen and pelvis had the highest rate of discordant reads, while CT C-spine had the lowest rate of discordant reads. Patients with more severe injuries also had higher rates of discordant reads. Discordant CT C-spine and CT A&P reads had the highest rate of resulting in a major treatment change; however, none of the CT A&P reads would have resulted in a change in the transfer status of the patient. Only Table 4 This table displays the results of a multivariable analysis on the odds of having disagreement between pediatric radiologist and transferring hospital radiologist reads. Imaging type, age group, and ISS group were included in the model. Characteristic OR 95 % CI p-value Imaging type) CT abdomen & pelvis — — CT cervical spine 0.19 0.10, 0.35 <0.001 CT chest 0.66 0.40, 1.09 0.11 CT head 0.63 0.42, 0.96 0.029 CT lumbar spine 0.33 0.12, 0.77 0.017 CT thoracic spine 0.33 0.12, 0.76 0.016 Age groups 0.086 <10 years — — 10—16 years 0.68 0.48, 0.97 0.034 17 years 1.07 0.52, 2.06 0.8 ISS groups Mild (ISS<9) — — Moderate (9—15) 1.45 0.94, 2.29 0.10 Severe (16—25) 2.51 1.53, 4.18 <0.001 Major (>25) 2.04 1.09, 3.76 0.023 Table 5 This table shows the clinical details of pediatric trauma patients with discordant CT head or cervical spine interpretations. For each patient, the initial OSH diagnosis, subspecialist over-read diagnosis, nature of the discrepancy, hospital stay duration, associated injuries, and clinical outcomes are listed. “Overcalled” indicates an over- diagnosis at the OSH, while “Missed” indicates findings identified only on subspecialist review. CT = Computed Tomography, OSH = Outside Hospital, Dx = Diagnosis, EDH = Epidural Hematoma, SDH = Subdural Hematoma, ED = Emergency Department, MRI = Magnetic Resonance Imaging. Patient Number Age CT Body region OSH Dx Over-read Dx Difference Hospital Stay (days) Other Injuries Outcome 1 5 years Head Epidural Normal Overcalled EDH 1 Femur fracture Discharged from ED 2 13 years Head Subdural hematoma Midline shift Missed Midline Shift 1 None Admitted to ICU, Follow up imaging 3 6 months Head SDH Artifact Overcalled SDH 1 None Discharged from ED 4 8 weeks Head Skull Fracture Vascular Groove Overcalled Skull Fracture 1 None Discharged from ED 5 11 months Head SDH, Skull Fracture Midline Shift SDH, Skull Fracture, No midline Shift Overcalled Midline shift 1 Frontal Scalp Hematoma Admitted to Floor 6 11 years Head Normal Skull Fracture, Extra- axial hematomas Missed Skull Fracture and Hematoma 2 None Admitted to ICU, Repeat imaging 7 8 years Head Frontal Lobe contusion Normal Overcalled hemorrhagic contusion 1 None Admitted to floor 8 13 Month Head Occipital Lobe Fracture Normal Overread skull Fracture 1 None Discharged From ED 9 3 years Head SDH Normal Overcalled SDH 1 None Discharged From ED 10 3 years Head Subarachnoid hemorrhage Streak Artifact hemorrhage Overcalled SAH 1 None Admitted to Floor 11 18 months Head Skull fracture, Extra axial hemorrhage Normal Overcalled Skull Fracture & Extra axial hemorrhage 1 None Admitted to ICU, X-rays of L spine and C spine 12 13 years C-Spine Trace anterolisthesis C5-6, with slight anterior subluxation at facet joints. Reversal of lordosis and widening of interspinous distance Mild loss of cervical lordosis secondary to cervical collar Overcalled concern for ligamentous injury 4 None MRI confirmed ligamentous injury and patient had operative intervention with Neurosurgery 13 21 months C-Spine Atlantooccipital ligamentous injury with widening of atlantooccipital intervals Normal Overcalled the atlantooccipital ligamentous injury 1 None Neurosurgery consulted and patient was discharged from the ED J.H. Scaife, D.J. Martinez, C.E. Clinker et al. / Journal of Pediatric Surgery 61 (2026) 1627654

discordant reads for CT head and C-spine would have prevented unnecessary transfers. Our overall discordance rate findings are supported by a pre- vious study by Miller et al., who found a discordance rate of 14.1 % among a similar population. This agreement suggests that the discordance rate among overreads remains around 15 % and is consistent across regions. However, there were significant differ- ences in the discordance rates by body type imaged, with CT A&P having the highest discordance rate at 26 %. Previous studies by Onwubiko et al. assessing the discordance rate in CT A&P per- formed in transferred patients found a discordance rate of 19 % [7]. The increased discordance rate in our study may be due to the widened definition of discordance compared to Onwubiko et al. Additionally, we found that CT A&P had the highest rate of sig- nificant disagreements, which were more likely to have major impacts on the patient's treatment plans. The rate of major dis- agreements found in other states has been between 18 and 35 %. In our study, the rate was significantly lower. This result may be due to the fact that we have only looked at traumatic injury, while the other studies have included patients transferred for non-traumatic reasons. The misses on outside reads can have significant conse- quences, as shown by the rate of major treatment changes caused by the findings on the overreads. However, none of these dis- agreements would have resulted in a transfer decision change when accounting for the child's other injuries. Therefore, post- poning a transfer while awaiting an overread is likely unwarranted. We found that cervical spine CT discordances, while less frequent overall, were highly clinically significant. Our study found that 75 % of discordant cervical spine interpretations resulted in major treatment changes. Additionally, discordances in C-spine imaging were associated with the highest proportion of potential changes in transfer status (56 %) and actual changes (12.5 %). These findings are significant given the vulnerability of the pediatric cervical spine [11,12]. Pediatric cervical spine in- juries are rare but carry a disproportionately high risk of morbidity and mortality, particularly at the craniocervical junc- tion (occiput—C2), where injuries such as atlantooccipital dissociation and atlantoaxial subluxation are more common in children due to anatomical and biomechanical differences [13]. Timely pediatric radiology review is therefore critical, as failure to identify injuries such as atlantooccipital interval widening, C1—C2 subluxation, or facet fractures can result in inappropriate collar removal, missed surgical indications, and preventable neurologic decline [14]. In our cohort, multiple discordant reads involved instability patterns, including atlantooccipital interval widening, ligamentous injury with MRI recommendation, and subtle anterolisthesis. Most of these children with disagreeing reads were rapidly discharged, indicating an unnecessary trans- fer; however, one patient with a CT scan read as normal required operative treatment for atlantooccipital interval widening, indi- cating that the overread was incorrect, while the initial outside read was correct. These findings suggest that pediatric radiology review of cervical spine imaging is extremely valuable for guiding management. Still, when the original and overread disagree, given the significant consequences, it would be recommended to treat the patient conservatively by giving fidelity to the abnormal scan. We found that despite CT heads having a lower rate of discor- dant reads than CT A&P, CT head discordant reads represented most of the overreads that could have prevented an unnecessary transfer. Our study found a rate of discordant reads of 18 %. This rate is within the previously reported ranges of 14—34 % for CT head studies [6,8]. Additionally, in our study, disagreeing reads led to major treatment changes in only 4 % of overread scans in comparison to the 16 % found by Eakins et al. [6] Once again, this difference may relate to our study only including trauma patients in comparison to all comers, which may indicate that there are a lower number of major finding differences in patients with trau- matic injuries. However, to our knowledge, no study has assessed the impact of overreads on transfer decision-making. In this study, 10 patients were unnecessarily transferred, while one patient was under-transferred based on changes in the read made by the pe- diatric radiologist. Most unnecessarily transferred patients were discharged from the ED, and none had a hospital stay longer than 1 day. This finding indicates that in children with isolated head in- juries, an overread by a pediatric radiologist may be warranted to ensure that the patient is not unnecessarily transferred. Additionally, we found that more severely injured patients were significantly more likely to have discordant reads. This finding is consistent with previous findings in the adult literature assessing the accuracy of CT in adult trauma patients. Banaste et al. found that a higher clinical severity class and the presence of more than two injured body parts were independent predictors of missed injuries on whole-body CT in trauma patients [15]. Simi- larly, Eurin et al. and Kok et al. found that higher ISS was signifi- cantly associated with missed injuries [16,17]. These findings indicated that in more severely injured patients, it is likely that injuries are missed and these scans deserve to be overread; however, waiting for an overread to guide transfer decision-mak- ing in these patients is not necessary, as these patients’ stability is tenuous and their known injuries already necessitate transfer. We must address a significant number of limitations to this study. First, pediatric radiologist interpretations were considered the reference standard, but no third-party arbitration was used, so the accuracy of either interpretation cannot be definitively established. However, we did not comment on the validity of either read and instead only noted instances in which the reads disagreed, but previous studies have indicated that fellowship- trained pediatric radiologists are more accurate than those without pediatric-specific training [18,19]. Additionally, other radiologic factors were not evaluated, including the experience level of outside radiologists and whether discordance rates varied based on the imaging facility. Without this data, we are unable to ensure that outside reads are not read by radiologists with pedi- atric-specific training. Finally, this is a retrospective study at a single institution, which limits the generalizability of this study, but given that the rate of discordance found in this study is in alignment with other previous studies, it is unlikely that the rate found in this study differs significantly at other institutions [7,8]. 5. Conclusion Among injured children, we found a significant rate of discor- dance between initial CT imaging interpretation and reinterpre- tation by an attending pediatric radiologist at a Pediatric Level I trauma center. CT A&P had the highest rate of discordant reads and led to the most major changes in the treatment plan. However, cervical spine and head CT discordances, while less frequent, were more likely to result in major management changes and could have prevented unnecessary transfers or altered transfer de- cisions. These data support the need for injured children to receive overreads of CT scans of the head, cervical spine, and abdomen and pelvis upon arrival at a Level I pediatric trauma center. Over- reading prior to transfer might have prevented transfer in a small number of patients. J.H. Scaife, D.J. Martinez, C.E. Clinker et al. / Journal of Pediatric Surgery 61 (2026) 162765 5

Conflict of interest The authors declare no competing financial interests or per- sonal relationships that could have influenced the work reported in this manuscript. References [1] National hospital ambulatory medical care survey: 2022 emergency department summary tables. 2022. [2] Goldstick JE, Cunningham RM, Carter PM. Current causes of death in children and adolescents in the United States. N Engl J Med May 19 2022;386(20): 1955—6. https://doi.org/10.1056/NEJMc2201761. [3] Myers SR, Branas CC, French B, Nance ML, Carr BG. A national analysis of pediatric trauma care utilization and outcomes in the United States. Pediatr Emerg Care 2019;35(1):1—7. https://doi.org/10.1097/ pec.0000000000000902. [4] Melhado C, Hancock C, Wang H, Eldin MM, George N, Miller JA, et al. Pediatric readiness and trauma center access for children. JAMA Pediatr Feb 3 2025. https://doi.org/10.1001/jamapediatrics.2024.6058. [5] Turner A, Ricketts T, Leslie LK. Comparison of number and geographic dis- tribution of pediatric subspecialists and patient proximity to specialized care in the US between 2003 and 2019. JAMA Pediatr 2020;174(9):852. https:// doi.org/10.1001/jamapediatrics.2020.1124. [6] Eakins C, Ellis WD, Pruthi S, Johnson DP, Hernanz-Schulman M, Yu C, et al. Second opinion interpretations by specialty radiologists at a pediatric hos- pital: rate of disagreement and clinical implications. AJR Am J Roentgenol Oct 2012;199(4):916—20. https://doi.org/10.2214/AJR.11.7662. [7] Onwubiko C, Mooney DP. The value of official reinterpretation of trauma computed tomography scans from referring hospitals. J Pediatr Surg Mar 2016;51(3):486—9. https://doi.org/10.1016/j.jpedsurg.2015.08.006. [8] Miller JP, Ivanics C, Zalewski K, Mody SS, Kannikeswaran N. Rates and clinical impact of discordant X-ray and CT imaging in transfers to a pediatric emergency department. Am J Emerg Med Nov 2021;49:166—71. https://doi. org/10.1016/j.ajem.2021.05.063. [9] Stevens KJ, Griffiths KL, Rosenberg J, Mahadevan S, Zatz LM, Leung AN. Discordance rates between preliminary and final radiology reports on cross- sectional imaging studies at a level 1 trauma center. Acad Radiol Oct 2008;15 (10):1217—26. https://doi.org/10.1016/j.acra.2008.03.017. [10] R: a language and environment for statistical computing. R Foundation for Statistical Computing; 2023. https://www.R-project.org/. [11] McAllister AS, Nagaraj U, Radhakrishnan R. Emergent imaging of pediatric cervical spine trauma. Radiographics Jul-Aug 2019;39(4):1126—42. https:// doi.org/10.1148/rg.2019180100. [12] Jeong D, Beno S, Drake JM, Freire G. Clinical characteristics of upper cervical spine injuries in children: a retrospective case series. J Neurosurg Pediatr May 1 2024;33(5):452—60. https://doi.org/10.3171/2023.12.Peds23197. [13] Little SB, Sarma A, Bajaj M, Dennison J, Brahma B, Pruthi S. Imaging of cra- niovertebral junction instability, fixation, and stenosis in children. Radio- graphics Apr 2025;45(4):e240075. https://doi.org/10.1148/rg.240075. [14] Riascos R, Bonfante E, Cotes C, Guirguis M, Hakimelahi R, West C. Imaging of Atlanto-Occipital and atlantoaxial traumatic injuries: what the radiologist needs to know. Radiographics Nov-Dec 2015;35(7):2121—34. https://doi.org/ 10.1148/rg.2015150035. [15] Banaste N, Caurier B, Bratan F, Bergerot JF, Thomson V, Millet I. Whole-body CT in patients with multiple traumas: factors leading to missed injury. Radiology Nov 2018;289(2):374—83. https://doi.org/10.1148/ radiol.2018180492. [16] Eurin M, Haddad N, Zappa M, Lenoir T, Dauzac C, Vilgrain V, et al. Incidence and predictors of missed injuries in trauma patients in the initial hot report of whole-body CT scan. Injury Jan 2012;43(1):73—7. https://doi.org/10.1016/ j.injury.2011.05.019. [17] Kok D, Oud S, Giannakopoulos GF, Scheerder MJ, Beenen LFM, Halm JA, et al. Delayed diagnosed injuries in trauma patients after initial trauma assess- ment with a total-body computed tomography scan. Injury May 2024;55(5): 111304. https://doi.org/10.1016/j.injury.2023.111304. [18] Martino AM, Yeates EO, Grigorian A, Chinn J, Young H, Colin Escobar J, et al. Comparing accuracy of night radiology interpretations for pediatric trauma: radiology residents versus attending teleradiologists. Am Surg Oct 2024;90 (10):2436—41. https://doi.org/10.1177/00031348241248794. [19] Yeates EO, Grigorian A, Chinn J, Young H, Colin Escobar J, Glavis-Bloom J, et al. Night radiology coverage for trauma: residents, teleradiology, or both? J Am Coll Surg Sep 1 2022;235(3):500—9. https://doi.org/10.1097/ XCS.0000000000000280. J.H. Scaife, D.J. Martinez, C.E. Clinker et al. / Journal of Pediatric Surgery 61 (2026) 1627656