Evaluation of Missed Radiological Diagnosis in Multiple Trauma Patients With Full-Body Computed Tomography in the Emergency Department

Introduction: This observational, cross-sectional, and retrospective study was conducted at the Dokuz Eylül University Emergency Department in İzmir, Turkey, after obtaining ethical consent (Dokuz Eylül University Medical Faculty Ethics Committee, approval no. 2019/15-37). In this study, we aimed to determine missed radiological diagnoses and their effects on mortality and morbidity by comparing the ED diagnoses of patients and radiology reports of these patients who presented to the emergency department (ED) with multiple traumas and scanned full-body computed tomography (CT). Materials and methods: This observational, cross-sectional, and retrospective study was conducted at the Dokuz Eylül University Emergency Department in İzmir, Turkey. Adult patients who presented to the ED with trauma between July 1, 2016 and June 30, 2018 and who had a full-body CT were included in the study. Radiology reports of CTs and ED electronic file information were compared. Missed diagnoses were determined for all body parts. Results: In this study, 1,358 patients who had scanned full-body CT in the ED were evaluated. A total of 369 diagnoses were missed in 248 (18.3%) of the patients. The diagnosis-to-patient ratio was 0.27. In the process of individually evaluating pathological diagnoses in all body regions, it was low only in brain edema, pneumomediastinum, bladder injury, and mesentery injury. At least, there was one missed diagnosis in 88 (9.7%) of 907 (66.8%) discharged patients. At least, there was one missed diagnosis in 18/23 (78.3%) patients who died within the first 48 hours. Among the patients who have missed diagnosis, the rate of the discharged patients was 35.5%, patients called back from home was 1.2%, intensive care unit admission was 20.2%, hospitalization was 65.7%, and death was 8.9%. Among the patients who did not have missed diagnosis, the rates were 73.8%, 0%, 5%, 26.9%, and 0.8%, respectively. Conclusion: Thoracic region pathologies are the most frequently missed pathologies, and orthopedics was the most frequently consulted department related to the missed diagnoses. Patients who have a missed diagnosis had lesser discharging from the ED than the other patients and had higher rates of in-hospital deaths, hospitalization, and intensive care unit admission.


Introduction
Missing diagnosis or misdiagnosing in multiple trauma patients delays effective treatment, prolongs hospital stay, and increases mortality [1].In a study examining the misdiagnoses in multiple trauma patients in the literature, the misdiagnosis rate was 16.2%, and the anatomical regions misdiagnosed were the extremities and pelvis, abdomen and pelvic organs, and thoracic regions, in order of frequency [2].However, with the progression in imaging methods, it has been reported that the rate of misdiagnosis of vascular and heart injuries has increased [3,4,5].
Brain, cervical, thorax, and abdomen imaging is often performed together in patients who present with highenergy multiple trauma to the emergency department (ED), and computed tomography (CT) is preferred as the imaging method.In this study, we compared emergency department diagnoses (EDDs) and diagnoses in radiology report (RDs) in multiple trauma patients with full-body CT.We aimed to identify missed radiological diagnoses and determine the effects of these diagnoses on mortality and morbidity.
There was an "almost perfect agreement" between the RDs and EDDs in detecting the presence of any pathology in brain CT imaging, any intracranial hemorrhage (epidural hematoma, subdural hematoma, subarachnoid hemorrhage, intraventricular hemorrhage, and contusion diagnosis), and any fracture (any of the diagnoses of partial/displaced fracture and linear fracture presence).Weak overlapping was found in the diagnosis of cerebral edema.Except for contusion, cerebral edema, and subcutaneous foreign body diagnoses, there was no difference between the RDs and EDDs on brain CT imaging results (
A substantial and almost-perfect agreement was found between the EDDs and RDs of any pathology, C1 fracture, C2 fracture, C3-7 fracture, facet dislocation, occipital condyle fracture, and foreign body in the cervical region (Table 2).There was no difference between the EDDs and RDs in all diagnoses on cervical CT imaging except cervical spinous process fracture (  ) had sternum fracture, and 3/8 patients (37.5%) had pneumomediastinum.Meanwhile, pericardial effusion was correctly diagnosed in 3/7 patients (42.8%) in accordance with the EDDs and RDs.The EDDs and RDs were consistent in all diagnoses of aortic dissection (n = 3), diaphragm rupture (n = 3), foreign body in thorax (n = 1), thoracic vertebral dislocation (n = 1), foreign body in muscle tissue (n = 1), alveolar hemorrhage (n = 1), and bilateral humeral dislocation (n = 1).Thoracic vertebral lamina fracture in three patients and sternoclavicular dislocationacromioclavicular joint dislocation and bronchial rupture in one patient were missed in the ED.
An almost-perfect agreement (k = 0.829) was found for the presence of any diagnosis between the EDDs and RDs in the thorax CT.The fair agreement was found in the diagnosis of pneumomediastinum (Table 3).
There was no difference between the EDDs and RDs in diagnoses other than thoracic pathologies and pneumothorax (  ) were correctly detected with the abdomen CT imaging in the ED.Among the vertebral fractures, lumbar vertebral corpus fracture in 44/51 patients (86.3%), lumbar vertebral lamina fracture in 2/4 patients (50%), lumbar vertebra transverse process fracture in 84/100 patients (84%), lumbar spine spinous process fracture in 7/10 patients (70%) were diagnosed in the ED in accordance with the RDs.Among the pelvic bones, ilium fracture in 20/25 patients (80%), ischial fracture in 4/8 patients (50%), pubis fracture in 58/65 patients (89.2%), acetabular fracture in 29/34 patients (85.3%), and dehiscence of the sacroiliac joint in 3/4 patients (75%) were diagnosed in the ED in accordance with the RD (Table 4).All diagnoses of femur dislocation (n = 5), intestinal perforation (n = 3), foreign body in two patients, renal cyst rupture and pelvic hematoma, traumatic bowel herniation, perianal injury, pancreatic injury, and diasthesis pubis in one patient were all diagnosed correctly in the ED.On the other hand, diagnoses of perirenal hematoma, ovarian cyst rupture, and penile injury were missed in one patient.
Correct diagnosis was made in one each of two patients with psoas hematoma and scrotal injury.An almostperfect agreement (k = 0.845) was found between the EDDs and RDs in evaluating the presence of any pathology.There was no difference between the EDDs and RDs in all abdominal CT diagnoses, except retroperitoneal hematoma and sacrum/coccyx fracture (

Patient outcomes
Among the 1,358 patients included in the study, 907 (66.8%) were discharged, and 88 (9.7%) of these patients had at least one missed diagnosis.Three of these patients were recalled from home for reevaluation, after realizing that the diagnosis was missed.Among the 23 patients who died within the first 48 hours of the application, 18 (78.3%)were patients who had at least one missed diagnosis.Whether death was related to missed diagnoses could not be examined in our study.The discharge rate of the patients who had a missed diagnosis was 35.5%, the rate of being called from home was 1.2%, the rate of admission to the intensive care unit was 20.2%, the rate of hospitalization was 65.7%, the mortality rate was 8.9%, and these values of those without a missed diagnosis were 73.8%, 0%, 5%, 26.9%, and 0.8%, respectively.It could not be reached to the current data of six patients, and they were excluded from the analysis in mortality and morbidity assessments.

Effect of missed diagnoses on morbidity
Among the 248 patients who have missed diagnoses, 13 (3.8%) of them realized that they have a missed diagnosis after hospitalization and additional medical intervention (six patients on anti-edema treatment, two patients on oxygen support, and one patient on low-molecular-weight heparin, shoulder arm sling, sitting ring, Philadelphia-type cervical collar, and thoracic-lumbar-sacral orthosis (TLSO) corset) was performed.Eight of the patients had additional surgical interventions (four patients had tube thoracostomy, two patients had intracranial decompression surgery, one patient had splenectomy, one patient had femur fracture operation).

Discussion
The number of studies in the literature about the effect of missed diagnoses on mortality and morbidity in multiple trauma patients with full-body CT is limited [1,3,6,7,8].
Full-body CT is frequently performed on patients with multiple trauma, and the most common trauma mechanism in similar studies in the literature is traffic accidents.In our study, more than half of the patients (61.8%) presented with in-vehicle and pedestrian traffic and motorcycle accidents, and similar results were found in the studies in the literature.In the studies of Banaste et al. and Yang et al., the rate of patients presenting with falling, which is the second most common mechanism, was 20.7% and 12.4%, respectively; in our study, the rate of patients who presented with falling from a height or the same level was 27.3% [2,9].
In our study, 369 diagnoses were missed in a total of 248 (18.3%) patients in the CT images of 1,358 patients.The diagnosis-to-patient ratio was 1.49.This ratio was 1.48 in the study of Yang et al. and 1.4 in the study of Buduhan et al. [2,3].In the literature review of Pfeifer and Pape, the rate of missed diagnosis varied between 1.3% and 39% [6].This variable ratio in studies in the literature may depend on many factors, such as the functioning of the emergency health system (Anglo-American and Franco-German models), characteristics of the hospital where the patients are evaluated (such as being a trauma center, being a primary/secondary/tertiary hospital, being a university hospital), or the characteristics of physicians evaluating trauma patients (e.g., emergency medicine specialist, general practitioner, other specialists).Our hospital is a fully comprehensive tertiary university hospital.With these features, it is a reference hospital where especially critical patients with poor general conditions from rural districts and cities are referred.This situation is one of the reasons that increase the mortality and morbidity of our patients.In our study, mortality, hospitalization, and intensive care unit admission rates were higher in patients with missed diagnoses.In the study of Yang et al., as the severity of trauma increased, the more rate of missed diagnosis increased [2].We could not give any results about the relationship between the trauma score and mortality and morbidity as trauma scoring was not used in our study.However, we think that the rate of critical patients may be higher than those in other hospitals because we are a reference hospital.
According to the anatomical regions, it was found that the rate of missed diagnosis was the highest (9.3%) in the thoracic region and the least (1.2%) in the cervical region.This may be related to more frequent injuries in the thoracic region and more injuries to the thoracic region that need attention.Although there were minor injuries that did not require intervention, such as simple rib fracture and lung contusion in the diagnoses that were overlooked, there were also major pathologies, such as hemothorax and pneumothorax.It shows the need for training on thoracic CT in traumatic patients, as we have mentioned above.Moreover, it will be beneficial to reporting CT imaging results early in trauma patients.
In our study, it was found that the pathologies in thoracic CT, abdominal CT, and brain CT were missed at most.In the study of Yang et al., the most frequently missed diagnosis was the extremity and pelvis (40%), followed by the abdomen and pelvic organs (20%) and thoracic region (14%).We did not evaluate the extremities in our study.In the study of Yang et al., after the extremities were excluded, abdominal and pelvic structures (52%) were the most frequently missed diagnosis, followed by the thoracic region (23%) and head and neck (18%), respectively.In the studies of Buduhan et al. and Kalemoğlu et al., it was reported that the most frequently missed diagnosis is in the head and thoracic region, excluding the extremities [3,8].In the study of Houshian et al., there was a diagnosis mostly overlooked in the thorax and abdomen after the extremities [10].The head, thorax, and abdominal regions are the most common fatal injuries in multiple traumas.Therefore, it is important to examine the imaging in detail and to get radiological reports urgently.
In both our study and the study of Yang et al., the rate of missed diagnosis in the cervical region was lesser than the other regions (5.1% and 7.5%, respectively).It is an expected result that traumatic cervical pathologies are seen less frequently compared to other regions and missed diagnoses to be proportionally less frequently.
According to the results of brain CT imaging, the detection rate of any intracranial hemorrhage was 90.8%, and the compatibility of RDs and EDDs for intracranial hemorrhage was found almost perfect level (k = 0.870).On the other hand, it was noticed that mistakes can be made about the correct definition of bleeding (such as epidural, subdural, intracerebral, contusion, subarachnoid, and intraventricular), but these mistakes do not lead to clinical consequences that will change patient management.It was noticed that often these patients have intracranial hemorrhage, and they have consulted to the neurosurgery after all.No bleeding was considered in only six patients with subarachnoid hemorrhage, four patients with a cerebral contusion, two patients with subdural hematoma and intraventricular hemorrhage, and one patient with epidural hematoma.A similar situation is valid for the definition of fractures in the cranial bones.While the agreement rate between EDD and RD was almost perfect in recognizing partial/displaced fractures, this ratio was relatively low in linear fractures.On the other hand, linear fractures are often diagnoses that do not change the need for surgical or medical treatment.These results suggest that patients in the ED can be diagnosed with high accuracy.
It was found that there was a slight agreement in the diagnosis of cerebral edema and herniation in brain CT.There were three patients with missed herniation and 29 patients with missed cerebral edema diagnosis.Cerebral edema may be omitted because most of the patients with cerebral edema have other diagnoses, and it is considered relatively unimportant and not written in the file notes or focus on other diagnoses.Emergency physicians and related department consultants evaluating patients in the ED should be more careful in terms of missed diagnoses.
It was observed that injuries, such as C1, C2, cervical vertebra corpus, and occipital condyle fractures, that were diagnosed had a substantial agreement with RD in the ED in cervical CT imaging results.This result is gratifying.We think that early and correct diagnosis of cervical trauma may decrease mortality and morbidity.
Thorax CT imaging results showed that lung contusions were missed more often than pneumothorax and hemothorax.We think that this situation is related with the lesser recording file notes of lung contusions due to their less clinical significance.Accurate evaluation of diagnoses that cause serious mortality and morbidity, such as aortic dissection and diaphragm rupture in all patients, supports this view.In general, the rate of recognizing lung pathologies and their agreement level with RD were found to be substantial, which is also valid for bone structures.On the other hand, the agreement level between the diagnosis of pneumomediastinum in the ED was fair agreement.The reason for this situation may be the uncommon presence of pneumomediastinum and insufficient experience in tomographic evaluation.Less than 50% of the patients with pericardial effusion could also be diagnosed; this may be because of emergency physicians' use for evaluating pericardial effusion with echocardiography or USG with e-FAST protocol and do not pay attention to CT imaging.Attention should be paid to evaluate thoracic pathologies, such as pericardial effusion, pneumomediastinum, and lung contusion in the ED.
It is noticeable that accurate diagnosis of solid organ injuries to splenic injuries in abdominal CT imaging results is higher and more consistent than others.The spleen is a solid organ frequently injured in blunt abdominal trauma [11].In this perspective, it is important to be diagnosed correctly.It was found that the rates of correct diagnosis and agreement levels with RD in patients with mesenteric and bladder injury,

FIGURE 1 :
FIGURE 1: Assessment of discordant diagnoses in brain CT imaging.

TABLE 4 : Abdomen CT imaging outcomes
The diagnosis-to-patient ratio was 0.27.There diagnoses were 83 brain CT pathologies in 73 patients, 19 cervical CT pathologies in 16 patients, 156 thoracic CT pathologies in 126 patients, and 111 abdominal CT pathologies in 84 patients.The number of patients who have a missed diagnosis in the ED and the number and rates of missed diagnoses are shown in Table5.
* McNemar test ⱡ kappa analysis; EDD: emergency department diagnosis; RD: radiology report