Contribution of Brain MRI after Head CT in the Evaluation of Acute, Hospitalized Traumatic Brain Injury Patients: An Under-Utilized Resource? Brain after the Evaluation of Acute, Brain Injury Patients: An UnderUtilized Resource?.

Contribution of MRI Head CT in Traumatic Abstract Purpose: Non Contrast Head CT (NHCT) has long been the study of choice for initial imaging of patients presenting with acute Traumatic Brain Injury (TBI). However, MRI has been used more frequently as imaging times have decreased. The purpose of this study was to assess the efficacy of lesion detection between NHCT and MRI as well as to estimate the frequency of MRI utilization in routine clinical practice during the care of acute TBI patients. Methods: A retrospective cohort of patients presenting with acute TBI to a level 1 trauma center between November 2007 and December 2013 who received an NHCT and subsequently received brain MRI. The electronic medical record (EMR) was reviewed to assess indications


Introduction
Imaging plays a crucial role in the initial evaluation of moderate and severe Traumatic Brain Injury (TBI) patients. Non-Contrast Head CT (NHCT) has long been the initial step of imaging management in the acute traumatic brain injury patient [1] because it has high sensitivity and specificity for intracranial bleeds, cerebral edema, and skull fractures and it is quick and cost-effective. Despite its utility, head CT has limitations: it may underestimate the degree of injury in the acute setting, and very small amounts of blood may be missed due to volume averaging or artifact that can obscure hemorrhages [2,3]. Previous studies show the efficacy of MRI in patients who present with TBI. Prior retrospective data from emergency department patients presenting with TBI demonstrate 96.4% sensitivity and 63.4% sensitivity with MRI and CT, respectively [4,5]. MRI was more sensitive for detecting contusion, shearing injury, subdural hematoma, epidural hematoma, and sinus involvement. CT was more sensitive for detecting skull fracture [5].
Our hypothesis is twofold: first, we hypothesize that because of NHCT's limitations, brain MRI (bMRI) may be the superior study for acute TBI management. Second, we hypothesize that brain MRI is underutilized in clinical practice, given its sensitivity and specificity for lesion detection. Brain MRI's sensitivity may result in earlier detection of acute findings.

Procedures
Radiological Information System (RIS) produced a report that identified all patients who received a trauma protocol NHCT in conjunction with brain MRI between November 2007 and December 2013. The brain MRI could be performed any time after the NHCT and before the patient was discharged from the hospital.
If multiple NHCTs were ordered, we compared the NHCT that was performed immediately prior to the bMRI. We chose to compare findings from the most recent NHCT so that delayed findings (e.g. delayed or blossoming contusions) would be accurately identified on both imaging studies. In order to determine the frequency of MRI ordering in a random two-year period, we ran a query through the Boston Medical Center Trauma Database to determine the number of patients admitted for traumatic injuries. Initially, 1235 subjects met criteria for admission, and the search was refined to include subjects who had TBI included as one of their injuries.
TBI was defined by the treating clinician -if present, the clinician would subsequently fill out database fields documenting Glasgow Coma Scale (GCS). If absent, these fields are blank. Six hundred ten subjects met the criteria for TBI. The subjects with TBI were reviewed with the Philips Search Tool (query: CT, MRI) to determine which subjects received head CTs and bMRI.

Data Acquisition
The initial NHCT was performed on a GE Lightspeed VCT

Terms and Measures
We reviewed clinical notes to determine the patient presentation and brain MRI indications. Patients were categorized as moderate or severe TBI. Moderate TBI was defined as GCS9-12 and an NHCT without changes due to the mechanism of injury.
Severe TBI was defined as an abnormal NHCT (i.e., any finding that was deemed directly related to the mechanism of injury) and/or New MRI findings were defined as any finding not present on the initial head CT. We further went on to categorize findings on MRI

Statistical Analysis
To compare the incidence of findings from NHCT to overall findings from bMRI, a one-sample test of proportion was conducted to test the hypothesis that the observed incidence of findings from bMRI was different from the incidence observed with NHCT. Exact 95% confidence interval (95% CI) was calculated for the bMRI proportion. A P-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using SAS 9.4 (SAS Institute, Cary, NC).

Results
The most common clinical indications for obtaining a bMRI

Severe TBI Patients
There were 199 patients categorized as severe TBI (Figure 3).

Moderate TBI Patients
All 119 moderate TBI patients received NHCT and subsequent brain MRI. In the moderate TBI patients, 28/119 (23.5%) had findings evident only on brain MRI (Figure 3).

Discussion
In our study, we observed that patients with moderate and severe TBI often have imaging findings not visualized on the initial head CT. New findings on bMRI were more common in patients with severe TBI, but were still frequent in both groups (53.3% in severe TBI and 23.5% in moderate TBI). The most common finding is DAI (44.0%), the second most common finding is infarct (27.6%).
The bMRI visualized findings on 42.1% of all patients with TBI regardless of severity that was not seen on NHCT, and the overall findings on bMRI were significantly higher than on NHCT. These results support our hypothesis that brain MRI will reveal additional findings compared to admission head CT. Our study also shows that bMRI is likely underutilized for clinical care. Based on our convenience sampling data, in a two year period, we perform bMRI on 13.4% of all patients with disease severe enough to warrant inpatient admission. If our study findings are accurate; however, then we would expect 42.1% of all bMRIs to have additional findings.

This translates into findings on 257 (42.1% of 610) patients in a two
year time period. It also means that there may be as many as 157 [210 -(82+18)] patients with undiagnosed findings because they did not receive bMRI. While these calculations are speculative, they illustrate the point that there is likely a large gap of patients with missed findings that could have been captured with bMRI. Other studies support this estimate, quoting rates of additional findings on bMRI as high as 67%.
Early identification of DAI is important for patient management because it can influence family counseling and recommendations regarding prognosis. A prospective study performed on patients with confirmed DAI evaluated 6-month outcomes. The study associated DAI with increased risk for both mortality and dependence [6]. These associations can have a significant impact on patients and their families. Most trauma patients are young men in the workforce; the median age in one of the largest trials to date was 39 years [7]. Hence, even small cognitive changes may affect their capability for full-time employment. Because brain MRI findings such as DAI may guide prognosis, clinicians may start post-TBI education earlier with families. Early education allows families more time to plan for rehabilitation stays and possible loss of income [8]. Timely identification and treatment of stroke are paramount for best patient care. Stroke in the setting of trauma may suggest dissection or occult infection [9]. It may also suggest an underlying etiology preceding the traumatic event, such as a stroke leading to a car accident. Most patients who suffer from TBI associated ischemic stroke are polytrauma patients with fractures and abdominal-thoracic injuries. The decision making algorithms for these patients is often complex and requires specialist expertise [10]. Therefore, rapid stroke identification, early specialist consultation, and subsequent stroke treatment will result in better overall patient care.
Our findings are consistent with prior studies. Data from pediatric populations that presented with TBI also show that MRI has a superior ability to pick up radiographic findings missed on head CT [11]. Retrospective data from adult ED patients presenting with TBI showed higher sensitivity of MRI in identifying nonhemorrhagic lesions (DAI, contusion, brainstem injury) but also noted the advantage of CT in patients who are unstable that might need surgery [12,13]. There are several limitations to our study.
First, this is a single institution, retrospective study limiting the generalizability of our findings. A more extensive study, utilizing data collected from several different institutions would help rectify this limitation in the future. We did not collect any information about gender, past medical history, and pre-existing conditions. It is possible that our sample may have contained a disproportionate number of patients with conditions such as cancer or atherosclerotic disease, thus increasing the likelihood that they would suffer from subsequent intracranial pathology associated with their TBI.
We did not adjust our data to account for the lag time between admission head CT and bMRI acquisition. While less likely, prolonged lag times increase the likelihood that patients may suffer other events unrelated to the original trauma (e.g., second fall in the hospital, arrhythmias). In these rare instances, the brain MRI may be identifying findings outside of the trauma on admission, thus affecting our study results. A future multi-center, observational study with structured pre-specified data elements and imaging collection times would address many of our studies' limitations.

Conclusion
Patients presenting with acute trauma, more specifically presenting with severe TBI, are more likely to have additional findings on bMRI not visualized on the initial NHCT. Thus earlier imaging with MRI may be beneficial for clinical management, and imaging can be considered within the first week of admission [14].
Our results suggest that bMRI may be an underutilized resource in the acute care of moderate and severe TBI patients.

Conflicts of Interest
None of the authors have any relevant conflicts of interest to disclose.