The total-body CT (typically referred to as the pan-scan) has quickly become a key component in the initial evaluation of trauma patients presenting to the Emergency Department. Despite this, the evidence supporting the benefits of the trauma pan-scan is lacking. Its rise to prominence due more so to riotous indignation than true evidentiary support. This post will focus on the recent publication of the REACT-2 trial, the first high quality RCT to examine both the benefits and harms of the total-body CT imaging strategy (1).
Published online in the Lancet on June 28th 2016, Sierink et al randomized 1,403 patients to either selective scanning as per their pre-planned protocol or a total-body CT at presentation. The authors attempted to ensure only high acuity trauma patients enrolled in their cohort by utilizing a strict inclusion criteria (RR > 30 or < 10 , pulse >120, SBP< 100, GCS<14, estimated blood loss of > 499 mL, abnormal pupils, or high risk mechanism). Even with this narrow criteria, the authors excluded 203 patients after randomization, the majority of which were omitted due to experiencing only minor trauma.
Of the 1,083 patients included in the the final analysis, the authors found no difference in their primary endpoint, in-hospital mortality (16% in the selective scanning group and 16% in the total-body CT group). Nor did they find a difference in 24-hr mortality (6% and 8% respectively), or 30-day mortality (16% and 17% respectively). Sierink et al failed to find a benefit for whole-body CT strategy in either the subset of patients defined as polytrauma (ISS>16) or TBI (GCS of <9 and AIS head score of >2). Patients with polytrauma had an in-hospital mortality of 25% and 22% in the selective imaging and total-body CT groups respectively (p-value of 0.46). While those with TBI demonstrated an in-hospital mortality of 44% and 38% respectively (p-value of 0.31). These findings remained consistent when the authors examined the 30-day mortality. Patients in the polytrauma subset experienced a 30-day mortality of 24% and 23% respectively (p-value of 0.46), while those with TBI demonstrated a 30-day mortality of 41% and 39% respectively (p-value of 0.31).
The authors established that patients randomized to the selective imaging strategy received far fewer radiographic investigations. Only 46% of the patients in the selective imaging group underwent total-body CT imaging. This reduction did not result in any clinically important delays in identification of life threatening diagnoses (58 min vs 50 min).
Despite these negative findings, there are a number of nuances that supporters of the total-body irradiation strategy may take issue with. The first is that 46% of the patients in the selective imaging group ended up undergoing total-body imaging. The argument is that this crossover would dilute the results, obscuring any benefit the total-body CT strategy may produce. On the other hand 54% of the patients in the selective imaging group avoided total-body irradiation. This is not a weakness of the trial design, but rather a reality. No one is arguing that total-body radiographic evaluation should be eliminated, rather that the decision be based upon clinical judgment rather than reflexive impulse induced by the fear of missing clinically occult injuries. This is a pragmatic representation of reality and only strengthens the findings demonstrated in REACT-2.
The second and far more important point of contention is that patients randomized to the total-body CT group were sicker, as represented by the portion of patients determined to have polytrauma (61% vs 67%) at presentation. This initial imbalance may obscure the benefits produced by a total-body CT strategy. And while this position when taken at face-value seems reasonable, it is a mirage. Rather this imbalance is not the random error of sampling but the expected inflation of injury severity scores due to the total-body imaging strategy imposed on these patients.
The more radiographic studies a patient receives the more injuries they will be burdened with. And while these excessive diagnoses have little clinical meaning, they are exceptionally facile at augmenting the injury severity score (ISS) (2). The ISS assigns a numeric score (0-6) for each anatomic area. This score is then squared and the squared numbers are summed to give the resulting ISS. ISS ranges from 0-72. Since it is solely an anatomically defined score, patients who undergo total-body imaging are far more likely to have clinically insignificant injuries detected by these unnecessary scans (3). Given the exponential nature of the ISS, findings can artificially inflate the resulting score. This shifts a healthier cohort into a more severe ISS level creating the illusion of more severely injured cohort. This phenomenon is clearly demonstrated in the REACT-2 cohort. A larger portion of patients randomized to the total-body CT group were considered polytrauma (67% vs 61). As illustrated by fig 2 in the Lancet publication, this inflation was more prominent in patients with intermediate scores (ISS of 16-49)(1), as severely injured patients are likely to have enough serious injuries that incidental findings found on CT will have little influence on the already high ISS (2). 362 patients in the total-body arm and 331 in the selective imaging arm were considered polytrauma. And yet the absolute number of patients who died in each group was essentially identical (81 and 82 patients), suggesting the excess patients observed in the total-body CT group were not representative of a sicker cohort but rather the same cohort subjected to the artificial inflation of clinically insignificant radiographic diagnoses.
This is not the first trial to demonstrate that selective imaging is safe and feasible in a trauma population. As far back as 2000 Hoffman et al published the NEXUS criteria, demonstrating that select trauma patients could forego radiographic evaluation of the cervical spine if they met specific clinical criteria (4). Since then other decision aids have been published demonstrating similar findings for the head, chest and abdomen (5,6,7,8,9). The REACT-2 cohort is essentially a validation of such a strategy. In the selective imaging group, the authors used very similar criteria to the ones found in these decision aids to determine who underwent imaging . Patients underwent imaging of the cervical spine only if they were unable to be cleared clinically by the NEXUS criteria (1).
The methodological design of REACT-2 failed to address the effect total-body imaging strategies have on the Emergency Department as a whole. We have all seen the sick medical patients who truly require CT imaging delayed because of clinically stable trauma patients requiring their mandatory pan-scan. And while such effects are difficult to quantify, an inability to measure its presence does not eliminate its existence.
There is no doubt that a total-body imaging strategy is more convenient for the treating clinicians. Less time has to be spent at the bedside assessing and re-assessing the patients, as most injuries are likely to be identified on CT. Patient satisfaction is generally higher as clinicians who perform more tests are often perceived to be more thorough. The clinician is spared the often long and contentious discourse on the harms of excessive imaging and over-diagnosis. On the rare occasion that an injury is missed, blame cannot be placed for obtaining inadequate radiographic investigations. Furthermore, the ISS inflation caused by the total-body imaging strategy creates the illusion of better outcomes when trauma centers are compared to the national standard. And yet exposing patients to dangerous amounts of radiation for what amounts to our convenience is unjustifiable. The prior evidence supporting the use of the trauma pan-scan was at best exploratory supported only by blustery claims of hearsay and what ifs. With the publication of REACT-2 it is hard to imagine the current practice of total-body irradiation an ethically sustainable strategy.