Written by Christian Gerhart
Spoon Feed
In this prospective observational study of patients undergoing CT imaging of the abdomen/pelvis for blunt trauma, gestalt among Emergency Medicine residents/attendings and General Surgery residents was moderately accurate but not sufficiently sensitive to rule out clinically significant injuries when the acceptable miss rate was set at 2%.
Gestalt is pretty good… but not good enough.
Deciding which trauma patients need CT imaging is challenging. We want to avoid unnecessary radiation while not missing injuries. There are excellent decision rules (NEXUS, Canadian Head/C spine) to help us decide if trauma patients need imaging of the head, neck, and chest. However, decision rules for imaging of the abdomen/pelvis are lacking. These authors set out to determine how effective physician gestalt was in estimating the likelihood of a clinically significant injury (CSI).
This was a prospective observational study across four trauma centers (3 academic, 1 community) which included patients ≥ age 15, who were being evaluated after blunt trauma, and had an abdomen/pelvis CT ordered. The study team collected data on how likely the physicians (Emergency Medicine residents and attendings plus General Surgery residents but NOT General Surgery attendings) thought it was the patient would have a CSI after an abdomen/pelvis CT was ordered but before it was performed. Response options included the following: <2%, 2%–10%, 11%– 20%, 21%– 40%, and >40%. An acceptable missed injury rate of <2% was chosen based on previous literature and expert panel opinion. The authors did perform a verification bias assessment with 3-month phone and EHR follow up on a group of similar patients who did not have a CT, which did not demonstrate any CSI.
Overall, physician gestalt was moderately accurate, with an area under receiver operating curve (AUROC) of 0.7. Gestalt was very effective in the midranges of suspected injury probability, but performed poorly at the high and low ranges. Importantly, among the patients physicians estimated as having a <2% likelihood of CSI, there was a 5.6% (95% CI, 3.3%– 9.3%) rate of clinically significant injury, which translated to sensitivity of 95.2% (95%CI 91.7%– 97.3%) for this group.
How will this change my practice?
This is a well-done study with sound methodology that provides evidence that physician gestalt is probably not sufficient on its own to accurately decide who should get a CT for the abdomen/pelvis for blunt trauma. Importantly, only patients who were already getting a CT were included. Including patients whom physicians did not think needed a CT may have changed the accuracy, especially at the lower end of probability range. I agree with the authors that this study alone should not cause us to order more or fewer CTs but should prompt a focus on developing a high sensitivity decision rule to aid us in taking care of these patients.
Editor’s note: We performed a similar study in the past and found sensitivity of attending emergency physician gestalt for abdomen/pelvis was 96.9% (95%CI 94.2-98.4). ~Clay Smith
Source
Accuracy of physician gestalt in prediction of significant abdominal and pelvic injury in adult blunt trauma patients [published online ahead of print, 2023 Jun 26]. Acad Emerg Med. 2023;10.1111/acem.14768. doi:10.1111/acem.14768.
