Page 78 - JSOM Summer 2023

34.25 seconds (geometric standard deviation [GSD] = 2.2), with the American College of Emergency Medicine’s policy state-
no significant difference between Zone 1 (36.0 sec [GSD = 2.2]) ment and therefore is taught during residency under ACGME
and Zone 3 (32.6 sec [GSD = 2.3]). Correct answers were made guidelines. Second, from a technical standpoint, Zone 1 sono-
significantly faster (32.3 sec [GSD = 2.2]; n = 253) than were graphic windows are more challenging to obtain because of
incorrect answers (60.0 sec [GSD = 1.8]; n = 27) (Figure 5). the acoustical barriers presented by the nonconductive nature
of the pulmonary airspaces and bony rib cage. Lower speci-
24
FIGURE 5 The time it took for physicians to determine the accuracy ficity exhibited in Zone 3 may be secondary to the substantial
of REBOA placement by ultrasound, recorded in seconds*. amount of bowel gas seen in cadaver models, which gives off
artifact that may have led to FPs.

The cadavers used had a wide range of anatomic variants in
their physical characteristics, sexes, and vasculature, which
increases generalizability. Furthermore, the high rate of in-
tra-rater reliability supports the pressurized cadaver model.
The mid-range inter-rater reliability may be the result of the
wide range of experience levels among the 10 participants be-
cause they had an over 20-year range in experience. In ad-
dition, each participant had varying familiarity with REBOA
prior to the study, which may have been a contributing factor.
Physicians that performed poorly, regardless of career level,
may have benefited from additional training or practice with
REBOA localization. Additional training would also improve
confidence in localization. We found that correct answers
occurred twice as fast as incorrect answers. It’s possible that
providers more confident in their answers were able to an-
swer faster, but our study did not survey confidence level and
included too few providers to test this hypothesis.
*Of 280 placements, 253 were correct answers, and 27 were incorrect
answers. Limitations
Circles represent individual data points, the long middle bar represents
the geometric mean, and the error bars represent the 95% confidence REBOA placement confirmation has a limited evidentiary base
interval. because of its current narrow clinical indications. Additionally,
****Unpaired t-test; p < .0001 appropriate and realistic models are limited. Here, we chose a
pressurized cadaver model but encountered some challenges
Inter- and Intra-rater Reliability and limitations. It was difficult to maintain prolonged view-
Each US-trained physician performed localizations on two able US windows of the abdomen. The model required fre-
or three unique cadavers. To determine inter-rater reliability, quent fluid administration to maintain arterial and venous
the overall accuracy was pooled from each physician’s total stenting. If the model became overflooded with fluid, the sub-
localization attempts and compared with those of all other sequent edema and ascites made US views much more chal-
participating physicians. The resultant Light’s kappa was lenging to obtain. Carefully limiting the amount of hypertonic
0.45, representing fair or moderate agreement between raters. saline administered ultimately ensured the greatest longevity
Intra-rater reliability was calculated to determine the consis- of the cadavers throughout the study.
tency of data generated by each physician among the two or
three cadavers tested. Light’s kappa for each physician ranged Additionally, because the model was pressurized but not per-
from 0.75 to 1.0, which corresponds to good agreement (0.60 fused in a pulsatile manner, there was a lack of dynamic feed-
to 0.80) and very good agreement (0.80 to 1.0) within raters. back. Pulsatile flow would allow use of Doppler and color flow
on the US and has been shown to be an effective model for
25
REBOA training. A participant could use these additional US
Discussion tools to identify the lack of Doppler or color flow distal to an
In this study, we demonstrated the accuracy and reliability inflated aortic balloon, which may allow for improved accuracy
of EM physicians using US to confirm REBOA placement in when determining placement. This limitation may have led to an
Zones 1 and 3 of the aorta. With overall sensitivities and spec- underestimation of catheter placement accuracy, sensitivity, and
ificities approximately 80% in both Zones 1 and 3, and an specificity. Finally, the REBOA was already placed and inflated
overall accuracy of 80%, US could be a powerful tool in aus- prior to the US being used. Ultrasound allows for real-time pro-
tere or resource-limited environments when used by trained cedural feedback, and the inability to move the catheter may
ultrasonographers and where gold standard confirmation with have limited the success of the participants. Future work should
fluoroscopy is not available. investigate confirmation of REBOA placement using additional
US tools in both static and dynamic placement models.
The higher sensitivity found in Zone 3 compared with Zone 1
may be attributed to two primary factors. First, while EM phy- Conclusion
sicians’ scope of practice includes indications for both thoracic
and abdominopelvic US, most have notably more exposure to Based on the adequate, although not optimal, accuracy and
the latter. Ultrasound evaluation of the abdominal aorta is one high intra-rater reliabilities measured in this study, US localiza-
of the 12 core emergency US applications specifically listed in tion of REBOA by trained EM physicians may be an effective

76 | JSOM Volume 23, Edition 2 / Summer 2023