This prospective Institutional Review Board-approved study was conducted at a university hospital Level I Trauma Center with an annual ED census of 38,000 patients. All patients who received an emergency ultrasound by ED personnel were eligible. There were no exclusions.
From January 24, 2005 to August 15, 2005, ultrasound students were assigned to specific days of the week, with all days equally represented. The students worked from 7 am to 2 am. EM residents and faculty were not aware that the number of scans was being monitored, nor that students purposely were not scheduled on some days.
We recorded the number of patients scanned, the number of scans done, and if medical students were present that day. Some patients received more than one scan, for example, right upper quadrant and the kidney. We also recorded if a resident or faculty member participated in the scan. We defined participation as presence at the bedside during the scan, regardless of who held the probe. Of note, each trauma patient (approximately 7–8 per day) receives a Focused Assessment of Sonography for Trauma (FAST) regardless of medical student presence. At our institution, it is common to have several physicians/students at the bedside assisting in image acquisition and interpretation.
All scans were done with either a BK Medical Hawk 2102 or Sonosite TITAN machine. Each scan was recorded and reviewed in its entirety during weekly educational sessions in the division of emergency ultrasound led by the emergency ultrasound fellow and the fellowship director.
Research assistants added the total from each day and data were entered into Excel (Microsoft Corporation, Redmond, WA) and analyzed with Stata (version 9.2, StataCorp, College Station, TX). Because we did not expect the number of patients and scans to be normally distributed, we used nonparametric statistics. We compared the number of patients that received scans and the number of scans with and without a medical student present. The distributions of scans and patients with and without a medical student were compared using the Wilcoxon rank sum test. We calculated medians and 95% exact binomial confidence limits for scans and patients with and without a medical student present.