CHICAGO - An error-tracking system developed at Chicago's Rush University Medical Center can quantitatively measure radiology residents' performance and serve as an educational tool, according to its Cicero, IL-based creators. Researchers in Taiwan also focused on residents' performance, but used teaching case files to help reduce misreads.
At Rush, residents read all emergency department films at night, which are then reviewed in the morning by an attending radiologists, explained Dr. Bryon Natashi and Dr. Laurens Ackerman, Ph.D., in their poster presentation at the 2006 RSNA meeting.
The duo built a data collection program, based on their reading protocol, using PerlTK and Linux. In this pilot system, information from the attending's morning review was transferred to a MySQL database and stripped of all patient data. The program automatically generates a form that the attending fills out to quantify resident error rates.
Natashi and Ackerman used data from three residency classes over a 16-month period. In that time, 396 discrepancies were noted between the residents' first reads and the attendings' final reads. The majority (243) of those misreads were done by first-year residents. Third-year residents had the least number of discrepancies (26).
The majority of errors were classified as "missed item" followed by "overcall." The second most common were missed fractures, they stated.
This error-tracking system not only serves as an education tool but also gives a department the chance to compile error statistics, they said. These statistics can be individualized for each resident to facilitate discussion, they added.
In another study presented at the meeting, researchers in Taiwan looked at the usefulness of teaching files containing missed extremity fractures. Dr. Wan-Chen Tsai and colleagues at the Taipei Veterans' Hospital and the National Yang-Ming University selected 30 cases from 108 patients whose fractures were not identified in the initial imaging reports. These missed fractures included lower and upper extremity cases.
Eleven residents read two sets of x-ray images on two difference occasions, one month apart. They also took a one-hour training session on occult fractures between the two testing sessions. Some of the cases that were presented included radial head fractures, scaphoid fractures, and left femoral neck fractures.
Not surprisingly, all residents improved their results on the second test, according to the group's RSNA poster. The mean correct fracture identification rate for first-year residents jumped from 5% to 44%. For second-year residents, the correct rate increased from 44% to 67%, and the rate went from 60% to 78% for third-year residents. Fourth-year residents turned in a mean correct rate of 80%, up from 67%. Overall, the mean false-positive rate for these residents dropped from 16% to 14%.
By Shalmali Pal
AuntMinnie.com staff writer
November 30, 2006
Related Reading
Revisiting report errors serves as educational tool for residents, December 22, 2005
Automated report tracking serves clinical, educational goals, January 18, 2005
Rads urged to standardize reporting of vertebral fractures, November 23, 2004
Copyright © 2006 AuntMinnie.com
