CHICAGO - Less experienced residents have a tendency to underestimate abnormalities in computed radiography (CR) exams, particularly those of the thoracic area, according to investigators from the University of Maryland in Baltimore.
"Several studies have been performed looking at resident performance and evaluating various imaging exams and their preliminary findings.... Our purpose was to evaluate the performance of radiology residents in interpreting computed radiographic exams using a PACS system," said Dr. John Filigenzi during a talk this week at the RSNA meeting. "And if any information gleaned from this paper could be used for improvements in resident education," he added. Filigenzi is a third-year resident at the institution. His co-author is Dr. George Gross.
For this study, 6,185 CR exams for children, ages one day to 18 years, were interpreted by diagnostic radiology residents. More than 75% of the residents were in their first and second year because, at his hospital, these studies come from the emergency department during on-call hours, Filigenzi said. They all had between six and 48 months of residency training and one to 12 weeks experience in pediatric radiology.
All exams were reviewed on an IMPAX 3.5 PACS system (Agfa HealthCare, Greenville, SC), where the reading also was recorded. Resident interpretation was then compared to that of a pediatric radiologist (considered the gold standard in this scenario), and scored as follows: agreement, false-positive (FP), overcall of severity, false-negative (FN), and undercall of severity.
All disagreements were further divided into minor types (little or no clinical impact) and major types (altered clinical management), Filigenzi said.
There were 2,595 muculoskeletal exams, 2,484 thoracic exams, 501 of the abdomen, 383 of the spine, and 222 skull and neck CRs. The mean age of the patients was 8.5 years. Agreement between the resident read and the pediatric radiologist read occurred in 82.7% percent of the exams. Disagreement arose in 17.3% of the cases, of which 6.9% were considered major.
"Specificity remained constant over the four years of residency, in the low to mid 90%," he said.
The level of agreement increased from the first year to the second year of residency (81.4% to 83.8%), then remained stable, Filigenzi reported. Major disagreements decline from 7.8% to 6.4% from the first year to the second year before leveling off, he added.
There was 92% agreement in the spine, which dropped to 7% in the thorax. Similarly, major disagreement occurred in 1.2% cases in the spine and 10% in the thorax. Based on his personal experience, Filigenzi said he initially found chest films for younger patients particularly difficult to understand.
Finally, as the patients grew older, so did agreement between the residents and the seasoned reader -- 69% for ages one to three months, increasing to as much as 89.1% in kids age 13 and older.
Overall, false-negative reads and undercalls exceeded false-positive reads and overcalls by a 2:1 ration, he said. "We saw that there was an improvement in sensitivity from first- to second-year residents, with little change after that," Filigenzi concluded. "There seemed to be an increase in specificity over the four years of residency, with the exception of thoracic imaging. Residents are more likely to miss an abnormality than to call a normal study, abnormal."
Session moderator Dr. A. Gregory Sorensen asked if thoracic imaging proved to be a sticking point because of long gaps between exposure to pediatrics.
"I think part of it is that. From my personal experience, you have to get into a mode of reading pediatric films versus adult films," Filigenzi replied. In addition, third- and fourth-year residents spend less time on-call so they may not be exposed to as many pediatric chest x-rays.
Further areas of study would include the reasons for the sensitivity plateau between the second-, third-, and fourth-year residents; why chest imaging proved to be so troublesome; and a more detailed explanation of the relationship between improved resident reading and patient age, Filigenzi said.
By Shalmali Pal
AuntMinnie.com staff writer
December 2, 2004
Related Reading
Reducing interns' work hours decreases serious medical errors, October 28, 2004
Experience beats speed in emergency FAST training, June 2, 2004
Copyright © 2004 AuntMinnie.com
