JACR: Diagnostic accuracy drops at end of day

The ability of radiologists to detect fractures and maintain focus was reduced at the end of the day compared to early in the day, and they showed increased symptoms of fatigue and eyestrain, according to a study in this month's Journal of American College of Radiology.

As radiologists read more studies with more images in less time, they report symptoms of eyestrain, including blurred vision and difficulty focusing, which could affect diagnostic accuracy, shared lead author Elizabeth A. Krupinski, PhD, from the University of Arizona in Tucson. Krupinski and colleagues designed the study to evaluate diagnostic accuracy at the beginning and end of a day of diagnostic image interpretation.

Researchers recruited 20 attending radiologists and 20 radiology residents from the University of Arizona and University of Iowa, Iowa City, to read 60 test skeletal radiographic exams with two to four images in each. Thirty cases had no fractures, and 30 presented with one moderate to very subtle fracture of the wrist, hand, ankle, foot, long bones, shoulder or rib. Radiologists rated positive findings as definite, probable, possible or suspicious and assigned a confidence rating to each, wrote Krupinski.

Data were collected in the morning and late afternoon for each reader, and the radiologists completed the Swedish Occupational Fatigue Inventory (SOFI), which measures fatigue in the work environment. Researchers used a refkeratometer to measure visual accommodation, or strain.

Krupinski and colleagues measured accuracy for fracture detection via the area under the receiver operating characteristic curve and compared average areas using analysis of variance (ANOVA).

“There was a significant drop in detection accuracy for late versus early reading. The average areas under the curves were 0.885 for early and 0.852 for late reading,” wrote Krupinski.  Residents were somewhat less accurate than attending radiologists on hard cases, and the reduction in accuracy for late reading was evenly split among increased false-positives and decreased true-positives.

Visual strain analysis showed significantly greater accommodative error after the workday, which suggests that readers were more myopic and experienced more visual strain after the workday, according to researchers.

When Krupinski and colleagues analyzed five SOFI factors with an ANOVA by early and late session, they noted statistically significant differences in lack of energy, physical discomfort and sleepiness but not in physical exertion or motivation.

Reading time remained consistent for fracture detection in both early and late sessions for both easier and harder cases. However, the average reading time of 56.7 seconds for normal studies was significantly greater than studies with fractures at 46.9 seconds.

“The results of this study suggest that because of increased visual strain as reflected in their lowered accommodation measures, radiologists’ ability to focus on images was reduced, making them less accurate after a day reading diagnostic images,” concluded Krupinski, who added that the difference in accuracy of 4 percent was small. “Radiologists need to be aware of the effects of fatigue on diagnostic accuracy and take steps to mitigate these effects,” she stated.

The study was limited to radiographic exams, and CT and MRI exams may be more fatiguing than static images, cautioned Krupinski.