The accuracy of interpreting a breast cancer screening mammogram is not diminished as clinicians get toward the end of a session.
The accuracy of interpreting a breast cancer screening mammogram is not diminished as clinicians get toward the end of a session, and there were no differences in accuracy between the first session of the day compared with later sessions, according to the results of a UK-based study published in JAMA.
Analyzing screening mammograms is a repetitive and potentially difficult task that requires great attention to detail. Errors can result in a missed cancer and false positive recalls. Because similar repetitive visual search tasks in other settings such as airport baggage screening and laboratory work can result in diminished accuracy over time, Sian Taylor-Phillips, PhD, of the University of Warwick in the United Kingdom, and colleagues assessed whether radiologists also experience diminished accuracy when faced with many mammograms to evaluate in a single session.
Generally, two independent radiologists analyze each mammogram for cancer. In the current study, the researchers reversed the order in which two radiologists read a batch of mammograms so that each reader experienced his or her peak attentiveness at different points in the batch of mammograms.
“The intervention did not influence cancer detection rate, recall rate, or rate of disagreement between readers,” wrote the authors.
There was no significant difference in the rate of cancer detection between the standard method of mammogram reading (control group) and the experimental method. The cancer detection rate was 0.87% in the experimental group and 0.88% in the control group. The recall rates were 4.14% and 4.17% percent, in the experimental and control groups, respectively. Rate of reader disagreement also did not differ significantly (3.43% vs 3.48%).
“There was no pattern of decreasing cancer detection rate with time on task as predicted by previous research on vigilance decrements as a psychological phenomenon,” wrote the authors.
The rate of detecting breast cancer did not change with time spent on task, as represented by near identical odds of detecting cancer between the first and 40th case in a batch (odds ratio, 0.987; 95% CI, 0.929–1.048).
Rate of cancer detection did not change throughout the day either. For batches read first each day, the cancer detection rate was 0.83% (95% CI, 0.76%–0.89%) in the experimental group and 0.88% (95% CI, 0.81%–0.95%) in the control group. For batches read throughout the rest of the day, the cancer detection rate was 0.85% (95% CI, 0.82%–0.89%) in the experimental group and 0.85% (95% CI, 0.82%–0.88%) in the control group.
The study was double blind and randomized, conducted at 46 specialized breast screening centers from the National Health Service Breast Screening Program in England between 2012 and 2014. A total of 360 qualified readers-186 radiologists, 143 radiography advanced practitioners, and 31 breast cancer clinicians participated. These readers interpreted a total of 596,642 mammograms in the experimental group and 597,505 in the control group.
“This study combined the strengths of randomization with standardized, real-world practices in place throughout England,” wrote Elizabeth S. Burnside, MD, MPH, MS, of the University of Wisconsin School of Medicine and Public Health in Madison, Wisconsin, and colleagues in an accompanying editorial.
The editorial authors also highlighted differences in the mammogram analysis system between the United States and the United Kingdom: “In England, population-based screening offered with lower frequency to women in narrower age ranges results in fewer false positives but detects fewer cancers (ductal carcinoma in situ and small invasive cancers) as compared with the United States, which provides opportunistic screening at higher frequency over wider age ranges.”