From airport baggage screeners to assembly-line inspectors, humans tasked with repetitively or continuously scrutinizing separate but similar visuals for hours on end tend to lose their focus bit by bit. The effect is called vigilance decrement. Does it affect mammography readers to a dangerous degree?
U.K. researchers designed a study to find out—and have concluded that interpretation accuracy neither declines as breast-imaging readers go through their day nor suffers when, for experimental purposes, two readers switch up the order in which they read a batch of mammograms during a single session such that each reader is at peak attentiveness at different points in that session.
The report on their double-blind, randomized study is running in JAMA.
Sian Taylor-Phillips, PhD, of the University of Warwick and colleagues analyzed the interpretations of 186 radiologists, 143 radiography advanced practitioners and 31 breast oncologists.
Images were drawn from exams performed at 46 specialized breast screening centers that participated in England’s National Health Service Breast Screening Program between 2012 and 2014.
Readers in the experimental group—made up of the pairs of readers who reversed the order of their readings—interpreted 596,642 mammograms, while those in the control group (pairs of independent readers proceeding as normal) interpreted 597,505 mammograms.
The experimental group detected cancer at a rate of 0.87 percent, nearly a perfect tie with the 0.88 percent observed in the control group.
The recall rates were also close—4.14 percent in the experimental group and 4.17 percent in the control.
Reader disagreement rates were much alike, as well (3.43 percent vs 3.48 percent).
As for performance throughout the workday, continuity was maintained.
Batches read first each day had a cancer detection rate of 0.83 percent in the experimental group and 0.88 percent in the control group, while batches read throughout the rest of the day had a detection rate of 0.85 percent in the experimental group and 0.85 percent in the control group.
In their discussion, Taylor-Phillips et al. note that their trial results are “unexpected and contradict previous research on the vigilance decrement in other fields. The vigilance decrement phenomenon has been reported in many peer reviewed publications but was not observed in this large randomized clinical trial.”
The authors acknowledge several limitations in their study design, the main one being that reading conditions were not controlled. As a result, while they measured effectiveness in screening practice, they could not evaluate efficacy in ideal conditions.
“The experienced specialists in this study could be less prone to a vigilance decrement, as was found in experienced closed caption television operators reviewing a test film,” the authors postulate. “The vigilance decrement phenomenon may be associated with an increase in recall threshold rather than a reduction in performance.”
In accompanying commentary, the University of Wisconsin’s Elizabeth Burnside, MD, MPH, and colleagues write that the Taylor-Phillips study provides “not only a new avenue of investigation but also the opportunity for improving performance in all screening programs, including those in the United States.”
“However, perhaps the most important message conveyed by this study is the potential afforded by combining standardization and advanced information systems to use pragmatic study designs to answer important questions applicable to the usual care setting,” write Burnside and co-authors. “This opportunity promises to effectively narrow the gap between randomized trials and evidence-based interventions so that clinical advances can rapidly influence practice.”