After-hours imaging interpretations by trainees a potential source of confirmation bias in radiology

Mistakes are more common among after-hours imaging interpretations made by a trainee and then confirmed by a consultant radiologist, according to a new study published in Clinical Radiology. Could confirmation bias be to blame?

“Given the frequency of errors, a review of discrepancies within a standardized case review framework is a fundamental part of effective clinical governance and a necessary component of ongoing professional validation,” wrote A. Nanapragasam, with the department of radiology at Newcastle University in the U.K. “Discrepant cases should be reviewed anonymously and discussed at a regular departmental meeting, such that everyone can benefit from the learning points generated.”

The researchers explored data from 100 consecutive cases reviewed by a discrepancy forum at a tertiary neuroradiology provider. Overall, after-hours imaging—when a trainee writes the radiology report and their work is confirmed by a consultant—made up 36 percent of the problematic reports, though they only made up 18 percent of the overall number of reports. Fifty-two percent of the errors were from consultant-checked reports first created by a trainee.

In addition, radiology reports created by a trainee outside of normal business hours and then confirmed by a consultant are “over-represented” among discrepant reports “by a factor of twofold over the expected rate.”

“The principal finding of the present study is the association between a trainee report and subsequent consultant error,” the authors wrote. “Although trainee reports constitute only 18 percent of reporting activity, they make up 52 percent of erroneous reports. This provides a salient reminder of the biasing effect that a trainee report can have upon the subsequent consultant scan checking.”

Nanapragasam et al. noted that confirmation bias is likely responsible for these statistics. The consultant radiologists could be letting the trainee’s opinion influence their own decisions, leading to mistakes that would not have occurred otherwise.

So how can radiologists battle such bias without slowing down how they confirm reports? One of the biggest steps, it turns out, is just acknowledging that the bias exists.

“Simply raising awareness of the underappreciated error-associated reporting context that is trainee report checking may be sufficient to somewhat reduce the incidence of error while preserving reporting speed,” the authors wrote. “By being cognizant of the way in which we make decisions and of our susceptibility to cognitive biases, we give ourselves a mental cue to look out for those unrelated but important findings.”