Hindsight is 20-20: How preexisting info can impact radiologist interpretation—and malpractice suits

Radiologists expecting to find abnormalities based on existing knowledge of outcomes may show bias in their decision-making and in expert witness testimony in malpractice cases, according to results of a new study published online Feb. 19 in the journal Academic Radiology.

Research has shown that radiologists typically make few mistakes, and when errors do occur, existing safeguards help to quickly and efficiently rectify the problem. But when mistakes persist, patients may suffer negative consequences and radiologists may be subject to medical malpractice lawsuits involving other radiologists as expert witnesses, said lead author Stephen Littlefair, MSc, of the University of Sydney in Australia, and colleagues.

“In certain malpractice cases, the radiologist (defendant) who misses a chest lesion may only have been provided with limited clinical information,” the authors wrote. “However, at trial, the expert witness for the plaintiff is presented with all pertinent medical records and would therefore be cognizant of the location of perhaps a now-known tumor.”

With this scenario in mind, Littlefair and his team devised a test to investigate how the expectation of abnormality, along with prior case knowledge, influences the decision-making of radiologists.

To do so, the researchers performed a web-based perception experiment asking 12 board-certified radiologists to interpret 40 adult chest images, with 20 of those images being abnormal. The radiologists reviewed the images twice and evaluated the presence of pulmonary lesions. A general clinical history was given for all images prior to the first reading, known as the “defendants read.” The radiologists were asked to view the images again two weeks later, unaware that the data had not been altered. The researchers told the radiologists that “these images were reported normal but all of these patients have a lung tumor diagnosed on a subsequent radiograph 6 months later” and gave them the location of the newly diagnosed tumor, referred to as the “expert witness read.”

Not surprisingly, what they found was a significant variance in location-based sensitivity between the two reads, with nodule detection increasing in the expert witness lead along with specificity. No affects were reported regarding case-based sensitivity and case-based specificity.

“The findings of this study suggest that radiologists operating with the advantage of extra outcomes knowledge are significantly biased,” the authors wrote. “There is evidence that such bias, which may cause a radiological expert to challenge retrospectively the decisions of a defendant, is inevitable when said expert knows the outcome.”

Littlefair and his colleagues believe their results merit examination of how defendants and their legal counsel should proceed with regard to informing both judges and juries about the measurable effects of hindsight biases in malpractice cases.

“Such findings should be of particular concern to experts in the legal domain,” they concluded. “Further research into the value of asking an expert to clarify the phenomenon of hindsight bias is suggested.”