5 solutions for addressing errors in diagnostic radiology

Imaging experts are offering tips on how to reduce the number of potential errors in diagnostic radiology.

Omer A. Awan, MD, and colleagues detailed their advice in an Educational Perspective piece published Aug. 17 by Academic Radiology [1]. With increasing demand for imaging and more than 1 billion radiographic exams performed worldwide each year, a 4% error rate would result in about 40 million interpretive inaccuracies per year, they estimated.

Awan and co-authors urged peers to educate themselves about this topic and seek ways to reduce mistakes.

“By taking the time to understand the type of errors that occur and the cognitive process behind image interpretation, we can develop ways for radiologists to slow down, to ‘think about thinking,’ and to remain skeptical yet open-minded throughout the diagnostic process,” Awan, a University of Maryland professor and senior public health contributor for Forbes, and colleagues advised.

In diagnostic radiology, errors can come in two forms: perceptual and interpretive. The former occurs when a radiologist overlooks a sufficiently evident abnormality, often due to fatigue, high case volumes and distractions. Interpretive errors, meanwhile, occur when a radiologist spots the abnormality but misunderstands the clinical significance, often due to cognitive biases.

Awan and co-authors suggested five potential solutions to address these two types of errors:

1) Double reading: “When you consider the number of errors that are attributed to a radiologist getting distracted when reviewing an image or reading a scan too quickly, a logical solution for this perceptual error would be having a second radiologist view the study and confirm the previous findings. However, this can be resource intensive, so it is important to strike a balance between feasibility and higher quality interpretations of images.”

2) Detection with AI assistance: “With the advancements in artificial intelligence, there is a wide scope for using technology to aid the job of a radiologist and reduce medical error. Pattern recognition software, such as computer-aided detection, flag abnormal features on a scan and play a similar role to that of a double reading by a second physician.”

3) Cognitive debiasing: “What has been found to be an effective solution in increasing diagnostic accuracy and improving decision-making is having a heightened awareness of errors in the interpretation process. This phenomenon has been coined by some as cognitive debiasing. Deliberate efforts should be made to identify bias and subsequently focus on switching from pattern recognition to analytical thinking.”

4) Calibrating confidence: “When individuals are unable to accurately judge their own competence level, it can lead to a phenomenon known as the ‘better-than-average’ effect. This can contribute to both perceptual and interpretive errors. Those who overestimate their competence are more susceptible to a self-confirming bias. Feedback is an important means for physicians to reinforce strengths or learn from their mistakes.”

5) Random peer review: “Errors can be present because of a lapse in judgment or due to a limitation in knowledge. Through random peer review of images read by radiologists, such as through the ACR’s RadPeer program, data that can be used to provide valuable feedback and identify points for further education are collected.”

“While each of the methods outlined in this paper can significantly contribute to reducing medical error, it is important to recognize that continued work on effectively integrating each of these strategies into the lives of radiologists is necessary for the highest level of impact,” the authors concluded.