Cognitive bias in diagnostic radiology
Citation, DOI, disclosures and article data
At the time the article was created Andrew Murphy had no recorded disclosures.View Andrew Murphy's current disclosures
At the time the article was last revised Andrew Dixon had the following disclosures:
- Annalise AI, Sponsored discussion at Radiopaedia 2021 (past)
These were assessed during peer review and were determined to not be relevant to the changes that were made.View Andrew Dixon's current disclosures
Cognitive biases have a complex and significant impact on the perception of examinations within diagnostic radiology, with the clear and present danger of diagnostic errors. The following are some of the more common cognitive biases that can affect day-to-day decision making 1.
Alliterative bias (sometimes called satisfaction of report) is the tendency for your diagnostic impression to be unduly influenced by the prior interpretation made by another radiologist or clinician. It is a type of confirmation bias and it can result in the same incorrect impression being propagated from one radiologist to the next. Formulating your own interpretation before reviewing any prior imaging reports may help reduce alliterative bias.
Anchoring bias is the tendency for your initial impression to unduly influence your evaluation of subsequently collected information. Careful assessment of all imaging findings prior to formulating a diagnosis and consideration of alternate diagnoses may help minimize anchoring bias.
Automation bias is the tendency for reporters using computer-aided decision support to over rely on the software for the diagnosis, and to ignore their own opinions 2. Making your own assessment prior to reviewing the computer-aided findings, and knowing the limitations of the automated software, can help reduce this bias.
Availability bias refers to the tendency for your diagnostic judgements to be unduly influenced by information or experiences that are readily recalled in your mind. Use of information sources beyond your own personal experience, such as publications or an opinion from colleagues, may help to minimize availability bias.
The bandwagon effect (sometimes termed diagnosis momentum) refers to the tendency for you to do or think as others do, simply because that's what others are doing. Applying your own judgment and not dismissing your own opinion can help avoid this bias.
Having a predetermined diagnosis in mind, then looking for evidence that supports this predetermined idea. Alliterative errors 3, sometimes referred to as satisfaction of report errors, are caused by the tendency to overvalue previous reports, and can be conceptualised as a type of confirmation bias.
Framing bias refers to the tendency to be unduly influenced by how a clinical question is asked or how the problem is presented. For example, a request form that presents a succinct history that perfectly matches a particular pathology, may influence your interpretation of the imaging. Reviewing the images prior to reading the clinical indication can help avoid any influence. Obtaining more detailed clinical information via the electronic medical record may help provide you with a more balanced assessment of the clinical situation.
Hindsight bias is the tendency to overestimate the prior predictability of an event (eg. a diagnosis) after the event is known. In other words, the difficulty of making the correct diagnostic decision initially is retrospectively de-emphasized, after the diagnosis has been proven. It is also informally termed as the “I knew it all along” or "retrospectoscope" bias 5. It is important to be conscious of this bias so that you are not overly critical of yourself or others when quite reasonable errors are made.
A tendency to favor a less severe diagnosis based on empathy for a patient.
Making a judgment on an aspect of an image that is based on one's own perception of what that represents. Representativeness bias as the description suggest can also be heuristic in nature.
Search satisfaction is the tendency to prematurely stop searching after early findings satisfy the reader that the diagnosis or symptoms can be explained. Satisfaction of search (SOS) errors have been reported to account for 22% of diagnostic errors 4. A systematic approach to image interpretation and review of check areas and not-to-miss diagnoses can help to reduce this bias.
Zebra retreat bias
A reader will not make a rare diagnosis, which is otherwise supported by the available evidence due to a lack of confidence.
- 1. Bruno MA, Walker EA, Abujudeh HH. Understanding and Confronting Our Mistakes: The Epidemiology of Error in Radiology and Strategies for Error Reduction. (2015) Radiographics : a review publication of the Radiological Society of North America, Inc. 35 (6): 1668-76. doi:10.1148/rg.2015150023 - Pubmed
- 2. Goddard K, Roudsari A, Wyatt JC. Automation bias: a systematic review of frequency, effect mediators, and mitigators. (2012) Journal of the American Medical Informatics Association : JAMIA. 19 (1): 121-7. doi:10.1136/amiajnl-2011-000089 - Pubmed
- 3. Alliterative Errors. (2012) American Journal of Roentgenology. 174 (4): 925-31. doi:10.2214/ajr.174.4.1740925 - Pubmed
- 4. Young W. Kim, Liem T. Mansfield. Fool Me Twice: Delayed Diagnoses in Radiology With Emphasis on Perpetuated Errors. (2014) American Journal of Roentgenology. 202 (3): 465-70. doi:10.2214/AJR.13.11493 - Pubmed
- 5. Lindsay P. Busby, Jesse L. Courtier, Christine M. Glastonbury. Bias in Radiology: The How and Why of Misses and Misinterpretations. (2017) RadioGraphics. 38 (1): 236-247. doi:10.1148/rg.2018170107 - Pubmed
- 6. Itri J & Patel S. Heuristics and Cognitive Error in Medical Imaging. AJR Am J Roentgenol. 2018;210(5):1097-105. doi:10.2214/ajr.17.18907