Background Many authors have implicated cognitive biases as a primary cause of diagnostic error. If this is so, then physicians already familiar with common cognitive biases should consistently identify biases present in a clinical workup. The aim of this paper is to determine whether physicians agree on the presence or absence of particular biases in a clinical case workup and how case outcome knowledge affects bias identification.
Methods We conducted a web survey of 37 physicians. Each participant read eight cases and listed which biases were present from a list provided. In half the cases the outcome implied a correct diagnosis; in the other half, it implied an incorrect diagnosis. We compared the number of biases identified when the outcome implied a correct or incorrect primary diagnosis. Additionally, the agreement among participants about presence or absence of specific biases was assessed.
Results When the case outcome implied a correct diagnosis, an average of 1.75 cognitive biases were reported; when incorrect, 3.45 biases (F=71.3, p<0.00001). Individual biases were reported from 73% to 125% more often when an incorrect diagnosis was implied. There was no agreement on presence or absence of individual biases, with κ ranging from 0.000 to 0.044.
Interpretation Individual physicians are unable to agree on the presence or absence of individual cognitive biases. Their judgements are heavily influenced by hindsight bias; when the outcome implies a diagnostic error, twice as many biases are identified. The results present challenges for current error reduction strategies based on identification of cognitive biases.
- Cognitive biases
- Diagnostic errors
- Patient safety
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Correction notice This article has been edited since it first published Online First. Minor edits have been made to κ coefficient values in the Results section and table 4 has been updated.
Twitter Follow Laura Zwaan at @laurazwaan81
Contributors All authors made substantial contributions to the conception or design of the work and to the acquisition and interpretation of data. All authors were involved in either drafting the work or revising it critically for important intellectual content. All gave their final approval for the version to be published. All authors agree to be accountable for all aspects of the work and for ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Ethics approval McMaster Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.