Physicians often take shortcuts in diagnostic reasoning by being selective in the information that they gather and follow-up on. Although necessary, these shortcuts are susceptible to cognitive biases and may cause diagnostic errors. The aim of this study is to examine the occurrence of inappropriate selectivity in the information-gathering and information-processing stages of the diagnostic process and study how it relates to diagnostic errors and patient harm in clinical practice. Expert internists reviewed the patient records of 247 dyspnea patients of five acute-care hospitals in the Netherlands, to detect reasoning faults, diagnostic errors and patient harm. The cases with reasoning faults were discussed with the treating physicians. Based on the record review and the clarifications from the treating physicians, the occurrence of inappropriate selectivity in information-gathering and information-processing was established and related to the occurrence of diagnostic errors and patient harm. Inappropriate selectivity in the diagnostic reasoning process occurred in 45.7% (113 of 247) of the cases. Specifically, selective information-gathering occurred in 33.2% of the cases and selective information-processing in 12.6% of the cases. Diagnostic errors occurred in 18.3% of the cases with selective information-gathering, and in 35.5% of the cases with selective information-processing. Patient harm occurred in 11.0% of the cases with selective information-gathering and in 38.7% of the cases with selective information-processing. The results showed that inappropriate selectivity in the diagnostic process occurred in a substantial number of cases. Particularly inappropriate selective information-processing was related to diagnostic errors and patient harm. Prevention strategies should include an increase in promoting the falsification strategies in the diagnostic process.
Keywords: Cognitive biases; Diagnostic error; Diagnostic reasoning; Dyspnea patients; Patient safety.
Copyright © 2013 Elsevier Ltd. All rights reserved.