21 - Expert Consensus on Cognitive Biases Affecting Child Physical Abuse Evaluations in Pediatric Emergency Medicine: A Modified Delphi Study

Poster Number: 21
Publication Number: 21.405

Katherine A. Harmon, Children's Hospital Los Angeles, Corona Del Mar, CA, United States; Todd P. Chang, Children's Hospital Los Angeles, Los Angeles, CA, United States; Alan L. Nager, Children's Hospital Los Angeles, Los Angeles, CA, United States; Karen Kay Imagawa, Children's Hospital Los Angeles, Los Angeles, CA, United States; Anita R. Schmidt, Children's Hospital Los Angeles, Los Angeles, CA, United States; Phung K. Pham, Children's Hospital Los Angeles, Los Angeles, CA, United States

Background: Cognitive biases are known to increase diagnostic error. Child abuse can result in significant morbidity and mortality and is often missed. However, little is known about which cognitive biases influence diagnostic errors in the evaluation of child abuse in Pediatric Emergency Medicine (PEM).

Objective: A modified Delphi study was conducted to obtain consensus on relevant cognitive errors in child physical abuse evaluations in PEM.

Design/Methods: We designed a modified Delphi study in which 15 nationally recognized experts in PEM, Child Abuse Pediatrics, and/or cognitive bias in medicine were purposefully sampled and recruited to participate. The study was conducted online via Qualtrics, with an a priori goal of 7 to 9 cognitive errors derived by consensus. For Round 1, we compiled 32 cognitive errors and their definitions from the literature and created a PEM-specific vignette for each one. Using a 9-point Likert scale, participants rated the importance (1= not important, 9= extremely important) and frequency (1=never, 9=always) of each cognitive error in child physical abuse evaluations. Consensus was defined as the top 2/3 of the cognitive errors in rank order based on their median importance and mode (measure of central tendency). Median frequency scores were used as a tiebreaker between cognitive biases, if required. Each of the three survey rounds yielded a truncated list of cognitive errors.

Results: 13/15 (86.6%) experts completed all 3 rounds of the modified Delphi process (Table 1), agreeing on 9 cognitive errors as the most important and frequent in child physical abuse evaluations: anchoring bias, ascertainment bias, confirmation bias, framing effect, fundamental attribution error, outcome bias, premature closure, unpacking principle, and visceral bias. Each cognitive bias received a final median score of 7 for importance.Conclusion(s): The most relevant cognitive errors in child physical abuse evaluations in PEM should guide medical education to reduce diagnostic errors in the evaluation of child physical abuse. Consensus on relevant cognitive errors may inform other educational approaches, such as simulation, to better determine if the most important cognitive errors identified occur more frequently in cases of child abuse vs other medical diagnoses.
Katherine Harmon CVKatherine Harmon CV 2022 PDF.pdf