Approaching the Limits of Knowledge: The Influence of Priming on Error Detection in Simulated Clinical Rounds

Author: Elie Razzouk, MD

Primary Advisor: Trevor Cohen, MBChB, PhD (co-author)

Committee Members: Khalid Almoosa, MD (co-author); Vimla Patel, PhD, DSc (co-author)

Masters thesis, The University of Texas School of Biomedical Informatics at Houston.


Errors are inevitable in all clinical settings, posing substantial risk to patients. Studies have shown detection and correction are essential to error management. This paper documents the use of Opensimulator, a virtual world development platform, to create a virtual Intensive Care Unit where error recovery can be studied in a controlled, yet realistic environment. Subjects participated in rounds presented by computer-generated characters. Errors were embedded in these presentations, and subjects were evaluated for their ability to detect them. Eight subjects were asked to evaluate two cases and answer related knowledge-based questions under two conditions: primed (forewarned of the presence of errors) and un-primed. Subjects frequently failed to detect errors despite having the prerequisite knowledge. Priming significantly improved detection, suggesting a role for interventions that aim to shift clinicians’ error detection toward the limits of their knowledge. Such interventions may provide means to decrease adverse events resulting from human error.