Intervening to Prevent Contextual Errors in Medical Decision Making

We enrolled 139 internal medicine residents at 2 VA hospitals, Jesse Brown and Hines, in a randomized controlled design. Half participated in a 4 hour seminar series integrated into their ambulatory curriculum. Each month a total of 8 residents participated. Following the intervention there were 3 levels of assessment: (1) All participants, intervention and control, participated in a brief exercise interviewing 4 standardized patients (SPs). Note that we separately enlisted the assistance of 8 attending physicians to assist with case development for these SPs. (2) The research team subsequently enrolled 3 real patients from each physician's practice with "red flags" such as poor adherence, or missed visits, suggestive of contextual issues that need to be addressed. Physicians were scored on their performance at identifying the underlying contextual factors that account for these red flags and on formulating an appropriate plan of care. (3) The coders prospectively defined successful vs. unsuccessful outcomes for each case. At the follow up visit data was collected on whether the desired outcome was achieved. The analysis compared the skills, performance and outcomes of the intervention compared with the control group to determine the efficacy of training residents to individualize care.