Statement of problem: Although there is increasing appreciation of the importance of avoidance of diagnostic error in maintaining patient safety, healthcare institutions have struggled with the best means of addressing the issue. The very identification of these errors remains problematic as does the most appropriate method for analyzing such errors. The complexity inherent in diagnostic error, including the multitude of disparate causes, would seem to lend itself well to analysis by the root cause analysis (RCA) process. As currently constructed, however, the structure of the standard RCA is not likely to capture or identify some of the more common contributors to diagnostic error, including affective bias and cognitive mistakes. We thus aimed to redesign the RCA process to more closely reflect the causes of diagnostic error and applied this construct to two sentinel events in which diagnostic error was thought to play a key role.
Description of the intervention or program: The RCA classification was modified from elements of the Diagnostic Error Evaluation Research taxonomy tool and the system proposed by Graber et al. The scheme was then used as the basis for RCA of two cases of diagnostic error that had been identified as sentinel events.
Findings to date: Use of the modified classification scheme allowed for the identification of multiple causes of diagnostic error that otherwise may have been overlooked. These included affective error, incomplete physical examination, and poorly defined roles among multiple consultative services. Specific interventions designed as a result included a departmental curriculum in affective error, an algorithm for the evaluation of patients with a specific presenting complaint, and an institutional consultation protocol.
Lessons learned: A reorganized RCA process centered on diagnostic error may allow for a detailed analysis of cases of suspected diagnostic error and the development of interventions designed to decrease the likelihood of recurrent errors. The best method for identifying sentinel events that are appropriate for such analysis, however, is unclear.
Figure One
See more of: The 33rd Annual Meeting of the Society for Medical Decision Making