Background: Missed diagnoses are an important area of care quality that results in significant morbidity and mortality. Determining rates and causes has been limited by difficulties in screening, including the effort of manual chart review. We developed an innovative semi-automated review tool to expedite screening for diagnostic errors in an electronic medical record (EMR).
Methods: We retrospectively reviewed all patients seen in the emergency department (ED) of a teaching hospital over 30 days, using an automated screen to identify those who had a prior ED or ambulatory visit in the 14 days preceding their sentinel ED visit. We collected prior and subsequent notes from the institution’s EMR, then constructed a Microsoft Access database with a visual display designed for rapid comparison of previous visits to the sentinel ED visit. Each sentinel visit was assessed for potential missed or delayed diagnosis, and rated as “definite, probable, possible, unlikely or none” by a general internist. All “definite, probable, possible, or unlikely” cases were re-reviewed by another internist to verify and reach consensus.
Results: A total of 2385 minutes were required to review 1223 cases, for a mean time of 1.9 minutes/case. Of 5092 unique ED visits, 1523 had a prior ED or outpatient visit within 14 days; 1223 of these visits were randomly chosen for review. Out of the 1171 cases remaining after removal of 52 visits which ended with elopement from the ED, 32 were felt to “definitely” or “probably” have had a missed diagnosis (2.7%), while another 101 were felt to have had a “possible” or “unlikely” missed diagnosis (8.6%). Pneumonia was the most frequent diagnosis among cases felt to “definitely” or “probably” represent missed diagnoses. Reducing the preceding interval from 14 to 4 days improved the yield to 4% from 2% and captured 72% of the “probable” and “definite” errors.
Conclusion: Combining electronic notes with a new dedicated review tool was found to permit exceedingly rapid “chart” review and efficient screening for diagnostic errors. 2.7% of all visits to this ED with a preceding healthcare visit within two weeks had evidence of a prior encounter in the EMR where a diagnosis was likely missed, while another 8.6% had a potentially missed diagnosis. A number of types of diagnoses were missed, with pneumonia the most frequent. Taking advantage of downstream outcomes to illuminate upstream care and decision-making represents a powerful paradigm for uncovering diagnostic error and identifying areas for improvement.
See more of: The 33rd Annual Meeting of the Society for Medical Decision Making