Presentation
A Descriptive Analysis of Physician Diagnostic Work in the Emergency Department Using SEIPS
DescriptionBackground/Introduction:
Diagnostic error (DE) has been established as a source of patient harm, morbidity, and mortality for patients. From a provider perspective DE results in high malpractice payouts and serves as a source of burn out. Despite the negative effects, we have yet to create meaningful and sustainable solutions that help either patients or providers. We theorize much of the reason is the lack of understanding of the complex and dynamic interactions that influence diagnostic process and outcomes. Literature published to date focuses on physician faulty logic or slips regarding diagnostic result follow up in the system. Discussion within the literature about the contributions of the sociotechnical system factors to diagnostic harm, reflects a limited understanding of system interactions that create outcomes observed. Indeed, the literature continues to focus on error, despite a lack of a consistent common grounding on how that is defined. There is a lack of appreciation of the necessity of process deviations, what contributes to process failures, and the interactive nature of the diagnostic process, misdiagnosis, and harm. Through an observational study of attending physicians in a pediatric emergency department, we used the Systems Engineering Initiative for Patient Safety (SEIPS) framework to characterize the complexity and ambiguity of the environment not previously described in the literature. The pediatric (ED), was specifically chosen for the high density of diagnostic work. We aim to lead an important shift in thinking away from a confused focus on error, towards an understanding of interactions and adaptations that might be amenable to more effective work systems interventions.
Methods:
Direct observations were conducted by trained observers of the attending physicians’ work associated with “index patients". Observations were completed in the summer of 2023 in 50-min blocks, capturing time and frequency associated with tasks, people, tools/technologies, and physical locations. Data were recorded in Redcap using a template developed specifically for this project and informed by SEIPS 2.0.
Results:
We observed 19 physicians (males = 5) who completed 2,836 tasks over 66 index patient encounters. Physicians’ work required rapid task switching, completing on average 43 tasks in 50 minutes, with each task lasting on average 1.12 minutes. Specifically for the index patient, the physician spent an average of 8.75 minutes completing 7.2 tasks, interacting with 12.6 people and utilizing 3.5 tools/technologies. 97.7% of the time, caring for the index patient included other people, consisting primarily of the patient and family, resident doctors, the nurse, and other hospital personnel. Consequently, the most common tasks were spending time with the patient (i.e., counseling and taking a history) as well as interacting with a resident doctor (i.e., hearing about the patient's story as well as teaching the resident doctor). Of the tools and technology used, 70% were computer-based tools located in the medical record, 21% were physical items such as a stethoscope or the physician’s own paper documents, while 5% was related to phone use, and 4% was using online decision support tools.
Conclusion:
The diagnostic work observed was complex and messy. Our results indicate that the physician work is indeed patient centered, prioritizing the patient and family, completing essential tasks of history taking and counseling. Nearly all of the time physicians were caring for the index patient they were also interacting with others illustrating the interdependency of people in physicians’ work. This dependence on human interactions demonstrates the limited perspective of the ‘human error’ attributions, or the focus on individual physicians and their cognitive and diagnostic abilities. Physicians completed tasks quicker than anticipated with many more interactions challenging the human limitations of the observers. When building the rhetoric for why misdiagnosis or harm occurs, we suspect that task demands and the interdependence of others dictating the cadence of physician work to be a future area of exploration for vulnerabilities in the diagnostic process. Much of the modeling in the diagnostic literature has been oversimplified and we feel neglecting to consider the complexity captured in this study. Moving away from simplistic conceptions of the diagnostic process towards understanding work-as-done will offer many new ways to enhance diagnosis, improve outcomes, and reduce harm.
Diagnostic error (DE) has been established as a source of patient harm, morbidity, and mortality for patients. From a provider perspective DE results in high malpractice payouts and serves as a source of burn out. Despite the negative effects, we have yet to create meaningful and sustainable solutions that help either patients or providers. We theorize much of the reason is the lack of understanding of the complex and dynamic interactions that influence diagnostic process and outcomes. Literature published to date focuses on physician faulty logic or slips regarding diagnostic result follow up in the system. Discussion within the literature about the contributions of the sociotechnical system factors to diagnostic harm, reflects a limited understanding of system interactions that create outcomes observed. Indeed, the literature continues to focus on error, despite a lack of a consistent common grounding on how that is defined. There is a lack of appreciation of the necessity of process deviations, what contributes to process failures, and the interactive nature of the diagnostic process, misdiagnosis, and harm. Through an observational study of attending physicians in a pediatric emergency department, we used the Systems Engineering Initiative for Patient Safety (SEIPS) framework to characterize the complexity and ambiguity of the environment not previously described in the literature. The pediatric (ED), was specifically chosen for the high density of diagnostic work. We aim to lead an important shift in thinking away from a confused focus on error, towards an understanding of interactions and adaptations that might be amenable to more effective work systems interventions.
Methods:
Direct observations were conducted by trained observers of the attending physicians’ work associated with “index patients". Observations were completed in the summer of 2023 in 50-min blocks, capturing time and frequency associated with tasks, people, tools/technologies, and physical locations. Data were recorded in Redcap using a template developed specifically for this project and informed by SEIPS 2.0.
Results:
We observed 19 physicians (males = 5) who completed 2,836 tasks over 66 index patient encounters. Physicians’ work required rapid task switching, completing on average 43 tasks in 50 minutes, with each task lasting on average 1.12 minutes. Specifically for the index patient, the physician spent an average of 8.75 minutes completing 7.2 tasks, interacting with 12.6 people and utilizing 3.5 tools/technologies. 97.7% of the time, caring for the index patient included other people, consisting primarily of the patient and family, resident doctors, the nurse, and other hospital personnel. Consequently, the most common tasks were spending time with the patient (i.e., counseling and taking a history) as well as interacting with a resident doctor (i.e., hearing about the patient's story as well as teaching the resident doctor). Of the tools and technology used, 70% were computer-based tools located in the medical record, 21% were physical items such as a stethoscope or the physician’s own paper documents, while 5% was related to phone use, and 4% was using online decision support tools.
Conclusion:
The diagnostic work observed was complex and messy. Our results indicate that the physician work is indeed patient centered, prioritizing the patient and family, completing essential tasks of history taking and counseling. Nearly all of the time physicians were caring for the index patient they were also interacting with others illustrating the interdependency of people in physicians’ work. This dependence on human interactions demonstrates the limited perspective of the ‘human error’ attributions, or the focus on individual physicians and their cognitive and diagnostic abilities. Physicians completed tasks quicker than anticipated with many more interactions challenging the human limitations of the observers. When building the rhetoric for why misdiagnosis or harm occurs, we suspect that task demands and the interdependence of others dictating the cadence of physician work to be a future area of exploration for vulnerabilities in the diagnostic process. Much of the modeling in the diagnostic literature has been oversimplified and we feel neglecting to consider the complexity captured in this study. Moving away from simplistic conceptions of the diagnostic process towards understanding work-as-done will offer many new ways to enhance diagnosis, improve outcomes, and reduce harm.
Event Type
Oral Presentations
TimeTuesday, April 11:52pm - 2:15pm EDT
LocationQueens Quay
Patient Safety and Research Initiatives (PS)