Research Projects

Cognitive Complexity and Error in Critical Care

Funded by: James S. McDonnell Foundation, St. Louis

Principal Investigator: Vimla L. Patel, Ph.D., D.Sc. (ASU)
Co-Principle Investigators
Trevor Cohen, M.D., Ph.D. and Kanav Kahol, Ph.D. at BMI (ASU)
Marshall Smith, M.D., Ph.D. and John Ferrara, M.D. at Banner Health Care
Jiajie Zhang, Ph.D. at UT-Houston
Timothy Buchman, M.D., Ph.D. at Washington University in St. Louis

Motivation and Background
In 1999 the Institute of Medicine published a report that suggested between 44,000 and 98,000 people die each year in the United States because of preventable medical errors. Even the more conservative estimate suggests that medical errors cause more deaths annually than motor vehicle accidents, breast cancer, or HIV. This report resulted in an unprecedented focus of attention on the problem of errors in medicine, but even so there is still little evidence of widely available improvements in patient safety since the report. The complex nature of healthcare work has been proposed as a primary barrier to the implementation of effective safety measures.

Conventional approaches to error are poorly suited to this complexity.  Within the culture of medicine, the traditional approach to error involves assigning blame to a single individual. This attitude towards error is exemplified by the litigious climate and medical malpractice claims in the United States today. However, the framework of individual accountability is poorly suited to address the problem of medical error, as it fails to address the complexity of the system within which medical error occurs. Medical error is only rarely the result of the actions of a single person, and several leading error researchers have raised the importance of systemic causes of medical error. Human error will always be a factor, but recurring systemic weaknesses are amenable to intervention and correction. A productive approach to error reduction and management requires a contextual understanding of how errors tend to occur.

Furthermore, contemporary error research suggests that the process of error commission, detection and correction are an integral part of cognitive work in any complex modern-day workplace. In our previous research we have confirmed this finding in the critical care domain, by capturing and analyzing audio-recordings of clinical discourse. This methodology allows for the analysis of incidents of error correction, which would not be captured by retrospective inquiry into the cause of medical error following an adverse event. In contrast to the traditional view that error represents the deviation from a normative sequence of correct actions, we propose that the detection and correction of error are an essential part of complex cognitive work. The study of this process will reveal the extent to which the complexity of a cognitive system confers resilience. 

Research
This research proposal brings together an interdisciplinary team of cognitive scientists, critical care clinicians, simulation experts, biomedical informaticians and complex systems scientists to develop a holistic research effort dedicated to modeling of complexity and errors in critical care ( emergency medicine and intensive care) environments. Our approach to medical error emphasizes the cognitive properties of the “system” within which the error occurs. Thought processes underlying critical care decisions do not exist solely in the mind of a single individual. Rather, they are spread or distributed across the minds of many types of clinicians and providers, and across artifacts (physical objects such as notes and computer equipment). The notion of distributed cognition shifts the focus of cognitive science from the study of individuals in controlled settings to the study of groups of individuals in their real-world context.

Outside the confines of the laboratory, the thought processes of a single individual form part of a complex cognitive system with greater capacity than its individual components. The behavior of the system as a whole is considerably more complex than that of any of its parts. Decisions and actions in such an environment are always dependent on other systemic events; they are an emergent property of the system as a whole. By applying this approach, we reframe medical decision-action events as the product of the interactions of interdependent agents. In keeping with this design philosophy, we seek to model complexity and errors at varying levels of granularity (individual level, team level and health care system level).

 Medical teams from Banner Health, University of Houston and University of Washington St Louis will work with the cognitive team from Arizona State University (ASU) and University of Houston to develop protocols and model medical error. A unique opportunity lies in studying the electronic Intensive Care unit (eICU) and its effects on error models. This cognitive research will form the basis of a more comprehensive understanding of medical error.

A key element of the proposed research is to employ the developed models of complexity and distributed cognition into suggestions of relevant intervention strategies that can both predict and prevent future error. As compared to generic approaches to intervention, the knowledge gained through complexity modeling and its relation to errors will enable focused interventions designed to reduce error of a specific type. Led by the Simulation Education and Training Center at Banner Health we will develop team training protocols, simulators and on-the-job tools to aid in prevention of medical error. By studying the emergence of and recovery from error in context and developing focused intervention strategies we expect to identify new approaches for error prevention. These strategies will be based on increasing the visibility of the boundaries of safe practice, thereby enhancing error detection and correction within the complex cognitive system that underlies critical care decision-making.

 In addition to providing generalizable methodologies for the development of contextually appropriate error interventions, the proposed research will have enduring impact through the development of a robust theoretical model of error (Science aspect of the proposed research) that incorporates concepts derived from the study of complex systems, distributed cognition, and medical error. This model will facilitate the definition of cognitive complexity within a distributed cognitive system, as well as the characterization of the relationship between this complexity and vulnerability to error.

 

Dr. Shortliffe, Dr. Patel and Dr. Bruer, President of the James S. McDonnell Foundation