Viewpoint
Life, Death and Taxes:
Risk and Decision Analysis in Patient Safety
By Robert C. Lee
Health care systems can be extremely complex. They consist of interacting and constantly changing human and equipment (devices, hardware, software) elements, and thus are associated with a high degree of uncertainty and are prone to failures. Hippocrates admonished health care providers to "do no harm," but, for example, a recent study [Baker et al. 2004] reported that 185,000 adverse events occur annually in Canadian hospitals (an approximate 7.5 percent error rate!). Similar rates have been found in other countries. This is unacceptable by any measure.
As a result of such studies, there has been a massive overhaul of most health care systems in the developed world to implement a "culture of safety." Health care institutions have looked to industries such as the airline and chemical industries as examples of institutions that have been successful in reducing adverse events, and have begun to make great strides. However, thorny issues remain with regard to rare events that can be catastrophic in impact to both the customers and the industry itself.
A chance meeting between myself and a senior administrator in the field of cancer treatment has led to a research program designed to take a systems approach to these issues [Lee et al. 2004]. As an example, treatment of cancer with ionizing radiation has clear benefits in terms of survival of patients, but can be subject to systematic errors that have resulted, in some rare cases, in hundreds of patients being exposed to inappropriate treatments. The consequences can be disability and death.
A major problem has been that a systems approach has not been historically applied in radiation therapy (or most other health delivery scenarios), and no one person or unit in a particular institution has knowledge of the entire system and the complex interactions of people and equipment, thus resulting in a classic scenario for errors. Furthermore, although large human and financial resources are expended for quality control, to date there has not been a systematic process for efficient allocation of these resources. This is particularly important in a publicly funded government health care delivery system such as Canada's where taxes pay for health care, annual budgets are fixed, and strategic planning is typically rudimentary.
It was immediately apparent, once we formed a team of administrators, specialists and staff, that simple risk assessment methods that are being used in health care such as root cause analysis and failure mode and effects analysis (FMEA) were not sufficient to inform these decisions. I have a background in environmental and engineering risk and decision analysis, and saw that this risk management and resource allocation scenario was amenable to state-of-the-art methods in probabilistic risk and decision analysis (Figure 1).
Figure 1: Probabilistic Risk and Decision Analysis (PRADA) model.
Defining the problem and the system — the first step in any careful analysis — took considerable time, as this had never been done for the purpose of risk management in this institution. Our efforts have resulted in a system map that provides the basis for quantitative modeling (Figure 2). We have taken a fault tree approach for disaggregating all the important sources of error (Figure 3), and we are using influence diagrams to model uncertainties. Simultaneously, we are implementing an incident tracking and learning system to inform the model and the organization, and vice versa. Figure 4 shows how this analysis framework is integrated.
Figure 2: Radiation therapy system map.
Figure 3: High-level fault tree structure.
Figure 4: Integrated risk management approach.
We have discovered that typical means of modeling risk such as fault and event trees are not efficient in capturing the complexities and uncertainties associated with this system, thus we are exploring the application of Bayesian networks and discrete event simulation in our longer-term research program. Furthermore, typical economics-based decision criteria such as "cost-effectiveness" do not capture the multi-attribute nature of these risk management decisions; thus we are building capability for this type of assessment.
Our research to date has been invaluable to the cancer treatment administrators and professionals in terms of understanding the system and the tradeoffs involved in risk management. The process of organizational change will take a long time, as most health care delivery individuals and institutions are not accustomed to systems thinking. We hope that our research will provide a model for other scenarios in health care and beyond.
Robert C. Lee (rclee@ucalgary.ca) is an assistant professor in the Departments of Community Health Sciences and Oncology, Faculty of Medicine, University of Calgary; a member of the Centre for Health and Policy Studies, Southern Alberta Cancer Research Institute; a research fellow at the Institute of Health Economics; and director of the Calgary Health Technology Implementation Unit, Foothills Medical Centre.
References
- Baker, G.R., Norton, P.G, 2004, "Adverse events and patient safety in Canadian health care," CMAJ, Vol. 170, No. 3, pgs. 353-354.
- Lee, R.C., Kelly, K.L., Newcomb, C., Cooke, D., Ekaette, E., Craighead. P., and Dunscombe, P., "Quantitative approaches to patient safety: Research in risk analysis and risk management as applied to radiotherapy," Alberta Heritage Foundation for Medical Research Health Technology Assessment, report series, www.ahfmr.ab.ca.
OR/MS Today copyright © 2005 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Rd., Suite 220, Marietta, GA 30060 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com
URL: http://www.lionhrtpub.com
Web Site © Copyright 2005 by Lionheart Publishing, Inc. All rights reserved.