Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 139
Building a Better Delivery System: A New Engineering/Health Care Partnership Engineering Tools and Procedures for Meeting the Challenges
OCR for page 140
Building a Better Delivery System: A New Engineering/Health Care Partnership This page intentionally left blank.
OCR for page 141
Building a Better Delivery System: A New Engineering/Health Care Partnership Systems Engineering: Opportunities for Health Care Jennifer K. Ryan Purdue University Systems engineering involves the design, implementation, and control of interacting components or subsystems. A system consists of interacting, interrelated, or interdependent elements that form a complex whole, a set of interacting objects or people that behaves in ways individuals acting alone would not. The overall goal of systems engineering is to produce a system that meets the needs of all users or participants within the constraints that govern the system’s operation. The objectives can generally be divided into two broad categories: service and cost. Service can be measured by a variety of criteria, such as availability, reliability, quality, and so on. Cost is usually measured by how much costs can be reduced or at least controlled. A final objective of systems engineering is to gain a better understanding of system behavior and the problems associated with it. Models enable us to study the impact of alternative ways of running the system—alternative designs or controls and different configurations and management approaches. In short, systems engineering models enable us to experiment with systems in ways we cannot experiment with real systems. Systems engineers generally prefer to work with analytical or mathematical models rather than with conceptual models because they are generally better defined, have more clearly defined assumptions, and are easier to communicate, manipulate, and analyze. We begin with a graphical representation of the system, which often includes a diagram showing the flow of information and resources. We then create a mathematical description that includes objectives, interrelationships, and constraints. The components of the mathematical model can be divided into four categories: (1) decision variables, which represent our options; (2) parameters or givens, which are the inputs to the decision-making process; (3) the objective function, which is the goal, the function to be optimized; and (4) the constraints, which are the rules that govern operation of the system. When dealing with large complex systems, we often deconstruct it into smaller subsystems that interact with one another to create a whole. The decision-making structure provides natural breaks in the system. We model and analyze the subsystems and then connect them in a way that recaptures the most important interdependencies between them. Systems engineering requires a variety of quantitative and qualitative tools for analyzing and interpreting system models. We use tools from psychology, computer science, operations research, management and economics, and mathematics. The quantitative tools include optimization methods, control theory, stochastic modeling and simulation, statistics, utility theory, decision analysis, and economics. Mathematical techniques have the capability of solving large-scale, complex problems optimally using computerized algorithms. Mathematical models clarify the overall structure of a system and reveal important relationships. They enable us to analyze the system even when data are sparse. Models, combined with analyses, reveal the most critical parameters and enable us to analyze the system as a whole. Sensitivity analysis involves testing out trade-offs. Before we can convert a model solution to an implementable solution, we must test and validate the model to ensure that it actually predicts the behavior of the system. A logistics system can be defined as a network of suppliers, manufacturing centers, warehouses, distribution centers, retail outlets, and end consumers. The system includes raw materials, work in process, inventory, finished products, all of the materials in the system, all of the information that flows within the system, and all of the resources in the system (e.g., people, equipment, etc.). Logistics-systems engineering can be defined as the planning, implementation, and control of the system to ensure the efficient, cost-effective flow and storage of all materials and information from point of origin to point of consumption for the purpose of meeting customer requirements. Our goal is to ensure that the right
OCR for page 142
Building a Better Delivery System: A New Engineering/Health Care Partnership amount of materials or resources is in the right place at the right time at minimum cost. We deliberately leave the definition of service (i.e., meeting customer requirements) somewhat vague so we can define the needs and requirements of different customers in different ways. Logistics-systems engineering involves the difficult problem of simultaneously improving customer service and quality, improving timeliness, reducing operating expenses, and, if possible, minimizing capital investment. We are also interested in answering strategic questions, such as where we can expand capacity or what types of collaboration with customers or suppliers would be most beneficial. Systems engineering problems have some common characteristics. They tend to be interdisciplinary, involving both technical and nontechnical fields. They require multiple, high-level, or strategic metrics or performance measures, often measurements of nonquantitative factors (e.g., customer satisfaction). They involve many participants with different value systems and many decision makers; therefore, we have to find optimal solutions that meet conflicting criteria. The systems and issues tend to be hierarchical and complex, but the systems also evolve and change over time; they generally involve significant uncertainties. Much of the current research in logistics is driven by the needs of public and private organizations, such as health care systems, that operate in environments characterized by intense competition, constant change, and a strong focus on customer needs. Health care delivery systems, for example, consist of a variety of health care organizations, caregivers, and patients. State and federal governments are involved, as well as a variety of other organizations. These complex systems also involve a large number of interconnections between the components and the system—multihospital systems and provider networks with linkages between hospitals, physician groups, insurers, and others. There are also many decision makers who often have conflicting criteria, and there are complex interactions between participants. The effective organization and management of a health care delivery system requires careful management of resources to ensure that the necessary staff and equipment are in the right place at the right time. The problem is complicated by uncertainties and system complexity. Some aspects of the health care delivery system, such as government intervention, the level of uncertainty, and the nature of the demand, appear to be unique to health care. But similar problems can be found in other industries, such as the telecommunications and electricity industries, which also have to factor in government intervention. The nature of the uncertainties may be different, but they have similar effects on the system. Both the telecommunications and electricity industries have used logistics models to their advantage. Systems engineering models can provide structured, quantitative methods of studying alternative control policies and system designs for almost any industry. The methods can be used to help coordinate information systems, operations, and capital investment; develop control policies; predict and evaluate outcomes; and evaluate the benefits and costs of a given program or system design. The elements included in the model depend on the question or problem to be solved. For the output of the model to be useful, it must mimic the expected behavior of the real system. To control the behavior of one part of the system, the incentives driving that aspect of the system must be built into the model. A good deal of literature is now available on research in this area. Operations research tools and systems engineering tools have been used to address a wide variety of problems, from the operation of a hospital to higher levels of complexity, such as incentives, efficiency, and payment schemes. Quantitative models can provide important input for making decisions that involve complex societal, ethical, and economic issues.
OCR for page 143
Building a Better Delivery System: A New Engineering/Health Care Partnership Supply-Chain Management and Health Care Delivery: Pursuing a System-Level Understanding Reha Uzsoy Purdue University In recent years, effective supply-chain management has emerged as a significant competitive advantage for companies in very different industries (e.g., Chopra and Meindl, 2000). Several leading companies, such as WalMart and Dell Computer, are differentiated from their rivals more by the way they manage their supply chains than by the particular products or services they provide. A supply chain can be defined as the physical and informational resources required to deliver a good or service to the final consumer. In the broadest sense, a supply chain includes all activities related to manufacturing, the extraction of raw materials, processing, storing and warehousing, and transportation. Hence, for large multinational companies that manufacture complex products, such as automobiles, machines, or personal computers, supply chains are highly complex socioeconomic systems. The ability of successful firms to make the effective management of supply chains a source of competitive advantage suggests that there may be useful knowledge that can provide a point of departure for the development of a similar level of understanding of certain aspects of health care delivery systems. Similar to the supply chains in manufacturing and other industries, the health care delivery system is so large and complex that it has become impossible for any individual, or even any single organization, to understand all of the details of its operations. Like industrial supply chains, the health care “supply chain” consists of multiple independent agents, such as insurance companies, hospitals, doctors, employers, and regulatory agencies, whose economic structures, and hence objectives, differ and in many cases conflict with each other. Both supply and demand for services are uncertain in different ways, making it very difficult to match supply to demand. This task is complicated because demand for services is determined by both available technology (i.e., available treatments) and financial considerations, such as whether or not certain treatments are covered by insurance. Decisions made by one party often affect the options available to other parties, as well as the costs of these options, in ways that are not well understood. However, almost all of these complicating factors are also present, to one degree or another, in industrial supply chains; the progress made in understanding these systems in the last several decades is a cause for hope that some insights and modeling tools developed in the industrial domain can be applied to at least some aspects of health care delivery systems. In general, a centralized approach to controlling the entire system is clearly out of the question, although centralized decision models may be useful for coordinating the operations of segments of the larger system controlled by a single decision-making body. Designing decentralized models of operation that render the operation of the overall system as effective as possible is the main challenge for both health care delivery and industrial supply chains. In the following section, I shall briefly discuss how the study of industrial systems has evolved from individual unit processes to considerations of complex interactions among many different components of an industrial supply chain. I shall then describe some examples of modeling approaches that have been applied to supply chains and close with some comments on how these tools might be adapted for the health care delivery environment. FROM UNIT PROCESSES TO SUPPLY CHAINS If we examine how industrial operations, particularly manufacturing operations, have evolved since the beginning of the nineteenth century, we can see that many efforts were motivated by a desire to understand and optimize individual unit processes (see, for example, Chandler, 1980). These efforts led to many innovations, among them the development of improved machine tools and fixtures, a significantly better understanding of the chemistry of processes (e.g., steel-making), and through the work of the early industrial
OCR for page 144
Building a Better Delivery System: A New Engineering/Health Care Partnership engineers, such as Frederick Taylor and Frank and Lillian Gilbreth, the optimization of interactions between workers and their environment. As the understanding of unit processes developed, engineers began to consider larger and larger groupings of unit processes, trying to understand interactions between them and optimize the performance of entire systems, sometimes to the detriment of individual components. Hence, from considering individual unit processes, we progressed to considering departments of factories that perform similar operations, entire manufacturing processes from raw materials to finished products, and eventually, the operations of entire firms, as well as their suppliers and customers. It has often been observed that most significant new opportunities, both for cost reduction and the generation of new products and services, have been based on an understanding of interactions between different subsystems, or different agents, operating in the supply chain. Among today’s leading companies, examples abound. Many automotive companies, for instance, have developed joint ventures with transportation firms; the objective is to optimize the interface between the production and distribution functions and facilitate the just-in-time operation of automakers’ final assembly plants. Software companies that provide supply-chain planning software for multilocation companies is another strong indicator of the advantages companies perceive will accrue to them by the effective management of the various elements of their supply chains. The strong trend in industry to outsource noncritical functions has increased the need for companies to effectively manage and clearly understand their relationships with other companies. As a final example, we can point to the collaborative forecasting, planning, and replenishment initiative in the retail sector; retailers work closely with major suppliers to develop demand forecasts for products through information-sharing and joint planning processes. Clearly, the basic process of improving a system by a detailed understanding of the most fundamental unit processes, in other words the “atomic” elements of the system, and steadily extending that knowledge to interactions among larger and larger groupings of these elements is directly applicable to health care delivery systems. The individual unit processes in this case include the processing of a patient in an emergency room, the process by which a medical insurance claim is approved, and the scheduling of hospital operating rooms to optimize their performance. The need for a better understanding of how the operations of individual elements affect each other is apparent; these interactions can be quite complex because of long time lags between cause and effect. For example, the decision by a regulatory agency to disallow a certain kind of preventive procedure for infants may result in the emergence of an unexpectedly large number of children with special needs in the elementary school system several years later. The same kinds of problems are present to some degree in industrial supply chains, and a significant body of knowledge has been developed over the years to address them. Based on the history of industrial enterprises, we know that the development of today’s enterprises required substantial organizational innovations, such as capital budgeting to allocate scarce capital between competing activities, cost accounting to develop an understanding of factors contributing to product costs, and the development of multidivisional corporations with complex structures of management incentives and coordination mechanisms. An important development in recent years has been the recognition of the need for a cross-functional view of supply-chain operations. All aspects of a firm’s operation, from the design of a product to the specific timing of marketing promotions, have a direct effect on the operation of the supply chain. Therefore, different functional specialties must actively collaborate to develop solutions to optimize the performance of the overall system. Similarly, in health care delivery a number of different constituencies, such as doctors, government agencies, insurance providers, and patient groups, are all involved in the operation of the health care delivery supply chain. KNOWLEDGE OF SUPPLY-CHAIN MANAGEMENT In the domain of industrial supply chains, it is probably safe to say that we have developed a fairly good understanding of the operation and economics of individual unit processes, including functions such as transportation, distribution, warehousing, and information processing. In particular, we have developed a substantial understanding of the often complex dynamics of capacity-constrained systems subject to variability in both demand and process (Hopp and Spearman, 2000). However, in general we are only beginning to learn how to integrate the solutions to these individual elements to reach a reasonable understanding of the operation of the overall supply chain. Integrated planning models based on linear and integer programming have been applied to the segments of the supply chain controlled by a single company for at least four decades (e.g., Johnson and Montgomery, 1974). Although these models have been successful in many instances, they have not been effective in addressing the needs of a supply chain that involves many different companies with potentially conflicting objectives. In recent years, considerable efforts have been made to use some of the tools of economics, such as contracts, as a mechanism for coordinating the operation of complex supply chains (Tayur et al., 1998). However, these models are generally subject to long-run, steady-state assumptions that can be carefully evaluated relative to market conditions. Conventional Monte Carlo simulation techniques (Law and Kelton, 1991) have proven extremely effective for systems in which the operational dynamics can be described at a high level of detail, such as segments of manufacturing processes or hospital operations. The difficulty with these
OCR for page 145
Building a Better Delivery System: A New Engineering/Health Care Partnership models is that for large-scale systems the level of detail required to unequivocally model system behavior accurately becomes prohibitive in terms of both data collection and computation time. Systems dynamics models used to model large systems work by establishing input-output relationships for their components and simulating their operation through time using techniques based on the techniques used for the numerical solution of differential equations (Sterman, 2000). Although these techniques are capable of modeling large, complex systems, they usually do so by specifying aggregate input-output relationships for large subsystems, which must be validated and whose parameters must be estimated carefully. Nevertheless, these models can capture many critical aspects of supply-chain behavior, such as the “bullwhip effect,” in which variability in orders is amplified as it passes down the supply chain from the consumer towards the producers of raw materials (Forrester, 1962). RESEARCH NEEDS AND FUTURE DIRECTIONS At the risk of overgeneralizing, it appears that most of the tools required for analysis of the individual unit processes in health care delivery, such as efficiency of hospital facilities, have been developed in the engineering literature and have, in fact, been applied intermittently to a variety of systems over the last several decades (e.g., Pierskalla and Brailer, 1994). However, if our experience with industrial supply chains is any guide, only limited improvements in health care delivery can be obtained by these means. Repeated experience has shown that far greater improvements can be obtained by a thorough understanding of the interactions between different elements of the system and restructuring them in a way that leaves all parties better off. This brings the modeling issues squarely into the region where current supply-chain research is weakest (the effective coordination of socioeconomic systems consisting of multiple, independent agents); but this is also the area that is developing most rapidly. The development of novel models at the intersection of conventional engineering and economics promises to provide a wide range of challenging research problems for many years to come. To support this agenda, the most pressing research need is for techniques that can be used to model systems at the aggregate level, where one can accept some level of approximation to obtain computationally tractable models that achieve the correct qualitative behavior and provide useful insights into interactions between systems. This means that the aggregate models must capture the often nonlinear relationships between critical variables correctly, which has not always been the case in supply-chain modeling. The literature on systems dynamics may be a good starting point for this initiative, but it must be complemented by a variety of other techniques, such as economic models of competition and collaboration and agent-based techniques for modeling complex systems. It is important to bear in mind that the purpose of these models is far more likely to be descriptive than prescriptive, that is, models are far more likely to be used, and arguably far more useful, to inform debate between the various parties involved in health care delivery than to deliver decisions to be executed. Hence, the development of large-scale computational simulations of different scenarios with different actors and interaction protocols between the actors appears to offer interesting research challenges. These tools would be extremely beneficial to decision makers in health care delivery. REFERENCES Chandler, A.D. 1980. The Visible Hand: The Managerial Revolution in American Business. Cambridge, Mass.: Belknap Press. Chopra, S., and P. Meindl. 2000. Supply Chain Management: Strategy, Planning and Operations. Englewood Cliffs, N.J.: Prentice-Hall. Forrester, J.W. 1962. Industrial Dynamics. Cambridge, Mass.: MIT Press. Hopp, W., and M.L. Spearman. 2000. Factory Physics. 2nd Edition. New York: McGraw-Hill/Irwin. Johnson, L.A., and D.C. Montgomery. 1974. Operations Research in Production Planning, Scheduling and Inventory Control. New York: John Wiley & Sons. Law, A., and W.D. Kelton. 1991. Simulation Modeling and Analysis, 2nd edition. New York: McGraw-Hill. Pierskalla, W.P., and D.J. Brailer. 1994. Applications of Operations Research in Health Care Delivery. Vol. 6, pp. 469–505 in Handbooks in OR & MS, S.M. Pollock, M.H. Rothkopf, and A. Barnett, eds. Amsterdam, The Netherlands: Elsevier Science. Sterman, J.D. 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World. New York: McGraw-Hill. Tayur, S., M. Magazine, and R. Ganesham, eds. 1998. Quantitative Models for Supply Chain Management. Amsterdam, The Netherlands: Kluwer Academic Publishers.
OCR for page 146
Building a Better Delivery System: A New Engineering/Health Care Partnership This page intentionally left blank.
OCR for page 147
Building a Better Delivery System: A New Engineering/Health Care Partnership The Human Factor in Health Care Systems Design Kim J. Vicente University of Toronto The simplest way to think about the discipline of engineering is that engineers design things that are useful to society and satisfy important needs based on what we know about the physical world. When a bridge fails, we do not usually blame the bridge. We look to its design, trying to find a mismatch between what we know about the physical world and the outcome. We should apply this same logic to people. But when a system is poorly designed, we often blame the person using it rather than the flaws in the system. For example, when we design a mechanical lathe, we must place the mechanical controls in a way that respects what we know about human bodies. But sometimes, if a lathe is poorly designed, we blame the user rather than the design. Although we know a great deal about teamwork and about human behavior at the organizational and political levels, that knowledge is not always taken into account by designers of health care systems and devices. Clearly, improvements could be made, and not just in terms of safety. The lack of respect for human nature in the design of health care systems causes injuries and deaths, but it also costs money. Contrast that to the field of aviation. Despite September 11, 2001 was not a bad year for aviation safety. The average number of deadly crashes for the previous decade was 48 per year. In 2001, however, there were only 34 deadly crashes—worldwide, not just in the United States. That’s the lowest number since 1946 when there were far fewer flights. One reason for the improvement is that aviation engineers pay attention to the human factor. A familiar example is the rather high rate of crashes in a certain type of aircraft that occurred because pilots tended to raise the landing gear as the plane was landing, causing the airplane to scrape along the runway. When Al Chapanis, an aviation engineer, studied the problem, he found that the controls for the landing gear and the wing flaps were right next to each other and that they looked and felt identical. He realized that pilots could easily grab the wrong control, but he also realized that he could not redesign the whole cockpit. He came up with an idea, now called shape coding. He did not move the controls, but he altered the feel of the landing gear control. The controls are still right next to each other, but the change eliminated the errors. It was as simple as that. Can we apply the same type of thinking to health care systems? Patient-controlled analgesic devices, which allow patients to self-administer analgesics (usually morphine), are a case in point. A number of parameters are programmed into these devices by the nurse, the most important being drug concentration. These devices rely strictly on the programming and cannot independently verify either the concentration or even the type of analgesic in the syringe. Therefore, errors in programming can mean underdoses or overdoses; and errors have enduring effects, that is, the problem lasts until the programming is corrected. For the particular device that we studied, programming errors were associated with five to eight reported patient deaths. Adverse drug events and adverse events in general in medicine are severely underreported—roughly only 1.2 to 7.7 percent are reported (Vicente et al., 2003). In other words, adverse events may be 13 to 83 times higher than the reported rate. We calculated that programming errors had lethal results for this particular device at least 65 times, and perhaps as many as 667 times, over a 12-year period. To put these numbers in context, the manufacturer reports that the device was used safely over 22 million times. We then examined the existing design using traditional human-factor principles to see if there was room for improvement. We also talked to nurses, the users of this device. One serious problem we found was that the layout of the buttons on the interface was confusing and counterintuitive. So we came up with a new design by resegmenting the buttons and changing some of the labels. The new design offered the same functionality but changed the mode of interaction between the programmer and the pump. The system now provided more feedback and gave the user an overview
OCR for page 148
Building a Better Delivery System: A New Engineering/Health Care Partnership of the programming sequence. The redesigned device told the programmer the drug concentration, what was coming up next, how to program the mode, and then showed the settings. In essence, the new programming sequence was much less convoluted. We tested the redesigned interface in a laboratory setting with professional nurses who had more than five years of experience programming the commercial device. With the commercially available design, there were eight programming errors for drug concentration, three of which were undetected. With the new interface, there were no errors in drug concentration. They were eliminated. Given the epidemiological data, the change was obviously important for safety reasons. But it was also important in terms of cultural attitudes. If the problem had originated with the person programming the device, then changing the interface should have made no difference in the error rate. In fact, changing the design did eliminate the errors. Therefore, we concluded that the problem was not with the people, or, at least, not only with people. Surprisingly, we had a great deal of difficulty getting this research published. One journal refused it because the editor took for granted that what we had scientifically demonstrated was not true. We went through some pretty hard times, both in terms of getting the work published and dealing with the response from the public. One reviewer even suggested that a lawyer look at the research because of potential legal action by the manufacturer. We had chosen the particular device because it was relatively new, but soon after our research was completed, the media began to report some deaths as a result of errors in programming the device. This example shows three important points. First, we know how to design technology that works for people because we know a lot about people at many levels—physical, psychological, team, organizational, and political. We do not always make the most of this knowledge when we design health care devices, but lack of understanding is not the problem. Second, not making the most of that knowledge results in a tremendous loss to society. Tens of thousands, perhaps even hundreds of thousands, of people are injured or die every year unnecessarily. Finally—the most difficult lesson—change is important and necessary, but there is a great deal of resistance that must be overcome before we can make progress. REFERENCE Vicente, K.J., K. Kada-Bekhaled, G. Hillel, A. Cassano, and B.A. Orser. 2003. Programming errors contribute to death from patient-controlled analgesia: case report and estimate of probability. Canadian Journal of Anesthesia 50(4): 328–332.
OCR for page 149
Building a Better Delivery System: A New Engineering/Health Care Partnership Changing Health Care Delivery Enterprises Seth Bonder The Bonder Group The health care delivery (HCD) system in the United States is in crisis. Access is limited, costs are high and increasing at an unacceptable rate, and concerns are growing about the quality of service. Many, including the Institute of Medicine, believe the system should be changed significantly in two ways: (1) HCD enterprises should be reengineered to make them more productive, efficient, and effective; and (2) substantially more effort should be devoted to a strategy of prevention and management of chronic diseases instead of the current heavy reliance on the treatment of diseases. Although operations research can make substantial contributions to both areas, the focus of this paper is on: (1) reengineering HCD enterprises, particularly areas in which operations research can provide valuable support to senior health care managers; and (2) enterprise-level HCD simulation models to determine the reengi-neering initiatives with the biggest payoffs before implementation. HCD enterprises are very large, complex operational systems comprised of large numbers of people and machine elements. Tens of thousands of people are involved as providers, patients, support staff, and managers organized into specialties, departments, laboratories, and other organizations that are considered independent service units (“stovepipes”). Machines include durable medical equipment, information technologies, communications equipment, expendable supplies, rehabilitation equipment, and so on. These elements are affected by many clinical and administrative processes (e.g., arrivals, testing, diagnosis, treatment, scheduling, purchasing, billing, recruiting, etc.), most of which are probabilistic (i.e., uncertain) and change significantly over time. Perhaps most important, these processes involve large numbers of interactions within units, among units, and across processes. Decisions by enterprise managers regarding one unit may have second, third, and fourth order effects, which may be more significant than the first order effect. HCD enterprises are driven by endogenous and exogenous human decisions made by providers, patients, insurers, administrators, politicians, government employees, and others. Demand and supply issues have complex feedback effects. A great many resources are required for the development and operation of an HCD enterprise. For example, the University of Michigan’s budget for its HCD enterprise is more than $1 billion; the Henry Ford Health System’s budget is $2.5 billion, and these are relatively small HCD enterprises. Billions of dollars have been spent on cost containment initiatives over the past 15 years by the Agency for Healthcare Research and Quality (formerly the Agency for Health Care Policy and Research), the U.S. Department of Defense, the Veterans Administration, National Institutes of Health, foundations, universities, and others to reengineer the HCD system. Nevertheless, costs continue to rise at double-digit rates. We need better ways of analyzing systems of this magnitude. The operations research community has been involved with HCD enterprises for more than 40 years working on a wide range of problems, such as inventory for perishables; management of intensive care units; laboratory and radiology scheduling; relieving congestion in outpatient clinics; nurse staffing, scheduling, and assignments; and layouts for operating and emergency rooms. These efforts have focused on the small, stovepipe units, referred to by Don Berwick as clinical and support “microsystems,” and have produced some useful information for unit managers but have not addressed enterprise-level reengineering and planning issues (the so-called “macrosystem”). Macrosystem issues have interactive effects across the enterprise and have large cost, access, and effectiveness impacts. Some of these interrelated issues are listed below: the mix of health services necessary to support a given population the staff required (e.g., specialties, numbers, locations) to provide necessary services the impacts of changing demands (e.g., aging populations, effects of preventive measures)
OCR for page 178
Building a Better Delivery System: A New Engineering/Health Care Partnership which cells change before genetic defects become evident as tumors, and the distinction between prevention and therapy may disappear as detectable genetic errors are treated long before they are expressed as lesions. The achievements and promise of genomics, proteomics, and molecular discoveries, however, have not been matched by advances in the organization and delivery of services. When patients are diagnosed with cancer, they often find navigating the medical care system a nightmare. A colleague I had not seen for a while said to me, “When I was diagnosed with Stage 3 melanoma, I thought everyone in the health system would swing into action and take care of me. I didn’t realize until much later that no one could or would. It was up to me to make sure things happened and that my doctors knew about it.” She is a patient in a world-class medical center in the Baltimore-Washington area. Despite her education, her considerable resources, her excellent insurance, and her husband who took full-time leave to help her, she was not able to make the system work. The processes by which a patient accesses care (because of a symptom or for screening), receives a diagnosis, makes decisions, and plans for care in a hospital or outpatient facility or arranges for services from community service and support groups or home care may include initial treatment (such as surgery), follow-up treatment (such as adjuvant chemotherapy or radiation therapy), palliative care, education and information about community services, monitoring as a survivor, and treatment for recurrent disease, continuing primary care, and if needed, timely and appropriate end-of-life care in a hospital, hospice, or home. It may also involve genetic screening, rehabilitation, and support for family and others during and after serious illness. It is easy to understand why when Lee Atwater, campaign manager for Ronald Reagan, was diagnosed with a brain tumor and began treatment, he is reported to have exclaimed, “I need a campaign manager.” One hears the same complaints from the medical side of health care. Ensuring Quality Cancer Care, a report by the Institute of Medicine National Cancer Policy Board, states emphatically, “There is no national cancer program, care program or system of care in the United States” (IOM, 1999). A pediatric oncologist commented, “In the standard model of delivery of care to pediatric cancer patients, the onus of negotiating all aspects of treatment falls on the patient and his or her family” (Wolfe, 1993). Figure 1 shows a very common model of health care for cancer. In this distributed model, with oncologists practicing in the community, the patient goes from one doctor and laboratory to another trying to integrate sometimes conflicting information. In addition, oncologists have difficulty obtaining information, which results in waste, duplication of effort, and delays; and the primary care physician often has little information about the patient’s treatments. Care is provided in multiple settings, not only at the time of diagnosis and primary treatment, but also over time through later FIGURE 1 Distributed model of health care for cancer. treatments and follow-up, as needed. Recently, interest has grown in the use of “patient navigator” programs to help patients schedule appointments and keep up with their treatment and progress, but I am not aware that such programs have been evaluated for effectiveness (American Cancer Society, 2002; Christensen and Akcasu, 1999). Figure 2 shows a different model based on care in a comprehensive cancer center, such as M.D. Anderson, Memorial Sloan Kettering, or Dana Farber, where oncologists and other caregivers are grouped together in one facility. Even in these settings, patients may still go from one caregiver to another, and their records may be quite separate. A care coordinator, such as a nurse oncologist, might help the patient coordinate his or her care, and patients in these centers are more likely to enter clinical trials with stringent protocols and follow-up. In this model, tumor boards or multidisciplinary conferences among oncologists and pathologists develop a plan for patient care. Such conferences, which may be held FIGURE 2 Comprehensive cancer center model.
OCR for page 179
Building a Better Delivery System: A New Engineering/Health Care Partnership periodically after primary treatment, may, but usually don’t, include the patient and his or her family (Joishy, 2001). Figure 3, a pediatric multisite model, was developed by Dr. L.C. Wolfe and his colleagues when he was at the New England Medical Center (Wolfe, 1993). The model attempts to remedy the boundary problems at the transitions between settings, particularly between the hospital and home, home care and some outpatient care, and outpatient care and inpatient care. When something goes wrong, people do not always know what to do or who to contact. The model addresses these problems by having the oncologist and the nurse spend time in the hospital together with the patient and then in the outpatient setting and then, as a team, continuing to care for patients who had been in the hospital. To ease the boundary problems between hospital and home care, Wolfe devised an electronic system that enables families to transmit problems and questions to their doctors. O’Connell and colleagues (2000) have critiqued other models of care that try to integrate the hospital-community interface. Only a few efforts to design better health care delivery systems have been reported. Last week, I attended the annual meeting of the American Society of Clinical Oncology, which drew 25,000 participants from all over the world. Of the more than 3,000 abstracts published, only two reported on programs for improving care. One was a report from France on the number of cancer patients who had attended a nutritional workshop; the other was on the costs and satisfaction of palliative care service in a hospital. This points up a stark contrast. The knowledge base for the science of cancer care has undergone a radical transformation, but little attention has been paid to ensuring the consistent translation of this knowledge to the health care setting—not just for patients in cancer centers on protocols, but for all cancer patients all the time. Indeed, the assumption seems to be that the results of clinical trials will be translated into practice without error and without specifying how services should be organized and delivered. FIGURE 3 Pediatric multisite model. Source: Wolfe, 1993. The lack of well designed systems can result in the loss of benefits to patients. In many systems, failures can and do occur that could have been addressed by operational engineering. One of the most common consequences is the failure to screen patients. A research project involving health maintenance organizations found that only 50 to 83 percent of women who were expected to have mammographies in a particular year actually had them (Taplin et al., 2002). In Colorado, a risk-management study of lawsuits for failure to diagnose breast cancer found that the average length of delay from symptom to detection or detection to diagnosis was 13.4 months (Marjie G. Harbrect, M.D., personal communication, April 2001). There were many reasons for the delay, but most of them were system problems. In some cases, the primary care clinics did not have systems for tracking or follow-up. In many cases, individuals thought someone else was following up with the patient. Sometimes a lump found in an exam was not visible on a mammogram, and there was simply no follow-up. Failure to diagnose was also found in the United Kingdom, where there was on average a seven-month delay between detection and definitive diagnosis. A study in New York hospitals on women who clearly should have had adjuvant breast therapy after treatment for early-stage breast cancer found that in hospitals that were part of the Mount Sinai system, only 18 to 33 percent of these patients, depending on the hospital, received their indicated adjuvant therapy for early-stage breast cancer (Bickell and Young, 2001; Bickell et al., 2000). This was not because of a lack of knowledge. After going through the medical records of these patients and talking to the surgeons, the study found that the surgeons simply did not know what had happened to these patients, they had simply “fallen through the cracks.” Another serious problem is failure to use the evidence base. Dr. Ezekiel Emanuel (2001) at NIH recently reported on an excessive use of chemotherapy for patients in the last months of life. He found that in the last six months, three months, and one month of life, as much chemotherapy was given for tumors that are known to be unresponsive to chemotherapy as for tumors that are responsive to chemotherapy. Other losses of benefits include: failure to ensure that the necessary information is available at the time of decision making and at the point of care; failure to help with transitions following active treatment; failure to monitor and manage symptoms, including pain; and failure to support dying patients and their families. A few health systems have reported their attempts to develop an integrated model of care—financially, organizationally, and in data management (Clive, 1997; Demers et al., 1998; Glass, 1998). Other reports include the development of disease-management models of inpatient and outpatient oncology care (Hennings et al., 1998; Piro and Doctor, 1998; Sagebiel, 1996; Uhlenhake, 1995), breast cancer centers (Frost et al., 1999; Kalton et al., 1997), psychosocial support services (McQuellon et al., 1996), support for
OCR for page 180
Building a Better Delivery System: A New Engineering/Health Care Partnership long-term cancer survivors (Hollen and Hobbie, 1995), and quality improvement teams (Frank and Cramer, 1998). A remarkable example of what can be accomplished is the use of logistical engineering in the United Kingdom for cancer services (Kerr et al., 2002; NHS Modernisation Agency, 2001; H. Bevan, personal communication, May 2001). The story began with a major comparative study that showed that survival rates in the United Kingdom were low compared to rates in the rest of Europe and the United States. The study also found that therapy was initiated at a much more advanced stage of disease than expected, which resulted in low five-year survival rates. One reason was the seven- to eight-month delay between (1) detection and (2) diagnosis and staging. Patients were also not able to get the radiation therapy they needed, even though 20 to 50 percent of the appointment slots were not used. By the time patients were seen, the plan of care was often outdated or no longer appropriate. Although the patients’ needs were predictable, they did not know what to do once they left the hospital. Further, the percentage of patients referred for abnormal exams or test results who will, in fact, have cancer can be predicted. Hence, services could be designed according to a known demand function. Using such information, the National Health Service (NHS) made improvements in cancer care services a priority. The program began with 50 teams from nine cancer networks; the program has now been expanded to all 34 networks. The project teams tested more than 4,400 changes in the first 12 months and implemented nearly 550 of them. They instituted multidisciplinary teams that meet regularly to manage the experiences of families and caregivers. They revamped services to meet patient and family needs. For example, tests that used to require three separate hospital visits are now done in one visit. As a result of this initiative, there was, on average, a 50-percent reduction in time to first appointment and a 60-percent reduction in radiology waiting times. The NHS believes the five-year cancer survival rate can be improved by 10 percent and is reengineering systems accordingly. Engineering can play a major role in accelerating improvements in the quality and efficiency of cancer care. The unique skills of practicing engineers should be applied in six major arenas of cancer care: Redesign care processes using engineering tools, such as the 80/20 rule, continuous flow, mass customization, production planning, and supply-chain manufacturing. Use information technology to make medical information and patient-specific information available when needed. The goal is to ensure that timely, accurate information is available to clinicians and patients when they need to make decisions. Redesign care to include the patient and family in decision making. Encourage the continuous acquisition of knowledge and skills by all health care workers to support multidisciplinary work. The health care workforce must have the expertise to manage complex tasks, which may require changes in training, education, and protocols and rules about which tasks are permitted. Human factors analysis, which has been used in other industries for crew resource management, shift management, ensuring patient and worker safety, and ensuring high-level, reliable performance in dynamic, high-risk settings, should be applied to the health care setting. Care should be coordinated across settings and over time using any engineering tools available. Measurement of performance and outcomes should be used to improve care. This entails measuring the results of practice and removing the distinctions between research and clinical practice environments so that all patients and patient care can increase our knowledge. REFERENCES American Cancer Society. 2002. Harlem cancer screening clinic embraces the medically underserved. ACS News Today. Available online at: http://www.cancer.org/docroot/nws/content/nws_1_1x_harlem_cancer_screening_clinic_embraces_the_medically_underserved.asp. Accessed 9/04/03. Bickell, N.A., and G.J. Young. 2001. Coordination of care for early-stage breast cancer patients. Journal of General Internal Medicine 16(11): 737–742. Bickell, N.A., A.H. Aufses, Jr., and M.R. Chassin. 2000. The quality of early-stage breast cancer care. Annals of Surgery 232(2): 220–224. Bleyer, W.A., H. Tejeda, S.B. Murphy, L.L. Robison, J.A. Ross, B.H. Pollock, R.K. Severson, O.W. Brawley, M.A. Smith, and R.S. Ungerleider. 1997. National cancer clinical trials: children have equal access; adolescents do not. Journal of Adolescent Health 21(6): 366–373. Christensen, J., and N. Akcasu. 1999. The role of the pediatric nurse practitioner in the comprehensive management of pediatric oncology patients in the inpatient setting. Journal of Pediatric Oncology Nursing 16(2): 58–67. Clive, R.E. 1997. Update from the Commission on Cancer. Topics in Health Information Management 17(3): 10–14. Demers, R.Y., R.A. Chapman, M.H. Flasch, C. Martin, B.D. McCarthy, and S. Nelson. 1998. The Henry Ford Health System. Cancer 82(10 Suppl): 2043–2046. DHHS (U.S. Department of Health and Human Services). 2001. 2001 Cancer Progress Report. NIH Publication No. 02-5045. Bethesda, Md.: National Institutes of Health. Also available online at: http://progressreport.cancer.gov. Emanuel, E. 2001. Use of Chemotherapy for Advanced Disease. Presentation at the Annual Meeting of the American Society for Clinical Oncology, San Francisco, California, May 2001. Frank, J., and M.O. Cramer. 1998. The development of an oncology services performance improvement team. Journal for Healthcare Quality 20(6): 26–32. Frost, M.H., R.D. Arvizu, S. Jayakumar, A. Schoonover, P. Novotny, and K. Zahasky. 1999. A multidisciplinary healthcare delivery model for women with breast cancer: patient satisfaction and physical and psychosocial adjustment. Oncology Nursing Forum 26(10): 1673–1680. Glass, A. 1998. Delivery of comprehensive cancer care at Kaiser Permanente. Cancer 82(10 Suppl): 2076–2080.
OCR for page 181
Building a Better Delivery System: A New Engineering/Health Care Partnership Hennings, M.N., D.S. Rosenthal, D.A. Connolly, M.N. Pollach, and M.M. Lynch. 1998. Collaborative approaches to purchasing and managing oncology services for a prepaid population. Cancer 82(10 Suppl): 2026–2034. Hollen, P.J., and W.L. Hobbie. 1995. Establishing comprehensive specialty follow-up clinics for long-term survivors of cancer: providing systematic physiological and psychosocial support. Supportive Care in Cancer 3(1): 40–44. IOM (Institute of Medicine). 1999. Ensuring the Quality of Cancer Care, M. Hewitt and J.V. Simone, eds. Washington, D.C.: National Academy Press. Joishy, S.K. 2001. The relationship between surgery and medicine in palliative care. Surgical Oncology Clinics of North America 20(1): 57–70. Kalton, A.G., M.R. Singh, D.A. August, C.M. Parin, and E.J. Othman. 1997. Using simulation to improve the operational efficiency of a multidisciplinary clinic. Journal of the Society for Health Systems 5(3): 43–62. Kerr, D., H. Bevan, B. Gowland, J. Penny, and D. Berwick. 2002. Redesigning cancer care. British Medical Journal 324(7330): 164–166. McQuellon, R.P., G.J. Hurt, and P. DeChatelet. 1996. Psychosocial care of the patient with cancer: a model for organizing services. Cancer Practice 4(6): 304–311. NCCN (National Comprehensive Cancer Network). 2001. Practice Guidelines in Oncology. Available online at: http://www.nccn.org/physician_gls/index.html. NHS Modernisation Agency. 2001. The Cancer Services Collaborative: Twelve Months On. Presented to the National Patients’ Access Team, Leicester, England, July 18, 2001. O’Connell, B., L. Kristjanson, and A. Orb. 2000. Models of integrated cancer care: a critique of the literature. Australian Health Review 23(1): 163–178. Piro, L., and J. Doctor. 1998. Managed oncology care: the disease management model. Cancer 82(10 Suppl): 2068–2075. Sagebiel, R.W. 1996. The multidisciplinary melanoma center. Surgical Clinics of North America 76(6): 1433–1439. Sateren, W.B., E.L. Trimble, J. Abrams, O. Brawley, N. Breen, L. Ford, M. McCabe, R. Kaplan, M. Smith, R. Ungerleider, and M.C. Christian. 2002. How sociodemographics, presence of oncology specialists, and hospital cancer programs affect accrual to cancer treatment trials. Journal of Clinical Oncology 20(8): 2109–2117. Taplin, S., M. Manos, J. Zapka, A. Geiger, M. Ulciskas-Yood, S. Weinman, S. Gilbert, and J. Mouchawar. 2000. Frequency of Potential Breakdowns in Screening Implementation among Women with Invasive Cervical and Late Stage Breast Cancer. Presented at the 8th Annual HMO Research Network Conference, Long Beach, California, April 9–10, 2002. Uhlenhake, R. 1995. Is patient-focused outpatient cancer care on target? Journal of Ambulatory Care Management 18(4): 32–42. Wolfe, L.C. 1993. A model system: integration of services for cancer treatment. Cancer 72(11 Suppl): 3525–3530.
OCR for page 182
Building a Better Delivery System: A New Engineering/Health Care Partnership This page intentionally left blank.
OCR for page 183
Building a Better Delivery System: A New Engineering/Health Care Partnership Patient Trajectory Risk Management Charles Denham HCC Corporation and Texas Medical Institute of Technology This paper addresses the notion of risk trajectory of individual patients and the resultant aggregate risk trajectory of the healthcare enterprise caring for populations of patients. It also describes the use of various engineering concepts applied to medicine. In the late 90’s, working with a team from the Institute for Healthcare Improvement (IHI) and Premier Inc. a group purchasing organization of 1,800 hospitals we focused our attention on medication management. The project involved collaborators from the Cleveland Clinic, Partners System, Harvard Medical School, Mayo Health System, a number of frontline hospitals and leading experts. Our goal was to identify the idealized design for medication management to reduce adverse drug events, a major cause of preventable death and disability in U.S. hospitals. To do that, we first had to identify achievable world-class performance, then the “is state” of frontline hospital performance, and finally processes and technologies that would enable us to close the gap between the two. We were surprised by our findings and gratified by the opportunities they revealed. Engineers are used to using process impact evaluations, risk analyses, and pattern recognition methods, however these are new to the practice of medicine at frontline institutions. Clearly, medicine has much to gain from engineering, and many benefits have yet to be realized. The Institute of Medicine report, Crossing the Quality Chasm (IOM, 2001), proposes that we must redesign healthcare so that it is patient centered, evidence based, and systems focused. As such we must have a much better understanding of “integrated performance”—i.e. operational, clinical, and financial processes and outcomes—of an individual patient’s care delivery through a healthcare episode. We must look at the performance/risk trajectories of common patient treatment process paths and examine the contributive impact to enterprise wide performance. Hospital administrators must step back from their traditional vertical business unit view and take into account their patient populations as they move through those vertical units so that they can recognize operational innovations that can eliminate process segment failures. The game of golf provides a powerful metaphor. The desired outcome is to deliver the ball to the hole. For a given link one golfer may take eight strokes and another might take three. Both reach the goal if the outcome measure was just “ball in hole,” however one expended more energy and time than the other. The golfer taking eight strokes has increased the risk of having mishap along the way. In a similar way, if a patient requires two or three extra days of care, the risk of having an adverse event is greatly increased due to greater exposure to the inherently dangerous hospital environment. To come up with an ideal design for medication management, we first mapped the clinical and operational processes involved in medication use. Next, we considered the products, services, and technologies involved that enable best or better practice (technologies might include process reengineering tools, for example). Then we identified their impact on the risk of adverse events and whether they closed the gap between typical performance and best achievable performance. Traditionally administrators and clinicians have been trained to define a medication error by violation of one or more of the “five rights”—the right patient, the right drug, the right time, the right dose, the right route. Such errors occur with virtually every patient admitted to hospital. Dr. David Classen a noted patient safety expert on our team demonstrated that the overlap between error and harm minimal using this definition of error—only a small fraction of harm is caused by error as defined by the “five rights.” A great number of errors do not cause harm, and more importantly a number of adverse drug events that cause death, disability, or require treatment would not normally be counted using the classical “5 rights” framework. During the idealized design process, we worked with a
OCR for page 184
Building a Better Delivery System: A New Engineering/Health Care Partnership number innovative healthcare technology suppliers; 70 to 80 percent of them were attuned to error. Few focused on harm. The deeper we explored adverse drug events it became more and more apparent that distinguishing between error and harm was critical. We focused on the most common causes of adverse drug events including transition zones between care teams and high impact intravenous infusion events. We did not ignore errors without harm, but we did not focus on them. After completing about 80 percent of a thorough, evidence-based review of integrated care and operational processes, with the guidance of a number of experts, the opportunities for mitigation started to become clear. Subsequently IHI led a number of very successful hospital collaborative initiatives using a “trigger tool” medical record review framework that helped identify adverse drug event (ADE) risk and performance gaps. We studied smart the Alaris smart infusion pumps that have now have the ability of capturing and even preventing the most serious IV adverse events, clearly a technology advance that will deliver dramatic speed to impact in reduction of ADEs. To illustrate the error-harm gap and the notions of patient trajectory and hospital risk trajectory we used the example case of anticoagulation management with our teams and collaborative groups. Anticoagulant drugs are often very poorly managed by clinicians and patients resulting in severe adverse drug events. In fact this is the area of the most common drug related malpractice claims and awards. Certain engineering concepts have great application to medicine. When engineers evaluate airplanes, they examine and discuss its performance envelope. We applied this concept to the management of anticoagulation. Warfarin is an anticoagulant drug used to manage patients. Its danger lies in the fact that the therapeutic envelope of safety relating dose to effectiveness and complications may change or shift. The patient’s diet (i.e., wine or vitamin K consumption), or liver function can shift the therapeutic window. The therapeutic envelope is always changing, posing huge risk to patients for overdose or under dose leading to clotting or bleeding disorders. Currently physicians try to manage patients undergoing anticoagulation by trying to interpolate and extrapolate the relative patterns of multiple lab values and historical factors. Application of the performance envelope delivers terrific pattern recognition opportunities. We also demonstrated the use of other aviation tools to communicate performance. For instance we created a mock up “digital dashboard,” illustrating how clinicians could recognize patterns, access relevant protocols, and in the case anticoagulation decide how to manage the patient. In collaboration with one of the nations leading anticoagulation experts we presented an example case study of a young adult admitted for treatment of a defective heart valve who experienced 11 typical and different adverse drug events, none of which was caused by a medical error (using the 5 right classification) and none of which would have been picked up by the typical methods we use to catch medical errors. Dose adjustments unique to the patient’s condition and omissions due to missed laboratory values would not typically be classified as a medication error. The patient eventually has a stroke. In this case, the potential for recognition of the risk for adverse events would have been picked up by a computerized physician order entry (CPOE), which integrates order entries with laboratory and historical information. We know from other studies that CPOE can reduce adverse events dramatically. In the future, we will have a decision-support systems that enable clinicians who are not specialists in anticoagulation to put that part of the treatment in the hands of a pharmacy team while being able to monitor potential adverse events. That is precisely what an information integrating device that pilots use called a flight director does. Flight information is provided as an input, the crew makes sure all the instrumentation is synchronized and the director follows the plan. If the workload becomes too heavy, the autopilot can be turned on. Today, 16 different types of specialists prescribe anti-coagulants; none are specialists in anticoagulation. Orthopedists, internists, and cardiologists are all administering the drug and are responsible. The risk trajectories such patients are not being managed well and adverse events such as preventable strokes and bleeding related complications are occurring in epidemic proportions. We used a mockup of the digital dashboard to study the young adult described earlier. His medical history and his recent history revealed a number of health problems that pre-disposed him to a bleeding and clotting disorder that made anticoagulation drugs extremely dangerous for him. When we asked what might have been done differently, we found that when the care data is reconstituted in a graphic it would allow us to recognize a pattern. Had the data presentation been like that presented in aircraft instrumentation we would have seen the window of safety narrowing and prevented catastrophe. Instead, we are caught by surprise driving from a view through the rear view mirror. Clinicians could be assisted by innovations that make patterns simpler to recognize. The average doctor in an intensive care unit can interpolate three or four trends. A patient on a respirator who is very ill might have could have 60 pertinent trends. Our slowest cognitive capability is in processing data, which is exactly what computers do well. Before retiring to focus full time on emerging technologies, I was a radiation oncologist with a very large practice, and I managed all of my patients all the way through therapy. I had a high volume of patients with common diseases, including colon, breast, lung, and prostate cancer. I had to navigate between the response of the tumor to radiation therapy and the response of normal tissue. I had to manage that patient through a safety window that would become narrower and narrower as we proceeded through care. As the dose was increased, the risk for a host of complications would increase
OCR for page 185
Building a Better Delivery System: A New Engineering/Health Care Partnership and continue intensify through out treatment. We knew that every treatment decision had a risk-benefit balance to it. Every patient had a unique trajectory based on historical data and how certain factors had impact as therapy progressed. These patients were managed based on tacit knowledge—we could tell when a patient was headed for trouble, we could link this to certain parameters. In working with healthcare technology suppliers, we have found that an evidence-based, patient centered, and systems performance targeted approach to “enabling” best or better practice allows innovations to be developed that improve clinical performance and reduce risk. In addition, they often deliver improved enterprise wide performance as a by-product of improved patient specific performance. If we had continuity of information with pattern recognition support we could examine the risk trajectory of patients with very complex disorders and create scenarios and real time forecasts, as we do in aviation. In the future, we might ask a medical student to use a computer model to run scenarios for a specific patient. We could graphically portray patterns and risk trajectories to assist in decision making before patients get into trouble. Is the patient’s cardiac function adequate? Will his kidneys clear everything? What-if scenarios can be run before events cascade. Engineers already provide wonderful computational support and pattern recognition solutions for many industries. These technologies will offer physicians a terrific opportunity to “think through” treatment scenarios. With an appropriate decision-support system, we could apply the lessons learned in other industries, such as aviation and aerospace, to complex medical problems. The principles of data analysis from engineering could be tremendously beneficial for health care. REFERENCE IOM (Institute of Medicine). 2001. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, D.C.: National Academy Press.
OCR for page 186
Building a Better Delivery System: A New Engineering/Health Care Partnership This page intentionally left blank.
OCR for page 187
Building a Better Delivery System: A New Engineering/Health Care Partnership Deploying Resources for an Idealized Office Practice: Access, Interactions, Reliability, and Vitality Thomas W. Nolan Associates in Process Improvement and Institute for Healthcare Improvement The goal of our initiative is to create an idealized design of clinical office practices (IDCOP) that offers the best possible solutions to the health care practice needs of our customers. When implemented, these solutions should lead a visiting patient to say, “They give me exactly the help I want (and need) exactly when I want (and need) it.” To accomplish this goal, we have to improve measures associated with: clinical outcomes; patient satisfaction; finance; and staff satisfaction. To simplify and further systematize the systems that emerge from IDCOP, we have developed a framework of four “themes” to guide the redesign processes as a whole: access, interactions, reliability, and vitality. Access. Timing is an essential component of health care. When things happen is almost as important as what happens. Of all forms of timing, patients almost certainly value most the timing of entry into the system—getting to care when the care is needed. Care in this context does not mean only encounters or visits. It means all appropriate forms of interaction, including access to information, support, dialogue, reassurance, treatment, and supplies, as well as all possible routes of delivery—not just face-to-face meetings, but also electronic, print, and other media of exchange. Interactions. Health care is fundamentally interaction. Interaction is not the price of or vehicle for care; it is the care. Those who regard health care as a list of resources—people, medications, machines, technologies, and so forth—are merely listing the “inert” ingredients that become care only when they are combined in interactions between patients and the system. The quality of care is the quality of interaction among resources, not the quality of the resources per se. Reliability. Reliability involves ensuring an exact match between knowledge and activity in the IDCOP practice. Ideally, “all and only” effective and helpful care is given. The IDCOP practice, therefore, aims always to give care that can help a patient and never to give care that harms or cannot help a patient. Reliability is the conscious attempt to avoid the defects in health care that the Institute of Medicine Roundtable on Quality summarizes as “overuse, underuse, and misuse” of care. (The Roundtable defines misuse as errors in care and threats to patient safety.) Vitality. IDCOP aims for a sustainable design. The new system would be financially viable and would provide a great workplace. In other words, the demanding performance standard is not realized at the expense of those who work in the practice and depend upon it for their livelihood. Vitality also implies renewal—continual innovation and improvement. The IDCOP practice is not a fixed, solved system; it is a learning organization with the capability, agility, resilience, and will to change over time as desires, environments, and knowledge change. Each of these themes or aspects of IDCOP requires certain activities, some familiar and some new. One of the initial steps to redesigning the system as a whole is the systematic examination of the current premises and beliefs concerning the activities performed and the people who perform them. Meeting each of the goals requires some resource deployment and scheduling. To achieve excellent access, the demand for visits and other interactions must be estimated beforehand, and capacity, for example for appointments, must be available to meet the demand. Conceiving of care as interactions between the patient and the system via multiple media means that resources must be deployed to enable these interactions. Reliability requires an exact match between knowledge and activity in the practice, knowing the activities that will meet the needs of patients and ensuring that these activities are performed in an orderly manner and at the proper time. The activities that contribute to the vitality of a practice, such as training and process redesign, might easily be put off in the face of pressing daily demands, but
OCR for page 188
Building a Better Delivery System: A New Engineering/Health Care Partnership these activities are essential. Hence, time must be scheduled for them. Besides helping with the daily deployment of resources, the development of a master schedule for the practice will facilitate the fundamental rethinking of the design of the practice. The following three tasks serve as a guide to the deployment of resources consistent with the IDCOP themes: Understand and define the work involved in caring for persons who depend on the practice. Assemble a team of people and resources to match the work. Develop a repetitive master schedule to optimize the use of resources relative to the needs of the population. Defining the work involves describing activities in the practice and then assessing them in terms of the four themes. The activities can then be adjusted to ensure that the practice has all four characteristics and the appropriate clinician matched with the work. Once the work and appropriate team have been identified, the practice can match the work to the members of the team on specific days of the week using a repetitive master schedule. REPETITIVE MASTER SCHEDULE The work of a clinical practice is varied and complex—no two patients are alike, insurance companies have different requirements, and the external environment is changing rapidly. Designing an IDCOP practice is impossible unless some sense of order is established in the midst of increasing demands and varying conditions. Developing and using a repetitive master schedule is one method of establishing order. Although the work varies, every practice has a natural rhythm—the length of time after which the work begins to repeat. Staff in a primary care practice often cite one week as the repetitive period. Up to a point, the work done in one week is similar to the work done the next week. Of course, the rhythm in a practice is also influenced by shorter periods, such as days, and longer seasonal periods that must also be taken into account. The practice must first establish the period for which a master schedule will be designed. For purposes of discussion, let’s assume the period is one week. That means that a master schedule for a “typical” week can be used with minor adjustments for any week. The definition of the repetitive period simplifies the task of deploying the resources of the practice because the schedule is built only for a short period of time. Once the period has been chosen, a master schedule can answer the questions of what work will be done, who will do it, when they will do it, and where they will do it. An IDCOP practice calls for forms of interaction in addition to one-on-one visits with the doctor. Who will be using e-mail? Who will provide chronic disease management and review registries? When will training and staff development take place? The master schedule should provide answers to these questions. The slogan for a master schedule with a period of one week is “do today’s work today.” Although there is some overlap in each day’s work, Tuesday’s work will not be exactly the same as Thursday’s. The practice may hold a group visit on Tuesday, for example, and review the chronic disease registries on Thursday. Daily work should be completed on the day it is scheduled. “Open access” requires that patients be scheduled within the master schedule cycle. Hence, practice-patient interactions are a very large component of the master schedule. Backlogs are defined as work that is not scheduled or completed within the master scheduling period. Consider a patient’s initial appointment in a behavioral health practice. Because the initial appointment requires that multiple providers see the patient during the visit, a practice may designate one morning a week for initial appointments. The “open access” philosophy requires that new patients be seen within a week. Backlogs of two or more weeks for new patients are inconsistent with the repetitive master scheduling approach. Open access and repetitive master scheduling are based on the general concept of “continuous flow,” which requires that the amount of work be predicted and resources deployed to complete the work in a specified period of time without backlogs. Continuous flow principles apply to weekly scheduling and even daily scheduling. The physician who sees a patient and completes the chart before moving on to the next patient within the specified activity cycle time is using continuous flow. Many practices already use some aspects of master scheduling. Practices with open access to visits and phone calls are well along in the development of a repetitive master schedule. For practices that wish to develop a master schedule the following steps should be considered: Implement an open access system for visiting patients. Define the care process for each of the top diagnoses to use as input to the master schedule. Include in the definition the desired time between when a patient first presents with the problem and when an effective plan of treatment is begun. List the services required to accomplish the themes and the internal processes required to support these services. Devise a master schedule of one to two weeks that addresses who, what, where, and when for the services and processes enumerated above. Use the following metrics to assess success in executing the master schedule: the degree of completion of the schedule and the reasons for not achieving it the percentage of time physicians are doing work that only they can do or that only they are legally allowed to do the time from patient presentation to treatment for the top 10 diagnoses
Representative terms from entire chapter: