NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US). Evidence-Based Medicine and the Changing Nature of Healthcare: 2007 IOM Annual Meeting Summary. Washington (DC): National Academies Press (US); 2008.

Cover of Evidence-Based Medicine and the Changing Nature of Healthcare

Evidence-Based Medicine and the Changing Nature of Healthcare: 2007 IOM Annual Meeting Summary.

Show details

4Contending with the Changes


As new and more complex medical interventions are developed and scientific knowledge about disease origins and progression continues to expand, the healthcare system will need to adopt approaches that ensure evidence generated is relevant to real-world patient populations and is incorporated effectively into clinical practice. Essential to these approaches will be new roles and responsibilities for the patients and providers at the front lines of care. Anticipated shifts in the behaviors, beliefs, and practices of these stakeholders are described in this chapter.

Because of the overwhelming amount of new evidence and information that healthcare providers must incorporate into their practices, William W. Stead suggests that the healthcare system will shift from expert-based practice, which is built around the extensive knowledge and experience of the physician, to a systems-supported practice centered on teams supported by well-defined processes and information technology (IT) tools. While both approaches rely on evidence for decision making, the difference is in how evidence is translated into action. Stead describes how this approach has been used at Vanderbilt University Medical Center (VUMC) to improve care of patients on ventilators and discusses the major barriers inherent to health care that might limit broader implementation of this approach, and possible solutions.

In his paper, Marc Boutin highlights the diversity of patients, each with differing life circumstances, cultural needs, preferences, and socioeconomic status. Broader acceptance of evidence-based medicine (EBM) will require an evidence base that appropriately accounts for these differences, and better communication to patients on the importance of best evidence in healthcare decision making. Strengthening the patient-provider relationship is also essential to ensuring the use of EBM results in the best medical outcomes and closes the quality chasm across geographic regions, treatment settings, and patients’ socioeconomic levels.


William W. Stead, M.D., and John M. Starmer, M.D.1


Most healthcare providers believe their practice is evidence-based. Their education includes the scientific basis of health and disease. They have been trained to use scientific literature to compare alternative approaches to diagnosis and treatment. They do their best to stay up-to-date through reading and conferences. Yet despite their attention to evidence, studies repeatedly show marked variability in what healthcare providers actually do in a given situation. When challenged about why they do not practice consistently, healthcare providers point out that health care is both art and science. Explicit evidence is available for only a portion of what they do.

Instead of focusing on the use or non-use of evidence, we contrast expert-based practice to a systems approach to practice. Both approaches use evidence. The difference between the approaches is the way in which the evidence is translated into practice. We provide a framework of steps for developing, using, and iteratively improving a systems approach to practice. We provide examples using VUMC’s approach to ventilator management. We conclude with implications of our experience with a systems approach to practice for healthcare workforce and infrastructure policy.

Expert-Based Practice

In expert-based practice, the focus is on the individual’s performance. The individual expert provides extensive knowledge and technical skill based on education and experience. The individual expert is expected to remember facts, assimilate data, recognize patterns, judge, and make decisions wisely. The individual expert’s opinion is valued. Disagreement among experts is expected. The result is no better than the performance of the individual expert. The individual expert is responsible for recognizing and learning from his or her mistakes.

System-Supported Practice

In system-supported practice, the focus is on the system’s performance. Teams of people, well-defined processes, and information technology tools work in concert to produce the desired result consistently. People provide compassion, pattern recognition, and judgment. Well-defined processes standardize and simplify work flow. IT tools decrease dependence on memory and force action when needed. Agreement among individuals is required. The desired result is expected every time. Each failure feeds back to support just-in-time correction or iterative adaptive design. The system of behaviors, processes, and tools makes it easy for the individual to do the right thing every time.

Figure 4-1 depicts our systems approach to practice. The left-hand circle represents cycles of iterative system development and refinement. We focus our efforts by working on one population at a time. By population, we mean every instance of the circumstance that we want to manage to a consistent outcome, such as patients on a ventilator. A patient is likely in multiple populations at once. The first system development step is selection and definition of a high-priority population to target. A population might be high priority because of risk for morbidity or mortality, such as patients with central lines who have a high incidence of nosocomial bloodstream infection. Another population might be a priority because of opportunity to reduce cost by streamlining throughput, eliminating unnecessary procedures, or using more cost-effective drugs. We try to make the definition of the population explicit. This definition consists of the environmental, clinical, or procedural characteristics that collectively frame the circumstance we want to manage consistently. We restrict the definition to characteristics that are present in our information systems or can reasonably be obtained through supplemental manual audit. Once we agree on such a definition, we can identify each member or instance of the population across our enterprise.

FIGURE 4-1. The systems approach to practice joins system development to system-supported practice.


The systems approach to practice joins system development to system-supported practice. The left-hand circle represents the four steps in each iterative cycle of system development. The right-hand (more...)

The second system development step involves gathering the evidence related to that population into a common fact base. We look for three types of evidence. We search the literature for clinical trials related to the target population. We obtain consensus practice guidelines related to the population from the literature and from sources such as the University Hospital Consortium (UHC) and quality improvement organizations. We obtain the pathways, protocols, and order sets from our practice groups that relate to the target population. We reduce this information into a table with a row for each explicit practice related to the target population and a column showing the recommendations for each practice for each of the above sources. This table highlights variance in available evidence. We then charge a core team of subject matter experts to develop a “straw person bundle” for use across the enterprise. The bundle is a set of standardized practices with specific process steps and measures of performance for each practice. The core team also drafts one or more overarching idealized processes that would result in consistent execution of the bundle. The core team is not a representative consensus-forming body. Instead, it is as small as possible while bringing critical information to the table across medicine, nursing, ancillaries, etc. These individuals work as a team, viewing each member as a partner in the solution, rather than as representative of an area. We complete the common fact base by documenting the performance of our units or practice groups against the “straw person” set of standardized practices through either electronic or manual spot audit.

The third system development step makes the jump from the shared fact base to cross-enterprise agreement. This agreement includes a set of standard practices; metrics to assess performance on each practice; explicit definitions for both practices and metrics; simplified standard work flows to implement the practices; IT tools; staffing and education; and implementation time line. When the needed agreement cuts across disciplines and care settings, we take time out for a day of cross-enterprise design. We bring together executives that will have to approve change in policy or resource allocations, medical and nursing leadership of each affected practice group or unit, representatives of affected ancillaries and subject matter experts from clinical areas, quality improvement process, informatics, and finance. After sharing the fact base, we identify points of disagreement and use breakouts to work alternative solution designs. We constrain the design by restricting suggestions to ideas that can be implemented across the enterprise in 6 weeks to 3 months. Longer-term suggestions are noted and parked for future consideration.

The fourth system development step involves monitoring performance at the population level and iteratively refining the system of practice as needed. Sentinel events are monitored as early indications of unexpected problems during the rollout of changes. Metrics provide an early indicator of where we are and are not achieving reproducible performance on the set of standard practices, and we adapt education, tools, or process as needed.

The rapid cycle iterative nature of the system development steps cannot be overemphasized. For example, an initial population definition focuses an initial search for evidence. Our review of the evidence may suggest modification of the population definition followed by a revised search for evidence. Similarly, the bundle of practices developed from the evidence guides the initial audit of our performance. Our review of the audit may feed back suggestions for refinement in the bundle.

The right-hand circle of Figure 4-1 depicts the related iterative cycles of system-supported practice in the care of an individual patient. System-supported practice is not cookbook medicine. At the start of each cycle, the clinical team assesses the patient, develops or refines the plan of evaluation and care, and orders the next actions. As a patient matches the definition for one or more of the populations for which we have developed a systems approach, the clinical team is alerted and prompted with orders appropriate to implement the related standard practices. As they round, they are shown the status of each of their patients on each of the standard practices (Stead et al., In press). As they start the next cycle of system-supported care, they take corrective action as needed in addition to updating the plan and orders to reflect new information and trends.

To this point, our system development steps yield reliable execution on standard practices reflecting the literature and consensus. These practices may or may not actually be the right thing to do. Even if they are, they will get out of date as the biomedical knowledge base changes. In time, we expect our systems approach to practice to become self-correcting as we add indicators of good and bad clinical outcomes to the metrics of performance on the standard practices for each patient. Whenever the clinical team elects to vary from the standard practice, in effect it creates an experimental group. If outcomes of that group appear better, or even no worse, we will be able to do targeted clinical trials leading to change in the set of standard practices if appropriate. In addition, whenever clinical outcomes for a population deteriorate or diverge in the wrong direction from external benchmarks, we will know to reassess the standard set of practices for that population.

Example of Expert Management of System-Supported Practice

VUMC selected the ventilator management bundle as one of the test cases for our systems approach to practice. The following examples from that work are presented to highlight the gap between the available evidence and the set of standard practices needed to consistently produce the desired result. We also provide examples of decisions by the experts as they manage the cycles of system development and apply judgment within the resulting system-supported practice. We show how IT can decrease dependence on memory and provide a forcing function to help close the gap between intent and execution. Throughout the examples, we provide an indication of the number of people involved and the elapsed time.

We selected the ventilator management bundle as a test case for our systems approach to practice because of evidence of high morbidity and mortality associated with ventilator-acquired pneumonia (Bueno-Cavanillas et al., 1994; Girou et al., 1998); evidence linking specific practices to reduction in risk for (Ibrahim et al., 2001) or incidence of (Doebbeling et al., 1992; Thompson, 1994) ventilator-acquired pneumonia; and use of ventilators by several specialties in many units across our enterprise. Past VUMC initiatives had focused on tracking ventilator-acquired pneumonia and unit-specific practices to reduce the incidence. Our approach this time was different in that we started with three executive-level agreements. We would not focus on ventilator-acquired pneumonia. Instead, we would focus on cross-enterprise agreement on a bundle of well-defined standard practices for ventilator management, on how we would measure performance, and on the processes we would use to quickly achieve consistent performance house-wide. These decisions sidestepped pitfalls such as clinical arguments about what does or does not constitute ventilator-acquired pneumonia and the tendency to say why a unit is unique instead of what units have in common. The constrained time horizon forced people to think of simple solutions instead of requesting complicated support systems to get better results without having to change what people do.

In this test case, the definition of the population was straightforward, a VUMC inpatient on a ventilator. Although aspects of ventilator management involve multiple information systems, nurse charting provides a single source for an up-to-date indicator that a patient is currently on a ventilator. We made the executive decision that we were ready to launch the effort in mid-December 2006 and assembled the core team just after the holidays. Since this effort was a cross-enterprise test case, the initial core team included the corporate strategy and nursing officers and the chief executive officer of the adult hospital. The chair of the Critical Care Committee, a physician, and the nurse director of the Surgical Intensive Care Unit, together with leaders from Clinical Improvement and Informatics, completed the team. Over the course of January, that team oversaw the compilation of the common fact base, obtained initial agreement of the medical directors of the nine intensive care units on the set of practices to include in the bundle, and identified the people to include in the cross-enterprise design day. The key decisions during this phase of the system development work included affirmation of the focus on the bundle of standard practices instead of the incidence of ventilator-acquired pneumonia; selection of goal-directed sedation monitored by the Richmond Agitation Sedation Score (RASS) (Thompson, 1994) as an alternative to sedation interruption since the latter is inappropriate for certain patients such as those with extreme burns; and preference for house-wide implementation of the bundle instead of unit by unit.

On the last Saturday in January 2007, 45 individuals from across VUMC participated in the cross-enterprise ventilator bundle design day. This group included medical and nursing leadership from each unit, postgraduate fellows, front-line nurses, pharmacy, respiratory therapy, infection control, nurse educators, informatics, evidence-based order set development, decision support and order entry, clinical documentation, business analytics, process reengineering, process audit, chief quality officer, and executives from the core team. Their objectives included a common understanding of the fact base; refined agreement on the set of practices; specific process steps and measures for each practice; and identification of the IT tools, education, and staffing needed for consistent execution. This design work was constrained to solutions that might be implemented house-wide by mid-March. We were able to agree on cross-enterprise practice standards by decomposing high-level guidelines into components and agreeing on an approach to each component that is both supported by evidence and practical in our environment. For example, we replaced the UHC-recommended standard of oral care, defined to include everything from teeth brushing to supra- and subglottic suctioning, with three VUMC standard practices (oral swabs, teeth brushing, and hypopharyngeal suctioning). The more granular approach permits more focused accountability, performance measurement, and refinement over time. In addition, when the standard practice should change with patient condition, we made the criteria for branch points explicit. Practices requiring patient-specific variation ranged from stress ulcer prevention to goal-directed sedation. We agreed on which team role would be responsible for specific actions. For example, the physician should order and reassess the target RASS, and the nurse should assess the patient’s condition against that target. We identified two ways our IT tools could support the clinical teams.

We would use a modular order set (Figure 4-2) to present the bundle of standard practices, together with definitions or patient-specific criteria directly in clinical work flow, and use exit checks for reminders if something was missed.

FIGURE 4-2. (a) Ventilator management order set with a module for each element of the VUMC bundle of standard practices.


(a) Ventilator management order set with a module for each element of the VUMC bundle of standard practices. (b) Expansion on selection of stress ulcer prophylaxis. (c) Expansion on selection of (more...)

In addition, we would create a process control dashboard, as illustrated in Figure 4-3, with a line for each ventilator patient on a unit and a column for each element of the bundle, with a red, yellow, or green (gray scale in figure) square to indicate the status of the patient for that element.

FIGURE 4-3. Process control dashboard.


Process control dashboard.

Finally, we identified the teaching materials needed to support the change to the bundle of standard practices.

Over the course of February and March, work proceeded according to the time line from the design day. Since the cross-enterprise agreements were in place, the executives dropped off the core team to let the work proceed close to the action. Order sets were revised and education materials developed and distributed. However, we did not get traction until the process control dashboard was available in mid-May. At that point, any members of the team could see where action was needed as they walked onto the unit. As people began to question the many red squares on their unit, the core team was able to decide if the problem reflected a poorly defined standard practice, education, the documentation used to derive the status of a patient relative to a practice, or the algorithm used to calculate whether the status was acceptable (green), trending out of control (yellow), or unacceptable (red). By early September we felt we had reached a point of face validity and decided to launch a targeted education effort to close additional performance gaps.

All of our work to date has involved starting from the evidence base and developing the agreements and infrastructure to achieve consistent performance on standard practices. The next step will add outcomes such as time on mechanical ventilation, unplanned extubations, failed extubations, and complications (pneumonia, stress ulcer, and deep venous thrombosis) to the measures of process performance. This outcome feedback will in time provide the evidence to guide continued refinement of the standard practices.

Implications for Healthcare Workforce and Infrastructure Policy

The demise of expert-based practice is inevitable. The complexity of biomedical information and technology will increasingly overwhelm an individual expert’s cognitive capacity. Specialization is not an answer because of the accompanying fragmentation. Fragmentation is incompatible with the personalization of care that is becoming possible with progress in genomics and systems biology. Even if its demise was not inevitable, we would want to move beyond expert-based practice. Other industries have shown that a standard process is the key to consistently producing the desired result. There is no reason to believe that health care can be an exception to this rule. A process that varies on a case-by-case basis according to the opinion of individual experts will not consistently produce the desired outcome.

The move beyond expert-based practice is not straightforward. Health care differs from other industries in three ways that make transfer of approaches to standardization difficult. First, the manufacturing plants or services of other industries handle fewer inputs and outputs than their counterparts in health care. For example, a microchip fabricating plant has limited inputs that are translated into limited outputs through a limited number of manufacturing processes. Most healthcare facilities and services handle much greater variety. For example, an emergency department handles all comers, even if only to stabilize and transfer patients beyond their capability or to treat and return to primary care those patients whose problems are non-emergent. Highly specialized healthcare facilities have achieved consistent performance by limiting services to a few related clinical conditions and mimicking manufacturing by standardizing the complete process end-to-end. End-to-end standardization works when handling many instances of the same clinical condition, one after another. It does not scale up to handle a variety of clinical conditions at once. How might health care consistently produce the desired result in the face of this clinical variety?

Second, most other industries deal with physical systems while most of health care deals with biological systems. Each instance of a physical system is identical, produced from the same blueprint and behaving consistently according to the laws of physics. Variation is evidence of an error in the manufacturing process. To continue the analogy of a microprocessor plant, if a variation occurs, the error is identified, the process is corrected, and the variants are discarded. In contrast, biological systems are inherently variable. They evolve through random change in DNA sequence and survival of the fittest. An individual’s environment and behavior affect his or her characteristics. Because of this variability, two individuals might present with the same condition, yet need different treatment. For example, the most effective drug might be safe for many, but hurt a few. This risk may or may not be known. Even if it is known, there may or may not be a way of testing individuals to see which group they are in. In addition, individual patients may place different values on the alternative outcomes. One might value cure enough to accept a significant risk while another might prefer to continue to cope with the illness rather than take the risk of treatment. How might health care consistently produce the desired result while accommodating biological variation, uncertainty, and differing value systems?

Third, other industries are able to isolate change and stage its introduction into routine production more systematically than health care. Model development and simulation minimize the need for production trial and error. Change can be isolated in major steps. For example if a new generation of chip becomes possible, the microprocessor factory can shut down and completely retool to accommodate the changes. The rate of discovery in the biological sciences and the rate of introduction of new healthcare technology continue to increase. Yet new approaches are tested in production healthcare settings. Many of the changes are incremental, changing part of an approach to diagnosis or treatment. Many such changes occur in parallel. How might health care consistently produce the desired result while accommodating both experimentation and rapid change?

If health care did not differ from other industries, we could move beyond expert-based practice by agreeing on a standard practice for each condition and its use by all healthcare providers. Given the three major differences outlined above, such a simplistic solution cannot be expected to work. How then can health care achieve consistent performance, accommodating variety in clinical problems handled, variability in biology and values, and the rate of change in biomedical knowledge? We suggest that the answer involves standardization around a systems approach to practice, not around specific practices. Continuous system development and refinement through iterative cycles of the system development steps might yield local standard practices, consistent with global knowledge yet adapted to local resources and capabilities, changing evidence, and system performance. The linked cycles of system-supported practice permit flexing of standard practice for individual patients based upon expert judgment, but under the control of monitors that can warn of problematic trends in real time. Data reflecting the improvement or deterioration resulting from such flexing in turn provide evidence at the local level. Global correlation of local lessons in turn might feed back into the collective evidence base. Simply put, we still need the experts. Instead of spending the bulk of their time managing each individual patient as experiments with an n of one, they spend most of their time developing and iteratively refining the system of practice for their organization and working within the resulting system-supported practice. In both modes, whenever explicit evidence does not provide the next step, they make an expert judgment. In contrast to expert-based practice, this judgment takes place within a systems approach that turns the decision and the resulting outcome into information to guide the next iteration.

If correct, our suggestion to standardize around a systems approach to practice instead of around specific practices for specific patient conditions has three implications for healthcare workforce and infrastructure policy. First, we need to communicate more clearly to policy makers and payers the characteristics of health care that make moving beyond expert-based practice challenging. Without this understanding they will continue to ask for and pay for changes that are unlikely to produce the desired result. Similarly we need to help healthcare providers appreciate the synergy of a systems approach and the expert. The systems approach provides the context and feedback for the expert. The systems approach does not replace or devalue the expert.

Second, we should call for health services and biomedical informatics research into techniques and technologies to support local development and iterative refinement of systems approaches to practice. For example, we might test approaches permitting “mass customization” of standard practices. If we try to define a guideline at such a high level that anyone can use it, many details must be left to expert interpretation. Instead we might define modules or components that are small and targeted enough to gain agreement on one approach. Local flexibility might then be achieved by mixing and matching components.

Third, we should call for direct payment to clinical and process experts for their work in the four system development steps. Since this work impacts all patients in their system, we can argue for payment at a multiple of payment for work with individual patients. Similarly, we should work to focus payment for work with individuals on the steps that require an expert—applying judgment within the system-supported practice or exerting technical skill. In parallel we could deemphasize payment for time spent working around the non-system. Collectively these changes would create strong economic incentives toward a systems approach to practice while highlighting a role for the expert that will stand the test of time.


Marc Boutin, Executive Vice President, National Health Council

To begin, I’d like to give you a simple illustration of one of the challenges of looking at “the patient perspective.” Imagine that you have just received a diagnosis of acute lymphocytic leukemia, a type of leukemia that progresses very quickly. Treatment can range from chemotherapy to radiation to a bone marrow transplant.

In one scenario, you are a 38-year-old parent who has three children at home under the age of 12. In another scenario, you are a 65-year-old individual who has recently retired from a career, with a husband or wife of 40 years who has also recently retired, and the two of you are looking forward to spending more time visiting your two grown children and three grandchildren. In yet another scenario, you are an 86-year-old widower with three children in their 60s and eight grandchildren.

Each of these patients has the exact same medical diagnosis on the surface, yet every person’s circumstance is different, illustrating that a key challenge is to develop the evidence base that acknowledges that, even with identical diagnoses, a patient’s life stage, underlying health, social support, attitudes about health and illness, faith, culture, and other factors will greatly influence what is for each individual “appropriate treatment.”

We’ve heard much about EBM from the point of view of many health-care stakeholders, but what about the people the healthcare system is supposed to serve? It seems that we have an underlying assumption that, of course, all these parties exist to serve the patient and have the patients’ best interest at heart, but does it really work that way? Is it possible for us to build an evidence base that takes into account the unique needs of each patient, delivering and ensuring the “right” health care for each person?

We know that when well used, in a strong provider-patient relationship, EBM can be a powerful tool to ensure the best possible medical outcome. EBM can indeed help close the quality chasm across geographic regions, treatment settings, and socioeconomic levels of patients. It can help us use resources where they are most effective. The challenge, however, is to balance our nation’s urgent need to ensure quality care and use resources wisely, with the understanding that different patients react differently to different treatments and, just as importantly, have different priorities and personal values.

At the National Health Council (NHC) we frequently hear from patients whose chronic conditions require ongoing treatment to maintain their quality of life and enable them to remain productive members of society. The NHC has a broad and diverse membership, but representing the needs of patients is our primary focus. We have heard from many of our members that so-called EBM has been used to deny coverage to Medicaid patients in several states for treatments including asthma, epilepsy, and depression. This short-sighted view may save money for the payer in the near future, but it often later results in costly emergency room visits and hospitalizations, not to mention physical and/or emotional suffering for the patient, often accompanied by financial loss, all of which might have been prevented—or certainly lessened. We have all heard of similar cases in which the precepts of EBM have been distorted to look at short-term cost efficiency as the primary criterion.

If EBM is to be implemented systematically through a variety of mechanisms, it must be structured with the realization that what works for 80 percent of patients may actually cause harm to or be inappropriate for the other 20 percent. In other words, as we embrace an epidemiological view and use public health decision models, we should also remember and embrace the promise of personalized medicine. In the patient-centered world of personalized medicine, we allow individual patient data, in the hands of an individual health professional, to be given equal standing with aggregated public health data: as the IOM Roundtable has stated, “to account appropriately for individual variation in patient needs.” That is our ideal.

We are encouraged and excited that many in the healthcare industry are coming together to create a healthcare system that is more consistently effective, safe, efficient, and affordable. Yet, as is often the case, many of these efforts have not really focused on the needs of the patient, or even on the simple concept that engaging patients more fully in their care can directly improve medical outcomes.

There are additional factors we must keep in mind as we consider how to go forward with EBM. One is that the quality of the evidence base is often not consistent—that is, some evidence is based on large, double-blind studies over long periods of time, while other research put forth as “evidence” is based on very small groups of as few as 20 patients in very short time frames. Also, of course we all can remember research results touted as strong evidence that were later discredited when new, more robust research was conducted. So we must remember that “all evidence is not equal.”

Another factor to consider, which may be harder to grasp, is that if patients do not perceive a problem, they will not utilize the so-called solution to that problem. They may have many complaints about the way they receive health care, and we have all heard many of them, but the NHC’s research among patients repeatedly shows that they do not think quality—or more specifically, lack of adherence to evidence-based guidelines—is the problem. So, if we want patients to be accepting of the concept of EBM, we must be willing to explain it and convince them it is something they need and something that will improve their health care and their health and well-being.

Without true patient engagement and clear and honest communication about EBM, it is likely that many—maybe even most—patients will perceive that “the system” is out to limit their access to the care they need. And it is it likely to be much more complicated and expensive to implement than it needs to be. We believe the key is to protect and preserve the patient-provider relationship, so that it is on equal footing with public health and epidemiological evidence. The NHC wants to see us work together to address the needs of payers, industry, providers, and patients and their families alike.


  1. Bueno-Cavanillas A, Delgado-Rodriguez M, Lopez-Luque A, Schaffino-Cano S, Galvez-Vargas R. Influence of nosocomial infection on mortality rate in an intensive care unit. Critical Care Medicine. 1994;22(1):55–60. [PubMed: 8124975]
  2. Doebbeling BN, Stanley GL, Sheetz CT, Pfaller MA, Houston AK, Annis L, Li N, Wenzel RP. Comparative efficacy of alternative hand-washing agents in reducing nosocomial infections in intensive care units. New England Journal of Medicine. 1992;327(2):88–93. [PubMed: 1285746]
  3. Girou E, Stephan F, Novara A, Safar M, Fagon JY. Risk factors and outcome of nosocomial infections: Results of a matched case-control study of ICU patients. American Journal of Respiratory and Critical Care Medicine. 1998;157(4 Pt 1):1151–1158. [PubMed: 9563733]
  4. Ibrahim EH, Tracy L, Hill C, Fraser VJ, Kollef MH. The occurrence of ventilator-associated pneumonia in a community hospital: Risk factors and clinical outcomes. Chest. 2001;120(2):555–561. [PubMed: 11502658]
  5. Stead WW, Patel N, Starmer JM. Closing the loop in practice to ensure the desired performance. Transactions of the American Clinical and Climatological Association. 2008 In press. [PMC free article: PMC2394683] [PubMed: 18596845]
  6. Thompson R. Prevention of nosocomial pneumonia. Medical Clinics of North America. 1994;78(5):1185–1198. [PubMed: 8078375]



This paper presents ideas developed through VUMC’s efforts at the intersection of quality improvement, evidence-based medicine, and informatics. C. Wright Pinson has provided executive oversight for quality, and Nancy Lorenzi has facilitated the informatics components of quality. Larry Goldberg and Marilyn Dubree provided executive leadership for the ventilator management initiative. Lee Parmley and the Critical Care Committee provided medical direction for ventilator management. Devin Carr prototyped change in nursing practice for ventilator management. John Bingham and the Center for Clinical Improvement supported process mapping and performance audits. John Doulis and the Informatics Center developed information technology tools.

Copyright © 2008, National Academy of Sciences.
Bookshelf ID: NBK52820


Related information

  • PMC
    PubMed Central citations
  • PubMed
    Links to PubMed

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...