Civetta, Taylor, & Kirby's: Critical Care, 4th Edition

Section I - Introduction/General Concepts

Chapter 8 - Clinical Decision Making

 

Jonathan B. Cohen

Jonathan D. Dreier

Decision making is a complex process that involves not only interpersonal aspects but influences from recognized and unrecognized outside sources. The psychology behind decision making seems to be just as important as the quality of the evidence the decisions should be based on. In this chapter, we discuss the factors that influence the clinician's decision-making process.

Identification of the Problem

For a physician to solve a problem, it must first be correctly and succinctly identified. The identification of the problem relies on gathering data from the medical history and physical exam, laboratory, and radiographic and other diagnostic testing. From this, a problem is identified and a question is posed. This question may relate to either diagnosis (i.e., in the patient with electrocardiogram [ECG] changes and complaints of chest pain, has this patient suffered a myocardial infarction?) or treatment (i.e., in this patient who has unequivocally suffered an ischemic stroke, what is the best treatment for him or her?). Figure 8.1 outlines the scope of what physicians attempt to accomplish when confronted with a clinical problem.

Clinical Knowledge

After the appropriate diagnostic or therapeutic question has been raised, the process for determining the solution is undertaken. The information required by the clinician to solve a particular problem has many different origins. Some of this information comes from our own requisite knowledge and is termed background knowledgeForeground knowledge, in contrast, is obtained through the analysis of clinical investigations and research.

It is unarguable that medical training is a lifelong process. As training is begun, the vast majority of our total knowledge of how to solve a problem is composed of background knowledge. This background knowledge consists of material retained from reading pathophysiology textbooks and attending lectures and that taught to us directly by our professors. The personal experiences and biases of those who write the texts, give the lectures, and lead clinical teaching rounds enter into, and become part of, this background knowledge.

Foreground knowledge is much more specific than background knowledge. The information is typically much more focused and answers a very particular question. This type of knowledge increases during a physician's career, as research is conducted and reading systematic reviews replaces the reading of standard textbooks. Evidence-based medicine (EBM) makes up the largest component of foreground knowledge and has been cited as the “major revolution” in foreground knowledge (1). EBM has been described as the integration of research, clinical expertise, and patient values (2).

At varying times during her or his career, the physician innately has varying levels of both background and foreground knowledge (Fig. 8.2) (2). In practice, both types are necessary to answer a clinical question, but the acquisition of knowledge is simply not enough. Application of the knowledge gained is necessary to formulate a clinical decision (Fig. 8.3).

Analysis and Application

Most clinical scenarios with which the physician is confronted are something that she or he has previously seen. The clinician compares the history and physical examination along with preliminary laboratory data to the previous clinical experiences and makes a diagnosis. This type of process is most frequently referred to as pattern recognition. The benefits to pattern recognition are that it is rapid, efficient, and usually correct. It allows the clinician to make a diagnosis without the prolonged considerations of differential diagnosis (3). Typically, these patterns take the form of illness scripts. An illness script is a combination of textbook knowledge of a disease as well as the way that it has manifested itself to the clinician based on past experience (4). The primary fault with pattern recognition is the same as its primary benefit: It is very unsophisticated. It has the potential to lure the clinician into the simplistic realm of pattern recognition when the medical problem itself may be very complex (5).

When a physician is confronted with a situation not previously experienced—that is, no pattern or illness script exists—analytic reasoning is needed to reach a proper diagnosis. Analytic reasoning is more complex and time consuming. It involves the asking of several questions that allow the clinician to reach a more thoughtful conclusion. How much a physician relies on pattern recognition and on analytic reasoning to arrive at a diagnosis depends on the physician's experience. A seasoned physician has a larger database of patterns from which to recognize a disorder as compared to, for example, a medical resident. Physicians at tertiary-care facilities have different databases of patterns as compared to country physicians. Nonetheless, when the physician arrives at a diagnosis, a new pattern is generated that may be used in the future (Fig. 8.4).

On arrival at a diagnosis, the clinician frequently reflects on it. He or she reconfirms that the diagnosis accurately fits with the available information from the history, physical examination, laboratory data, and other diagnostic studies. She or he decides that other diagnoses from the differential diagnostic list have been reasonably excluded as likely causes. The principle of Occam's razor is often taught in clinical diagnosis classes in medical school. Paraphrased, it states that “All other things being equal, the simplest solution is the best.” The premise of the Occam's razor principle is important in that a diagnostic theory that introduces the smallest number of uncertainties is likely to be most valid. Although the concept of Occam's razor is elegant and attractive, much like pattern recognition, it can make complex situations too simplistic and should be used with caution. Both Hickam's dictum and Saint's triad (6) have attempted to issue cautionary warnings to clinicians about the failure to consider concomitant diseases. Certainly, a patient is far more likely to have several common diseases, rather than one rare disease, to explain a group of symptoms. As populations age, the likelihood of patients having multiple medical conditions increases, and the likelihood of having more than one condition that explain a particular set of symptoms also increases.

001104

Figure 8.1. Clinical decision making in the diagnosis and therapy of a disease.

000165

Figure 8.2. Changes in foreground and background knowledge as clinical experience increases. (From: Sackett DL, Straus SE, Richardson WS, et al. Evidence-based Medicine. How to Practice and Teach EBM. 2nd ed. Edinburgh, Scotland: Churchill Livingstone; 2000, with permission.)

Tools for Helping Clinicians Make Decisions

Heuristics

The simplest definition of a heuristic is that it represents a rule of thumb. The use of heuristics typically leads to a rapid and efficient solution to a problem (7,8,9). Some of the more popular heuristics in medicine are “treat the patient and not the number,” and “when you hear hoofbeats, think horses, not zebras” (10). Evidence shows that heuristics are used extensively by experienced physicians (11). The use of heuristics is fraught with bias, however. The most commonly occurring biases are discussed below.

Anchoring or Focalism

Clinicians start with an implicit reference point, the most likely in the list of differential diagnoses or anchor; they then make adjustments to its likelihood of being the most correct diagnosis based on further data. This heuristic describes the common human tendency to rely too heavily—or anchor—on one piece of information when making clinical decisions. In other words, the clinician is reluctant to discard the initial diagnosis despite mounting evidence that refutes it.

Availability

In this case, the physician bases his or her prediction of the likelihood of a patient having a disease on how easily an example can be brought to mind. For example, a physician assigns a diagnosis to a patient based on the diagnosis that a recent patient had with the same set of symptoms.

Denial

If the outcome or diagnosis is too upsetting, the clinician may rate the likelihood of the patient having the diagnosis as less likely than what its true prevalence would be.

 

000168

Figure 8.3. The steps in clinical decision making.

Representativeness

This bias results when commonality between patients with similar symptoms is assumed. In this heuristic, the likelihood that a patient has a particular disease because she or he matches a pattern or illness script increases even though the disease itself may be infrequent. Several other heuristics have been postulated (Table 8.1). The above are the most frequently cited in psychological literature.

Standards and Guidelines

When issued by an officiating agency, a standard is a rule. The American Society of Anesthesiologists (ASA) considers a standard to be the minimum requirements for clinical practice. Standards are generally accepted principles for patient management (12) and may be modified only under extreme circumstances. The ASA has developed standards in three areas:

000947

Figure 8.4. Pattern recognition and analytic reasoning in the development of a clinical diagnosis.

1.   Preanesthesia care

2.   Postanesthesia care

3.   Basic anesthetic monitoring

A standard of care is uniform throughout medical practice but case specific and time specific, and thus necessarily will change over time (13). The American College of Medical Quality states that medical standards reflect both the art (consensus of opinion of clinical judgment) and science (published peer-reviewed literature) of medicine (13). Deviations from standards result in both underuse and overuse of medical resources. Standards are sometimes also referred to as standard operating procedures.

Standard operating procedures (SOPs), more specifically, are detailed written instructions to achieve uniformity of the performance of a specific function (14). A clinical guideline, sometimes called a medical guideline or clinical protocol, is a systematically developed statement designed to assist in clinician and patient decisions about appropriate health care for specific clinical circumstances (15). The content of a guideline contains recommendations that are evidence based and obtained from systematic reviews of the medical literature. The purposes of clinical guidelines are noted in Table 8.2.

Table 8.1 Lesser-known Types of Heuristics

Affect heuristics
Effort heuristic
Familiarity heuristic
Fluency heuristic
Recognition heuristic

Some argue that the term clinical protocol should be reserved for a more detailed and specific set of steps for the management of a single condition (16,17). Clinical guidelines are somewhat more flexible than standards. Deviation from guidelines to accommodate differences between patients is acceptable, whereas a standard (or standard operating procedure) should be applied uniformly to all patients. Certain characteristics have been identified that contribute to guideline use:

1.   Inclusion of specific recommendations

2.   Sufficient supporting evidence

3.   A clear structure

4.   An attractive layout (18)

Perhaps the disease process that has received the most significant amount of attention in terms of guideline development has been sepsis. The Surviving Sepsis Campaign has just released its 2008 guidelines for the management of severe sepsis and septic shock (19). Sepsis is a devastatingly common disease with a high mortality. The speed and appropriateness of therapy in the initial hours after presentation are likely to influence outcome (19). For these reasons, the development of easy-to-follow guidelines for clinicians, which summarize best practices, is very important and would be expected to have an effect on the morbidity and mortality of those suffering from sepsis. At the beginning of this important paper, the authors of the guidelines address certain issues. The authors are very careful to state that their recommendations “…cannot replace clinician's decision-making capability when he or she is provided with a patient's unique set of clinical variables.” The authors warn that local resource limitation may prevent the accomplishment of some of their recommendations. They also state that a strong recommendation does not imply standard of care. They provide insight into how differences in opinions involving evidence interpretation, wording of the guidelines, and strength of recommendations were solved. In short, valuable insight into the complexities and limitations faced while developing guidelines can be gathered from reading the front matter of this paper. The development of international guidelines on the management of a disease process as complex as sepsis is an onerous task, and the amount of time and human effort that goes into such a project is significant. The benefit of the guidelines and the impetus to continuously revise and update the guidelines need to be measured by the success that implementing the guidelines has on decreasing morbidity and mortality. At this time, several studies show favorable outcomes to the implementation of the Surviving Sepsis Campaign guidelines (20,21). One study that found no statistically significant decrease in mortality was inadequately powered to do so (22).

Table 8.2 Purposes of Developing Clinical Guidelines

To describe appropriate care based on the best available scientific evidence and broad consensus
To reduce inappropriate variation in practice
To provide a more rational basis for referral
To provide a focus for continuing education
To promote efficient use of resources
To act as a focus for quality control, including audit
To highlight shortcomings of the existing literature and suggest appropriate future research

From http://www.openclinical.org/guidelines.html#fieldandlohr.

Algorithms

An algorithm is a guideline that has been placed into a flowchartlike format. This allows for a linear approach to clinical information. At various points (nodes) during the algorithm, input from the clinician is required in the form of observations to be made, decisions to be considered, and actions to be taken. Answers at each of the decision-making points determines the further development of the diagnosis. The goal at each branching point is to further classify the disease state of the patient. As the algorithm progresses, the testing to confirm the diagnosis typically gets more complex, invasive, and costly. An example of this may be the diagnostic evaluation of chest pain. The initial evaluation begins with a thorough history and physical exam, and then may progress through electrocardiography, evaluation of cardiac enzymes, echocardiography, and, finally, coronary angiography.

Treatment algorithms are typically less complex than diagnostic algorithms since they are necessarily more focused to a single diagnosis. For example, the standard treatment of a venous thrombosis is systemic anticoagulation. The only branch point in this algorithm is if anticoagulation is contraindicated in a particular patient, such as would be the case if the patient had suffered a recent intracranial hemorrhage, in which case placement of an inferior vena cava filter might be considered more appropriate.

Typically, the diagnosis and treatment of disease are intertwined. These algorithms can become very complex but represent what occurs in daily clinical practice. Algorithms that contain pathways for diagnosis and treatment are referred to as management algorithms (23). Much like diagnostic algorithms, they classify patients into groups who may benefit from a range of broad diagnostic and therapeutic goals. An example of a management algorithm is the series of algorithms that constitute advanced cardiac life support.

Benefits of using algorithms include convenience, accessibility, and ease of use. Some studies have shown that algorithm use has resulted in faster learning, higher retention, and better compliance with established practice standards than standard prose text (24,25,26,27). In addition, algorithms form the basis for the programming behind computer-assisted decision making.

Algorithms also have been criticized (23). One criticism is that the format of the algorithm is too rigid. Patients do not always present with concrete signs and symptoms. In addition, patients are variable in their personal preferences to modalities of treatments. The Agency for Health Care Policy and Research (AHCPR) clinical guideline program attempts to address this latter limitation by inserting branching points in algorithms that recognize the importance of, and allow for, patient preference in decision making.

Another criticism challenges the clinical validity of the algorithms used in practice; however, this challenge is not unique to algorithms and also is a valid criticism for the use of some guidelines. Hardon (23) recommends the annotation of the nodes in the algorithm with links to the literature that, in turn, validates the basis of the algorithm's recommendations. This would allow the clinician to further research different points of the algorithm, allowing for more precise definitions, additional clinical detail, and identification of important gaps in the literature.

Algorithms may lack specificity. For example, a node in an algorithm may state, “Obtain cardiac output measurement.” Cardiac output obviously may be obtained from several methods of variable invasiveness. This may reflect a lack of consensus on the best method, or the best method under a particular set of circumstances, for determination of cardiac output. Nodes involving the gathering of information from the patient (i.e., the quality of pain) or other subjective information (i.e., the level of agitation of a patient) may result in user bias. Conversely, algorithms may be too specific. It is easy to imagine how a management algorithm meant to address a symptom, such as chest pain, can develop into a flowchart with over 100 nodes. The clutter can be distracting and counterproductive to the use of the algorithm. Simplicity and standardization have been advocated for successful algorithm development by the Society for Medical Decision Making (28).

In summary, algorithms are a method of representing guidelines for care consisting of nodes where observations, decisions, and actions occur. Depending on the input of the user, different pathways are taken to classify and identify a disease process, treat a disease, or both. Management algorithms allow for the simultaneous diagnosis and treatment of a disease process. Although many benefits to the use of algorithms exist, so do the drawbacks. Nonetheless, they play a central role in computer-based decision models. Medal (29) is a collection of over 11,000 algorithms that may be useful to clinicians or biomedical researchers.

Clinical Pathways

Clinical pathways are multidisciplinary plans of care designed to support the implementation of guidelines and protocols. They support clinical management, clinical and nonclinical resource management, and clinical audit, as well as financial management (30). Clinical pathways have four main components:

1.   A timeline

2.   The categories of care or activities and their interventions

3.   Intermediate and long-term outcome criteria

4.   The variance record (31,32)

Clinical pathways differ from guidelines, protocols, and algorithms, as they are used by a multidisciplinary team and their focus lies on the quality and coordination of care after clinical decisions have already been made to begin the therapy or diagnostic evaluation (30,33).

Bundles

A bundle is a group of interventions related to a disease process that, when executed together, result in better outcomes than when implemented individually. It is a structured way of improving the processes of care and patient outcomes: A small, straightforward set of practices, generally three to five, that, when performed collectively and reliably, have been proven to improve patient outcomes (34). All practices set forth in a bundle must be completed; they have been designed such that each practice can be completed on an all-or-none scale. That is, a practice cannot be almost completed—it either is completed or it is not completed. The practices are scientifically robust, rigorously scrutinized, and based on the highest level of evidence available at the time they are released. The goal of the Institute for Healthcare Improvement (IHI) when releasing the bundles was for the focus of clinicians to be on the implementation of the elements of the bundle as opposed to the content of the elements of the bundle. The key elements of a bundle are that it is made up of very few (but very important) practices, the accountability for its completion lies with an identified person or team, and the completion of the bundle in its entirety improves outcomes. Bundles that have been assembled by the IHI are noted in Table 8.3.

Guidelines from the Institute for Healthcare Improvement

Formed in 1991, the Institute for Healthcare Improvement (IHI) has worked to improve the delivery and execution of health care services for over a decade. Owing to the groundbreaking work of Deming, Juran, and Crosby (35,36,37), they have spearheaded a model for improvement over the past 17 years to assist health care systems in process and quality improvement.

Any discussion regarding quality improvement (QI) must first focus on the historical framework. QI is defined as a planned approach to transform organizations by evaluating and improving systems to achieve better outcomes. Intrinsic to this definition is the specification of program/production/service components, measurement, and identification of outcomes criteria; these consist of a number of components. Deming, Juran, and Crosby had slightly different ideas about QI; however, there were similarities between each of these quality innovators, which are exemplified in Table 8.4 (35,36,37).

Table 8.3 Institute for Healthcare Improvement (IHI) Bundles

Sepsis resuscitation bundle
Sepsis management bundle
Central line bundle
Surgical site bundle
Ventilator-associated pneumonia bundle

 

Table 8.4 Components to Quality

Deming's 14 Points

Juran's Quality Planning

Crosby's Implementation Program

Create constancy of purpose for improvement of product and service

Establish the infrastructure needed to establish and maintain the quality improvement (QI) program

Management commitment: Make it clear where management stands on quality

Adopt the new philosophy

Identify the specific needs for improvement—the QI projects

QI teams: Create QI teams

Cease dependence on mass inspection

For each project, establish a team with clear responsibility for bringing the project to conclusion

Measurement: Create quality measurement to provide for objective evaluation and corrective action

End the practice of awarding business on price tag alone

Provide the resources, training, and support needed by the team

Cost of quality: Define the ingredients of the cost of quality

Improve constantly and forever the system of production and service

Determine who the customers are: The vital few and useful many

Quality awareness: Increase awareness and commitment to quality by all employees

Institute training

Determine the needs of the customer

Corrective action: Provide a systematic approach to resolving problems

Institute leadership

Develop product features that respond to customer needs

Zero defects: Identify activities that must be conducted to implement a zero defects program

Drive out fear

Develop processes that are able to produce what the customer needs/wants

Supervisor training: Prepare supervisors to implement the quality program

Break down barriers between staff areas

Transfer the resulting plans to the operating forces

Zero defects day: Initiate the zero defects program

Eliminate slogans, exhortations, and targets for the workforce

Keep the planned process in its planned state

Goal setting: Work teams establish goals

Eliminate numeric quotas

Evaluate actual quality performance

Error-cause removal: Employees identify obstacles to achieving goals and producing quality goods or services

Remove barriers to pride of workmanship

Compare actual performance to quality goals

Recognition: To appreciate those who contribute in the quality effort

Institute a vigorous program of education and training

Measure: Statistical significance, economic significance, and trends

Quality councils: To coordinate the organization's quality program

Take action to accomplish the transformation

Act on the differences

Do it over again: To emphasize the QI program is continuous

IHI promotes a strategy of “changing health care together” and embodies this approach through their philosophy of “all teach, all learn.” This system reinforces the idea that committed individuals and organizations can, through collaboration, more quickly and efficiently improve health care delivery than any single individual or corporate entity. Figure 8.5 embodies IHI's strategy for transforming health care. At the core of their work is innovation, the creation and testing of new ideas and concepts for improving patient care. Here, they work intensely with cutting-edge organizations on a project basis to test new solutions to old problems. Once a promising change concept has been successfully developed in one setting, it will require being fully vetted and piloted in other settings.

Strategic Relationships

IHI has developed various closely aligned, strategic relationships with dozens of organizations that test and deploy these changes. These high-level partnerships focus on transforming entire systems of care by concentrating on strategic objectives and system-level improvement. IHI has accomplished this at the global level, consulting with health care organizations and countries throughout the world. It is accomplished through multiple methods including developing strategic partnerships as previously discussed and their IMPACT network, where health care organizations come together to achieve dramatic improvement results in clinical outcomes, patient and provider satisfaction, and financial performance, as well as in learning and innovation communities. Learning and innovation communities are collaborative change laboratories focused on front-line improvement. Participating organizations work with each other and with IHI faculty to rapidly test and implement meaningful, sustainable change within a specific topic area. Learning and innovation communities are the next-generation evolution of the Breakthrough Series, IHI's traditional methodology for collaborative improvement.

For example, the improving outcomes for high-risk critically ill patients community focuses on identifying and rescuing patients whose condition is clinically worsening; providing appropriate, reliable, and timely care to high-risk and critically ill patients using evidence-based therapies; creating a highly effective multidisciplinary team; integrating patient and family into care so they receive the care they want; and the development of an infrastructure that promotes quality care. Specific interventions include ventilator and central venous access bundles, rapid-response teams, glucose control both inside and outside of the ICU, sepsis resuscitation and management bundles, multidisciplinary rounds and daily goals, handoffs, and a palliative care team to assist with end-of-life care. As a result of focusing on these areas, the IHI predicts that hospital organizations will be able to decrease raw mortality by greater than 25%, intensive care unit (ICU) mortality by 20%, and ICU length of stay by 20%.

000689

Figure 8.5. The Institute for Healthcare Improvement's strategy for transforming health care.

Learning opportunities, the next layer of the IHI strategy, offers a wide variety of learning opportunities for health care professionals from expert faculty and experienced colleagues around the world. This is accomplished through seminars and Web-based and professional development programs that create opportunities for organizations and individuals to learn and implement best-practice ideas online. These programs are designed for leaders who seek to gain a particular set of skills that are required for an organization to succeed in its improvement agenda. Programs offered by IHI include training for board members, patient safety officers, improvement advisors, and operations managers, as well as personnel in other critical roles.

The final step in the IHI learning system is the broad dissemination of best-practice improvement knowledge, knowledge for the world. This is accomplished primarily through campaigns, the IHI Web site, professional education, and fellowship programs (38).

The Institute for Healthcare Improvement may be best known for its 100,000 Lives Campaign. The goal of the campaign, besides the saving of the lives of 100,000 hospitalized patients, was the building of a reusable national infrastructure for the implementation of evidence-based change. The IHI proclaimed at the conclusion of its campaign that 122,300 lives were actually saved by the implementation of their six interventions (Table 8.5). Criticism was generated relating to the statistical analysis that lead to calculation of the number of lives saved by the initiative, as well as the evidence that showed a benefit to rapid-response teams (39). At the conclusion, however, even the toughest critics believe that the campaign did save lives and was worthy of implementation (40).

Table 8.5 The Institute for Healthcare Improvement (IHI) Campaigns

Goals of the 100,000 Lives Campaign:

· Deploy rapid response teams at the first sign of patient decline

· Deliver reliable, evidence-based care for acute myocardial infarction to prevent deaths from heart attack

· Prevent adverse drug events (ADEs) by implementing medication reconciliation

· Prevent central line infections by implementing a series of interdependent, scientifically grounded steps

· Prevent surgical site infections by reliably delivering the correct perioperative antibiotics at the proper time

· Prevent ventilator-associated pneumonia by implementing a series of interdependent, scientifically grounded steps

Goals of the 5 Million Lives Campaign

· Prevent pressure ulcers by reliably using science-based guidelines for their prevention

· Reduce methicillin-resistant Staphylococcus aureus (MRSA) infection by reliably implementing scientifically proven infection control practices

· Prevent harm from high-alert medications starting with a focus on anticoagulants, sedatives, narcotics, and insulin

· Reduce surgical complications by reliably implementing all of the changes in care recommended by the Surgical Care Improvement Project (SCIP)

· Deliver reliable, evidence-based care for congestive heart failure to reduce readmissions

· Get Boards on Board defining and spreading the best-known leveraged processes for hospital Boards of Directors, so that they can become far more effective in accelerating organizational progress toward safe care

From www.ihi.org/campaign.

Perhaps the most important accomplishment of the campaign was its ability to unite clinicians, allied health care workers, and hospital administrators across the country in support of evidence-based guidelines with the purpose of reducing harm to patients. Currently, the IHI is in the middle of its 5 Million Lives Campaign, with six new interventions designed to eliminate five million cases of patient injury.

Computerized Decision Support/Fuzzy Logic

Intensive care medicine frequently involves making rapid decisions on the basis of a large and disparate array of often incomplete information. Intensivists typically rely on conventional wisdom, evidence, and personal experience to arrive at subjective assessments and judgments. Due to an increased focus on outcomes, physicians are being asked to adhere to explicit guidelines and bundles that have been agreed on by the medical community at large (41). A vast majority of these guidelines have an inherent logical structure and therefore make them suitable for computer implementation. As a result, there has been increasing interest in computer-based support tools to automate certain aspects of the medical decision-making process in the intensive care unit (42). Compared with the human brain, computers are well suited to make rapid calculations, allowing the creation of decision networks that support near-limitless complexity. The variable nature of disease and patient characteristics makes it difficult to decide what should be done in every conceivable set of circumstances. In these situations, physicians must depend on intuitive decision making, sometimes defined as the art of medicine; intuitive decision making is usually described as being unsuitable to computerization.

Subjective judgments generally defy description in terms of the kinds of deterministic mathematical equations that computers are well suited to solve. The methods of fuzzy logic are suited to this kind of task and can lead to algorithms that emulate the nonexplicit nature of clinical decision making (43,44). Fuzzy logic was first introduced by Zadeh in the 1960s (45) and is now a well-established engineering discipline (46,47). Given that fuzzy logic is particularly advantageous in areas where a precise mathematical description of the control process(es) is impossible makes it especially suited to support medical decision making.

Fuzzy logic has been successfully tested in areas of medicine that include the ICU (48). Nemoto et al. (48) were able to create a fuzzy logic controller to facilitate weaning on pressure support mechanical ventilation using patient vital signs and respiratory mechanics such as peak inspiratory and mean airway pressures. Huang et al. (49) created a fuzzy logic controller to control intracranial pressure using propofol sedation. Finally, models for treatment of septic shock have also been created (50). Fuzzy logic provides a means for encapsulating the subjective decision-making process in an algorithm suitable for computer implementation. It appears to be eminently suited to aspects of medical decision making. Further development using fuzzy logic in the ICU is underway, and FDA-approved, commercially available products are on the horizon.

Complexities in Decision Making

In actuality, decision making is much more complex than the search for knowledge and the application of answers. Most of the original research into the study of how humans make decisions focused on economics (51). The process of decision making in critical care medicine is unique from many decision-making processes in business for one major reason: Decisions may be acutely time sensitive. As hypoxia is progressing, one does not have the luxury of multiple diagnostic tests, obtaining a complete history, and a prolonged analysis of the data. If a working diagnosis and appropriate therapeutic measures cannot be instituted quickly, the patient may perish. Several other barriers to successful clinical decision making are discussed below.

Communication

Sir William Osler is attributed with stating, “If you listen carefully to the patient, they will tell you the diagnosis.” Successful communication between patients and their doctors is necessary for obtaining a medical history, as well as for judging response to treatment. It has been stated that physicians interrupt their patients on average of once every 18 seconds. Nonfiction books tell the stories of patients who have been misdiagnosed for years before an astute physician listened carefully and was allowed to correctly make the diagnosis. The amount of time spent in obtaining a history changes dramatically as the physician matures. With this maturation process, the focus of the questions becomes more narrow and succinct, partly as a result of experience and partly as the result of external pressures to see more patients in less time. This is true despite the fact that, at least in the outpatient setting, a disproportionate amount of information is gathered from the history rather than the physical examination or laboratory data (52). In fact, allowing for 90 seconds of spontaneous conversation at the beginning of an outpatient consult—the so-called 90-second rule—has been advocated (53).

Communication in the intensive care unit is quite different. Many of the patients are unable to communicate effectively as a result of severe illness and subsequent neurologic disability, sedative and analgesic medications, and the presence of an endotracheal tube or tracheostomy. This makes communication even more difficult, since the patient may not respond, for example, to abdominal pain on physical examination until it becomes quite severe. Communication may rely on gathering information from the first responders on the scene and the paramedic run sheets subsequently generated. Studies of paramedical personnel show that they can relatively adequately diagnosis cerebral vascular accidents (54), acute myocardial infarction (55), and those with difficulty breathing (56), even though there is a tendency toward overdiagnosis. Although one might think it somewhat better to overdiagnose than underdiagnose, the reliability of information gathered by the first responders may not always be reliable. Information gathered from spouses or next of kin may be variable as well. In the outpatient setting, a history given by the surrogates was accurate in terms of medical history and medication use (57), cigarette and coffee use (58), but not concerning alcohol use (58) or dietary data (59); knowledge of drug use or other illicit activities is more poorly correlated between the patient and a family member (60). In the above-mentioned studies, the information for comparison was gathered in an outpatient setting. The distress of having a loved one acutely and severely ill may affect the ability to recall details of the patient's medical history necessary for diagnosis and treatment. In practice, revisiting the patient's medical history with the family after the initial admission to the intensive care unit may be more fruitful.

Language Barriers

A gap in understanding between patients and health care workers can become quite problematic. Language has been cited as the most common barrier in any health care setting and has been found to be a risk factor to adverse outcomes (61). Interestingly, a recent study determined that nursing staff versus physicians found a language barrier to be more stressful (97% vs. 78%, respectively) and more of an impediment to the delivery of quality care (95% vs. 88%, respectively) (62). Also of note, adherence to medication regimens are more frequently a problem for non-English speakers (63). These data can likely be extrapolated to adherence with nonmedication regimens that have been advised by health care providers as well, although no specific studies have elucidated this point. As immigration continues to rise in the United States, we will be faced with, perhaps, more non-English-speaking patients than in previous years. Although a language barrier should not be a reason for inequities of care provided in a health care setting, there is no universally accepted solution to address this. Each hospital must develop its own resources for dealing with non-English-speaking patients. One method that is exceedingly popular is to have bilingual nurses translate for physicians. In a study evaluating the efficacy of nurse-translators in an ambulatory setting and comparing them to the videotaped interaction, which was translated by blinded medical interpreters, misinterpretations resulting in physician misunderstanding occurred in about 50% of cases (64). In many of these settings, the problems included nurses' further interpretation of the patients' words to become more consistent with the clinical picture and the use of cultural metaphors that did not translate accurately to English. This study demonstrates that even with a language as common as Spanish, misinterpretation that affects the clinician's ability to formulate an accurate diagnosis is still prone to occur. Speaking a nondominant language in a country that tends to be monolingual may also lead to medical interventions that are possibly unnecessary. Despite similar mechanisms of injury, the degree of hypotension during resuscitation, injury severity score (ISS), illicit substance use, alcohol use, and a higher Glasgow coma scale (GCS) score, Spanish-speaking trauma patients were more likely to be endotracheally intubated than their English-speaking counterparts (65).

Resource Limitation

Adherence to recommended guidelines may be well beyond the control of the physician if she or he does not have access to adequate resources or a referral network, and decision making may be affected. It has been suggested that the lack of the immediate availability of an anesthesiologist may interfere with the ability to adhere to consensus guidelines, decreasing the rate of elective cesarean deliveries (66,67). Guidelines aimed at reducing the risk of central venous catheter infections have been available since 1996 (68,69). Yet in 2003, less than 10% of American internists acknowledged using chlorhexidine gluconate for skin preparation prior to insertion. The major factor that determined the use of the antiseptic agent was its availability at the institution (70). Furthermore, lack of appropriate equipment was associated with lack of adherence to guidelines.

Fear of Litigation

The practice of defensive medicine involves using diagnostic and therapeutic measures as a safeguard, or self-protection, in case charges of medical malpractice are levied at some time in the future. Defensive medicine may result in additional, unnecessary testing and/or referrals to other health care providers, or it may result in the practitioner's refusal to treat certain groups of high-risk patients (71). In a 2005 survey, 93% of 824 physicians in Pennsylvania reported practicing defensive medicine (71). The most frequent form of defensive medicine practiced was ordering expensive imaging studies. A study of Israeli otolaryngologists determined that almost 80% of surgeons varied from the American Academy of Otolaryngology–Head and Neck Surgery recommendations regarding coagulation screening tests before tonsillectomy and adenoidectomy. Most of those surgeons that deviated from the practice guidelines stated that the reason for this behavior was the practice of defensive medicine (72). In a study of Illinois neonatologists, many perceived a “gray zone” of resuscitative practices related to the gestational age at which resuscitation would be used or withheld (73). At less than 25 weeks' gestation, the neonatologists were significantly more fearful of litigation should they not resuscitate. The conclusion of this study was that external influences may affect delivery room resuscitation practices. The practice of defensive medicine is not simply the harmless addition of a few unnecessary diagnostic tests. It contributes to the skyrocketing costs of health care and, in some instances, can worsen the expected clinical outcome of patients (74).

Personal Biases and Interindividual Differences

Physicians may make their future decisions biased by their previous experiences. Practicing medicine in the same manner as one's instructors was, and still is, commonplace. The emergence of evidence-based practice is attempting to exchange personal bias for objective scientifically based practice. A previously missed diagnosis may sort to the top of the differential diagnosis list when the physician is confronted with a subsequent patient with a similar presentation; this may be especially true if a bad outcome or medicolegal issue occurred with the previous patient. Although this is frequently assumed to occur, at least one study disagrees with this philosophy, suggesting that physicians with greater malpractice experience showed no systematic differences in initial management choice or subsequent test recommendations (75).

Several studies have examined the differences of the clinical decision-making process between individuals. Although the methods themselves are similar, some differences do exist. It has been found that the more expert diagnosticians ask fewer questions, consider less in their differential diagnoses, and arrive at the correct diagnosis in less time (3,76,77). Although the accumulation of medical knowledge, vast experience, and an excellent memory undoubtedly helps a clinician become a master diagnostician, simply attaining these three qualities does not guarantee diagnostic superiority. Ongoing research is focusing on how these qualities are individually and collectively used by clinicians.

Clinical Inertia

Clinical inertia refers to the practice of not intensifying treatments of patients who are not yet at goals defined by evidence-based medicine. Clinical inertia has been called a leading cause of potentially preventable adverse events, disability, death, and excess medical care costs. Traditionally, the focus of clinical inertia has been on chronic illnesses, such as diabetes, hypertension, and hypercholesterolemia. It appears, however, that clinical inertia is readily present in the ICU as well.

Intensivists in Germany, when polled, claimed that 91.6% used lung-protective strategies suggested by ARDSnet (acute respiratory distress syndrome network), 67.4% used intensive insulin therapy, and 79% used low-dose hydrocortisone therapy for septic shock. When the ICUs were actually surveyed, it was found that only 4.2% of patients were ventilated with ARDSnet-suggested strategies, only 8.8% had tight glucose control, and only 30.6% of patients with septic shock were, in fact, treated with low-dose hydrocortisone (78). Clinical inertia represents a specific phenomenon differing from lack of resources; in clinical inertia, the resources exist and are available—they are simply not used. Multiple causes have been proposed, including fallacious reasoning and overall complexity (79).

Inability to Locate Pertinent Information

Almost 20 years ago, Greenes (80) astutely observed that physicians were faced not with data overload, but information underload. That is, data is available from a multitude of sources, but the navigation through these sources to find relevant information is a harrowing task. It is well recognized that some, if not much, of the information contained in textbooks is outdated by the time the book is put into print. Bias can exist anywhere and everywhere within experimental design and execution, which can result in less-than-conclusive results (Table 8.6). If the data are collected and analyzed, and a manuscript is submitted, the research is subject to publication bias. Several factors have been identified as influencing rates of publication, including sample size, funding, quality, and prestige (81). It has long been felt that authors of smaller studies can boost their chance of publication by showing a stronger effect of their intervention. One answer to solving some of the publication bias is the creation of meta-analyses; however, significant problems can arise from combining studies that have different criteria for enrollment. As an example, consider the early studies conducted on patients suffering with acute respiratory distress syndrome (ARDS). ARDS had been known previously as shock lung, stiff lung, wet lung, and white lung. Prior to the 1994 American-European Consensus Conference on ARDS, different studies identified what we now commonly call ARDS by different criteria. This resulted in drastically different outcomes observed in response to similar treatment strategies. For a meta-analysis to have significance, the disease process must be understood and the definitions for patients included in the studies must be universal.

Table 8.6 Types of Bias That Can Exist in Research

Referral bias
Nonrespondent bias
Insensitive measure bias
Expectation bias
Recall bias
Attention bias
Verification bias
Contamination bias
Cointervention bias
Compliance bias
Withdrawal bias
Proficiency bias

A final problem is that no study may exist regarding the best treatment when patients have two disease processes with competing treatment goals—for example, the treatment of the patient with cerebral edema and concomitant cerebral vasospasm. The goals of therapy for each of those processes are evidence-based and well published. The appearance of a patient who is suffering from both processes at the same time is not infrequent, but the literature on the treatment of such a patient is nonexistent at the time of the writing of this chapter.

It should be noted that the limitations of searching computer databases have not been addressed. Identifying all possible search parameters prior to conducting a database search is necessary. For example, searching an online computer database such as PubMed for literature dealing with the treatment of retinopathy of prematurity and for that dealing with retrolental fibroplasia generates different results despite the fact that these represent the same clinical entity. Keeping current with changes in medicine is also important. Searching for treatments for Stenotrophomonas infections produces different results from those produced when searching for treatments for Xanthomonas infections. These bacteria are the same, but the name change is a result of a reinterpretation of the taxonomic position, which occurred in the early 1980s (82). Searching databases for synonyms for the same disease process can yield up to a tenfold difference in results! Although lack of information is definitely problematic, too much information may also be a problem. In one study, the introduction of additional options increased the difficulty of the physician to make decisions. In one scenario, the uncertainty in deciding between two similar treatment options led some physicians to avoid this decision altogether and recommend not starting either treatment regimen for the patient (83).

Fatigue

It has long been felt that stress has a profound impact on clinical judgment and medical decision making. Research into the role that stress and fatigue have on clinical performance is actively ongoing.

Libby Zion's tragic death on March 4, 1984—a result of complications due to serotonin syndrome—drew public attention to the conditions under which resident physicians work. Although a grand jury exonerated the physicians, it was discovered that residents were working consistently more than 100 hours per week, sometimes continuously for 30 to 40 hours, and with minimal supervision. Although other factors contributed to Ms. Zion's death, the role that fatigued physicians played was seen as a serious potential danger. In March 1987, the New York State Commissioner of Health appointed Doctor Bertrand Bell to oversee a committee to evaluate the findings of the grand jury. The Bell Committee, as it was later called, handed down recommendations for the limitation of resident work hours. These limitations, also referred to as the Libby Law, were initially instituted in New York, but were eventually adopted elsewhere. On July 1, 2003, the Accreditation Council for Graduate Medical Education (ACGME) instituted standards for all accredited residency programs, limiting the work week to 80 hours; this has been adopted by all residency programs to maintain accreditation. One study found that 35.9% more serious medical errors were made—including 56.6% more nonintercepted serious errors—during a traditional call schedule (one-in-three) than during the intervention schedule that limited call hours to those recommended by the Bell Committee (84). Decline in cognitive performance has been specifically reported in ECG interpretation (85), monitoring during anesthesia (86), and surgical performance. One report suggested that surgical complication rates were 45% higher among residents who had been on call the previous night (87).

One study found that staying awake for 24 hours continuously impairs cognitive performance to a similar degree as having a 0.1% blood alcohol level (88). One of the first qualities to be impaired by alcohol intoxication is insight, which may lead some to assume that insight is equally impaired by those who are sleep deprived. That is, they do not fully appreciate how much their practice is impaired by sleep deprivation. Although insight may be impaired, interestingly, subjective ratings of high pressure in the workplace and insufficient sleep are associated with an increase in self-reported omissions in patient care (89). Thus, it seems that the sleep-deprived physician may realize she or he is taking less than optimal care of a patient but does nothing about it.

Has the ACGME limitation of work hours resulted in decreased morbidity and mortality? A recent study indicates that it may have, noting that there was a decreased short-term mortality among high-risk medical patients in teaching hospitals, but no difference was seen among surgical patients (90). Similar results were found in another study performed in Veterans Administration Hospitals (91), finding a decrease in the mortality of medical patients but no associations with surgical patients.

000988

Figure 8.6. Performance versus arousal plot of the Yerkes-Dodson law. (From Yerkes RM, Dodson JD. The relation of strength of stimulus to rapidity of habit-formation. J Comp Neurol Psychol. 1908;18(5):459–482, with permission.)

Stress

Although stress and fatigue are inherently related, and too much of either is not beneficial, some degree of stress, or arousal, is necessary to perform well and make sound decisions and judgments. This idea was proposed by psychologists Robert M. Yerkes and J.D. Dodson in 1908 (92) (Fig. 8.6). Different tasks may require different levels of arousal. For example, intellectually demanding tasks may require a lower level of arousal so that concentration may be facilitated; on the other hand, tasks demanding stamina or persistence may be performed better with higher levels of arousal (to increase motivation). Excessive arousal, anxiety, or stress results in diminished performance in clinical scenarios as seen on the downward portion of the graph.

Summary

In many ways, clinical decision making resembles the practice of medicine as a whole; there is an artistic and a scientific component to both. How physicians acquire knowledge is well known. Knowledge gained from books is constantly supplanted by knowledge gained from critical reviews of the medical literature. How physicians apply that knowledge to arrive at a clinical decision is not as well understood and varies greatly between clinicians. Physicians use standards, guidelines, algorithms, clinical pathways, and bundles to assist in the provision of rapid, evidence-based care. Organizations such as the Institute for Healthcare Improvement assist in the identification and dispersion of best practices. As technology advances and information is shared both more readily and more rapidly, standardization of the practice of medicine will continue to increase. Although critics exist who fear the evolution of cookbook medicine, outcomes support the benefit conferred by the application of standards, bundles, and guidelines. The application of fuzzy logic and computerized decision support to the medical field is very exciting and has almost limitless utility at the bedside of the critically ill patient. One, of course, must realize that nothing can replace clinical judgment.

In the day-to-day practice of medicine, decision making is affected by other influences as well. Communication and language barriers, fatigue and stress, and the inability to locate pertinent information all impede a physician's ability to make accurate and efficient decisions. Nonetheless, hundreds of thousands of medical decisions are made in hospitals across the world on a daily basis that affect the lives of our patients. We have just begun to understand all the components involved in making those decisions.

References

1. Del Mar C, Doust J, Glasziou P. Clinical Thinking: Evidence, Communication and Decision-Making. Malden, MA: Blackwell; 2006.

2. Sackett DL, Straus SE, Richardson WS, et al. Evidence-based Medicine. How to Practice and Teach EBM. 2nd ed. Edinburgh, Scotland: Churchill Livingstone; 2000.

3. Dhaliwal G. Clinical decision-making: understanding how clinicians make a diagnosis. In: Saint S, Drazen J, Solomon C, eds. Clinical Problem Solving. New York, NY: McGraw-Hill; 2006.

4. Feltovich PJ, Barrows HS. Issues of generality in medical problem solving. In: Schmidt HG, De Volder ML, eds. Tutorials in Problem-based Learning: a New Direction in Teaching the Health Professions. Assen, the Netherlands: Van Gorcum; 1984:128–142.

5. Leape LL. Error in medicine. JAMA. 1994;272:1851–1857.

6. Hilliard AA, Weinberger SE, Tierney LM Jr, et al. Clinical problem-solving. Occam's razor versus Saint's Triad. N Engl J Med. 2004;350(6):599–603.

7. Tversky A, Kahneman D. Availability: a heuristic for judging frequency and probability. Cogn Psychol. 1973;5:207–232.

8. Tversky A, Kahneman D. Judgment under uncertainty: heuristics and biases. Science. 1974;185:1124–1130.

9. Tversky A, Kahneman D. Evidential impact of base rates. In: Kahneman D, Slovic P, Tversky A, eds. Judgment under Uncertainty: Heuristics and Biases. Cambridge, England: Cambridge University Press; 1982.

10. McDonald CJ. Medical heuristics: the silent adjudicators of clinical practice. Ann Intern Med. 1996;124:56–62.

11. Patel VL, Kaufman DR, Arocha JF. Emerging paradigms of cognition in medical decision-making. J Biomed Inform. 2002;35:52–75.

12. http://www.asahq.org/publicationsAndServices/standards/01.pdf. Accessed January 1, 2008.

13. http://www.acmq.org/policies /policies3and4.pdf. Accessed January 1, 2008.

14. U.S. Department of Health and Human Services Food and Drug Administration Center for Drug Evaluation and Research (CDER) Center for Biologics Evaluation and Research (CBER) Guidance for Industry E6 Good Clinical Practice: Consolidated Guidance April 1996, ICH.

15. Field MJ, Lohr KN, eds. Clinical Practice Guidelines: Directions for a New Program, Institute of Medicine. Washington, DC: National Academy Press; 1990.

16. http://www.openclinical.org/guidelines.html#fieldandlohr. Accessed January 1, 2008.

17. Delamothe T. Wanted: guidelines that doctors will follow [editorial]. BMJ. 1993;307:218.

18. Wollersheim H, Burgers J, Grol R. Clinical guidelines to improve patient care. Neth J Med. 2005;63(6):188–192.

19. Delinger RP, Levy MM, Carlet MD, et al. Surviving Sepsis Campaign: international guidelines for the management of severe sepsis and septic shock: 2008. Crit Care Med. 2008;36:296–327.

20. Hurtado FJ, Nin N. The role of bundles in sepsis care. Crit Care Clin. 2006; 22(3):521–529.

21. Kortgen A, Niederprum P, Bauer M. Implementation of an evidence-based “standard operating procedure” and outcome in septic shock. Crit Care Med. 2006;34(4):943–949.

22. Shapiro NI, Howell MD, Talmor D, et al. Implementation and outcomes of the Multiple Urgent Sepsis Therapies (MUST) protocol. Crit Care Med. 2006;34(4):1025–1032.

23. Hadorn C. Use of algorithms in clinical guideline development in clinical practice guideline development: methodology perspectives. AHCPR Pub. No. 95-0009. Rockville, MD: Agency for Health Care Policy and Research; January, 1995:93–104.

24. Grimm RK, Shimoni K, Harlon W, et al. Evaluation of patient-care protocol use by various providers. N Engl J Med. 1975;292:507–511.

25. Komaroff AL, Black WL, Flatley M, et al. Protocols for physicians' assistants: management of diabetes and hypertension. N Engl J Med. 1974;290:307–312.

26. Sox HC. Quality of patient care by nurse practitioners and physician's assistants: a ten year perspective. Ann Intern Med. 1979;91:459–468.

27. Sox HC, Sox CH, Tompkins RK. The training of physician's assistants: the use of a clinical algorithm system for patient care, audit of performance and education. N Engl J Med. 1973;288:818–824.

28. Proposal for clinical algorithm standards. Society for Medical Decision Making Committee on Standardization of Clinical Algorithms. Med Decis Making. 1992;12:149–154.

29. www.medal.org.

30. http://www.openclinical.org/clinicalpathways.html. Accessed January 1, 2008.

31. Hill M. The development of care management systems to achieve clinical integration. Adv Pract Nurs Q. 1998;4(1):33–39.

32. Hill M. CareMap and case management systems; evolving models designed to enhance direct patient care. In: Blancett SS, Flarey DL, eds. Reengineering Nursing & Health Care: Handbook for Organizational Transformation. Boston, MA: Jones and Bartlett; 1995.

33. http://www.oqp.med.va.gov/cpg/faqs.asp. Accessed January 1, 2008.

34. http://www.ihi.org/IHI/Topics/CriticalCare/IntensiveCare/ImprovementStories/WhatIsaBundle.htm, Accessed January 1, 2008.

35. W. Edwards Deming Institute (2008). http://www.deming.org/. Accessed January 1, 2008.

36. Juran Institute (2008). http://www.juran.com/. Accessed January 1, 2008.

37. Phillip Crosby Associated (2008). http://www.philipcrosby.com/pca/index.html. Accessed January 1, 2008.

38. Institute of Healthcare Improvement (2008). http://www.ihi.orh/ihi. Accessed January 12, 2008.

39. Wachter RM, Pronovost PJ. The 100,000 Lives Campaign: a scientific and policy review. Jt Comm J Qual Patient Saf. 2006;32(11):621–627.

40. Robeznieks A. IHI campaign draws more fire. Article questions methodology, rapid-response teams. Mod Healthc. 2006;36(49):14–15.

41. Kohn L, Corrigan J, Donaldson M, eds. To Err Is Human: Building. A safer health care system. Committee on Quality of Health Care in America, Institute of Medicine. Washington, DC: National Academy Press. 2000;63:101–107.

42. Hanson CW, Marshall BE. Artificial intelligence applications in the intensive care unit. Crit Care Med. 2001;29:427–435.

43. Steimann F. On the use and usefulness of fuzzy sets in medical AI. Artif Intell Med. 2001;21:131–137.

44. Helgason CM, Jobe TH. Causal interactions, fuzzy sets and cerebrovascular ‘accident’: the limits of evidence-based medicine and the advent complexity-based medicine. Neuroepidemiology. 1999;18:64–67.

45. Zadeh LA. Fuzzy sets. Inf Control. 1965;8:338–352.

46. Cox E. Fuzzy fundamentals. IEEE Spectrum. 1992;Oct:58.

47. Hess J. Fuzzy logic and medical device technology. Med Device Technol. 1992;6(8):37–46.

48. Nemoto T, Hatzakis GE, Thorpe CW, et al. Automatic control of pressure support mechanical ventilation using fuzzy logic. Am J Respir Crit Care Med. 1999;160:550–556.

49. Huang SJ, Shieh JS, Fu M, et al. Fuzzy logic control for intracranial pressure via continuous propofol sedation in a neurosurgical intensive care unit. Med Eng Phys. 2006;28(7):639–647.

50. Paetz J. Knowledge-based approach to septic shock patient data using a neural network with trapezoidal activation functions. Artif Intell Med. 2003;28(2):207–230.

51. von Neumann J, Morgenstern O. Theory of Games and Economic Behavior. Princeton, NJ: Princeton University Press; 1944.

52. Hampton JR, Harrison MJ, Mitchell JR, et al. Relative contributions of history taking, physical examination, and laboratory investigation to diagnosis and management of medical outpatients. BMJ. 1975;2:486–489.

53. Langewitz W, Denz M, Keller A, et al. Spontaneous taking time at start of consultation in outpatient clinic cohort study. BMJ. 2002;325:682–683.

54. Zweifler RM, York D, U TT, et al. Accuracy of paramedic diagnosis of stroke. J Stroke Cerebrovasc Dis. 1998;7(6):446–448.

55. Bright H, Pocock J. Prehospital recognition of acute myocardial infarction. CJEM. 2002;4(3):212–214.

56. Ackerman R, Waldron RL. Difficulty breathing: agreement of paramedic and emergency physician diagnoses. Prehosp Emerg Care. 2006;10(1):77–80.

57. Lipworth L, Fryzek JP, Fored CM, et al. Comparison of surrogate with self-respondents regarding medical history and prior medication use. Int J Epidemiol. 2001;30(2):303–308.

58. McLaughlin JK, Mandel JS, Mehl ES, et al. Comparison of next-of-kin with self-respondents regarding questions on cigarette, coffee, and alcohol consumption. Epidemiology. 1990;1(5):408–412.

59. Fryzek JP, Lipworth L, Signorello LB, et al. The reliability of dietary data for self- and next-of-kin respondents Ann Epidemiol. 2002;12(4):278–283.

60. Fernández Hermida JR, Secades Villa R, Vallejo Seco G, et al. Evaluation of what parents know about their children's drug use and how they perceive the most common family risk factors. J Drug Educ. 2003;33(3):337–353.

61. Aboul-Enein FH, Ahmed F. How language barriers impact patient care: a commentary. J Cult Divers. 2006;13(3):168–169.

62. Bernard A, Whitaker M, Ray M, et al. Impact of language barrier on acute care medical professionals is dependent upon role. J Prof Nurs. 2006;22(6):355–358.

63. Westberg SM, Sorensen TD. Pharmacy-related health disparities experienced by non-English-speaking patients: impact of pharmaceutical care. J Am Pharm Assoc (2003). 2005;45(1):48–54.

64. Elderkin-Thompson V, Silver RC, Waitzkin H. When nurses double as interpreters: a study of Spanish-speaking patients in a US primary care setting. Soc Sci Med. 2001;52(9):1343–1358.

65. Bard MR, Goettler CE, Schenarts PJ, et al. Language barrier leads to the unnecessary intubation of trauma patients. Am Surg. 2004;70(9):783–786.

66. Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA. 1999;282:1458–1465.

67. Kosecoff J, Kanouse DE, Rogers WH, et al. Effects of the National Institutes of Health Consensus Development Program on physician practice. JAMA. 1987;258(19):2708–2713.

68. Pearson ML. Guideline for prevention of intravascular device-related infections. Hospital Infection Control Practices Advisory Committee. Infect Control Hosp Epidemiol. 1996;17:438–473.

69. Pearson ML. Guideline for prevention of intravascular device-related infections, I: intravascular device-related infections: an overview. The Hospital Infection Control Practices Advisory Committee. Am J Infect Control. 1996;24:262–277.

70. Rubinson L, Wu AW, Haponik EE, et al. Why is it that internists do not follow guidelines for preventing intravascular catheter infections? Infect Control Hosp Epidemiol. 2005;26:525–533.

71. Studdert DM, Mello MM, Sage WM, et al. Defensive medicine among high-risk specialist physicians in a volatile malpractice environment. JAMA. 2005;293(21):2609–2617.

72. Toker A, Shvarts S, Perry ZH, et al. Clinical guidelines, defensive medicine, and the physician between the two. Am J Otolaryngol. 2004;25(4):245–250.

P.93

73. Weiss AR, Binns HJ, Collins JW Jr, et al. Decision-making in the delivery room: a survey of neonatologists. J Perinatol. 2007;27(12):754–760. Epub 2007 Aug 30.

74. DeKay ML, Asch DA. Is the defensive use of diagnostic tests good for patients, or bad? Med Decis Making. 1998;18(1):19–28.

75. Glassman PA, Rolph JE, Petersen LP, et al. Physicians' personal malpractice experiences are not related to defensive clinical practices. J Health Polit Policy Law. 1996;21(2):219–241.

76. Elstein AS, Shulman LS, Sprafka SA. Medical Problem Solving: An Analysis of Clinical Reasoning. Cambridge, MA: Harvard University Press; 1978.

77. Neufeld VR, Norman GR, Freightner JW, et al. Clinical problem-solving by medical students: a cross-sectional and longitudinal analysis. Med Educ. 1981;15:315–322.

78. Brunkhorst FM, Engel C, Jaschinsky U, et al, and the German Competence Network Sepsis (SepNet). Treatment of severe sepsis and septic shock in Germany: the gap between perception and practice—results from the German Prevalence Study. Infection. 2005;33(Suppl 1):49.

79. Miles RW. Fallacious reasoning and complexity as root causes of clinical inertia. J Am Med Dir Assoc. 2007;8:349–354.

80. Greenes RA. “Desktop knowledge”—a new focus for medical education and decision support. Methods Inf Med. 1989;28(4):332–339.

81. Fouque D. Producing systematic reviews of best quality: a prerequisite for evidence-based nephrology. J Nephrol. 1999;12:314–317.

82. Palleroni NJ, Bradbury JF. Stenotrophomonas, a new bacterial genus for Xanthomonas maltophilia (Hugh 1980) Swings et al. 1983. Int J Syst Bacteriol. 1993;43(3):606–699.

83. Redelmeier DA, Shafir E. Medical decision-making in situations that offer multiple alternatives. JAMA. 1995;273:302–305.

84. Landrigan CP, Rothschild JM, Cronin JW, et al. Effect of reducing interns' work hours on serious medical errors in intensive care units. N Engl J Med. 2004;351(18):1838–1848.

85. Friedman RC, Bigger JT, Kornfeld DS. The intern and sleep loss. N Engl J Med. 1971;285:201–203.

86. Denisco RA, Drummond JN, Gravenstein JS. The effect of fatigue on the performance of a simulated anesthetic monitoring task. J Clin Monit. 1987;3:22–24.

87. Marcus A. “Doctors Blunder When Lacking Slumber”. HealthDayNews, 2002.

88. Dawson D, Reid K. Fatigue, alcohol and performance impairment. Nature. 1997;388:235.

89. Feddock CA, Hoellein AR, Wilson JF, et al. Do pressure and fatigue influence resident job performance? Med Teach. 2007;29(5):495–497.

90. Shetty KD, Bhattacharya J. Changes in hospital mortality associated with residency work-hour regulations. Ann Intern Med. 2007;147:73–80.

91. Volpp KG, Rosen AK, Rosenbaum PR, et al. Mortality among patients in VA hospitals in the first 2 years following ACGME resident duty hour reform. JAMA. 2007;298(9):984–992.

92. Yerkes RM, Dodson JD. The relation of strength of stimulus to rapidity of habit-formation. J Comp Neurol Psychol. 1908;18(5):459–482.



If you find an error or have any questions, please email us at admin@doctorlib.info. Thank you!