Section I: Clinical Judgment

INTRODUCTION

Clinical judgment has always been a guide for good practitioners. Before physical therapists were able to use theoretically grounded and methodologically tested research hypotheses that provided scientifically sound evidence to guide clinical decision making, they sought guidance from well-developed practice knowledge and clinical reasoning. When good practitioners attempted new interventions, they carefully observed the situation and could recall successes and failures. Studies of expertise in physical therapy have helped us understand this important element in evidence based practice (EBP).^1,2,3,4,5 The deep practice-based knowledge of these experts gives them skill in active reflection.¹ Experts use all three elements of EBP but always modulate what they learn from the literature with their own expert judgment to best meet their patients' needs.

This section begins with Chapter 1, which provides an overview of the language of practice and the important developments over the past decades that allow us to communicate with each other and our health care colleagues. In this chapter we describe two important purposes for a common language in our profession: to describe our patients accurately and to describe our processes and outcomes of care. You will find a useful and context-rich description of models of disability and patient management as well as an introduction to the Guide for Physical Therapist Practice.

In Chapter 2 we present an overview of how physical therapists make clinical decisions. The types of decisions required in practice are highlighted along with an array of approaches to clinical reasoning used by physical therapists and some of their colleagues in medicine, nursing, and occupational therapy.

Chapter 3 presents a perspective on errors in clinical reasoning, drawing from reports in medicine, physical therapy, and occupational therapy. An understanding of the types of errors typically found in physical therapy practice and the sources of influences on errors is an important step in improving clinical reasoning.

Chapter 4, written by Jack Hershey, provides an understanding of how biases affect clinical decisions. Detecting bias in one's own thinking is difficult, but if it is done well and consistently, it will result in more accurate sharing of knowledge with patients.

Chapter 5 deals with the tools available to physical therapists to assist with clinical reasoning. You are encouraged to examine the way physical therapists think and their use of practice algorithms that might facilitate the best clinical reasoning. Policies that improve understanding of what good clinical practice is or what evidence to use in clinical decision making are also introduced in this chapter.

REFERENCES

INTRODUCTION

As human beings, we use language, all kinds of language, to communicate with each other. We speak of verbal and nonverbal language; of written, oral, sign, and symbolic languages; of languages of different peoples, countries, and times. Each occupation has also developed a language that describes the tools, the processes, and the products of that occupation. Some of these languages are easily translated or understood by those outside the occupation; others become more complex, perhaps even obscure and can sometimes serve to exclude others. But these languages occur for good reason—they serve to provide the framework for a way of thinking, for clarifying and articulating the lens used by its members to view the world. We are not speaking here of the many abbreviations, acronyms, or eponyms that have cropped up in the vocabulary of health care practitioners, but of the basic language that defines the philosophy of practice. There are at least two central components of a useful language of practice. We need to be able to describe our patients and then to describe the process and outcomes of our care for these patients. This chapter will present systems based on the biopsychosocial model to describe our patients and a patient management model that describes the process of our care.

DESCRIBING OUR PATIENTS

Accurate and comprehensive descriptions of patients are essential. The descriptive label given a patient can help us not only as health care practitioners but as sociologists, health services researchers, and others, striving to understand the concepts of health and illness and what they mean in human life. In this instance the label serves as shorthand for explaining the natural progression of the specific patient and the general condition of groups of people. A label also can help us choose proper interventions and predict outcomes for our patients. In this instance the descriptive label serves as a diagnosis.

The Biopsychosocial Model

The World Health Organization (WHO) has defined health as "a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity."¹ Since it was adopted in 1946, this definition, while presenting an ideal, has permeated much of the discussion on health care. It served as the foundation for the overarching model for understanding health and illness known as the biopsychosocial model (Fig. 1-1).² This model assumes that the person's response to any state of health or illness is a result of the interaction of the person's psychology and social environment with the biological determinants of disease and trauma.

Despite this long-standing recognition that health and illness exist far beyond the confines of pathology, the primary language used to describe patients has been to label the patient by using the patient's diagnosis of pathology. For example, we speak of the patient with viral rhinitis (the common cold), or the patient with a bladder infection, or the patient with an ankle sprain, or the patient with multiple sclerosis. While these labels do indeed describe the factor that initiates the need for health care, they do not necessarily describe the continuing status of health or illness the patient may experience, so they cannot help us determine the continuing health care needs of the patient. The label of viral rhinitis may well be sufficient to predict the care needs of the patient since it is a self-limiting disease that is "cured" with time. The label of bladder infection may well be sufficient, as long as it is precise enough to prescribe the appropriate antibiotic medication that then "cures" the disease. The label of ankle sprain may not be quite as useful, however, because depending on several factors (patient status, structures involved, nature of the injury, degree of the sprain, etc.) a sprain may never be fully "cured." The patient may well have residual alterations to soft tissue that will always leave him or her susceptible to reinjury. The label of multiple sclerosis becomes yet more problematic both in describing the patient and in contributing to determining the patient's care needs. There is not a known "cure" for this problem, nor is there a known method of controlling its progression. The efforts of the health care system related to this pathology are focused on reducing the intensity of the effects of the disease and on helping patients maintain maximum function.

DEVELOPMENT OF MODELS FOR ENABLEMENT/DISABLEMENT

For physical therapy, as for many of the fields in the area of rehabilitation, articulation of the concept of disability has given us the language to more fully describe our patients than the language used previously. Sociologists studying the sociology of health and illness recognized that using pathology to describe patients was simply not sufficient to fully understand and describe the life of a person with a chronic or progressive disease. This led sociologists to begin to develop disablement models. Nagi provided the first organized description of a concept of disablement as distinct from disease.^3,4 He identified that in order to understand disability, it was necessary to make distinctions among four separate but linked concepts: pathology, impairment, functional limitation, and disability.

The first concept is pathology, broadly defined to include a wide variety of diseases arising from different etiologies, including infection, trauma, and degenerative processes. Whatever the etiology, the pathology is seen as the start of a cascade of events and only one part of the full picture that explains the health status of the patient. All pathologies have the potential to cause problems in the human organism. The next stage, the consequences of pathology, occurring at the organ or system level, is called impairment. For example, the pathology of fracture results in impairments of pain and loss of motion. The next stage of the cascade is functional limitations. Impairment often results in the inability to carry out specific functional activities. For example, the pain that results from a fracture can in turn result in reduced ambulation endurance. Disability is the next step of the cascade. Disability moves beyond the biological events and begins to look at the person. A disability occurs when the person's functional limitations interfere with the person's role in life, be it work, family, or leisure roles. For example, the limitation of ambulation endurance that results from the fracture becomes a problem when ambulating long distances is part of the person's normal work or leisure activities and cannot be supplemented by the use of personal assistance or assistive devices.

Expansion of the Nagi Model

These concepts were expanded by later authors to add more dimensions to Nagi's work (Fig. 1-2).^5,6,7 One principal addition is the identification of the factors that can affect the cascade from pathology to impairments to functional limitations. This modified model shows that demographic factors, health history and habits, and personal characteristics can all affect the cascade. In addition, Guccione added the concept of handicap. Society places a handicap on a person when it refuses to make adjustments to accommodate the disability. This refusal may be an appropriate response based on available resources, but it still handicaps the person. This model was used to drive the creation of The Guide to Physical Therapist Practice⁸ and is therefore very familiar to many physical therapists in the United States.

International Classification of Functioning Model

At the same time that the language of the Nagi model and its evolutions were gaining a foothold in the conversation about health care in the United States, the WHO was developing the International Classification of Functioning, Disability and Health (ICF). At the time, the International Classification of Diseases (ICD), which classifies patient care based primarily on the patient's pathology and which was also developed by the WHO, was the most prominent of the classification systems. In the United States it is widely used to classify patients for many purposes, including reimbursement and categorization for required reporting. It has also been used for research and for prediction of resource allocation, especially for prevention and for management of chronic diseases. Just as with the Nagi-based models, it was recognized that this pathology-based classification scheme was inadequate by itself to meet all the needs of the health care systems around the world. The ICF is the response to this realization and to the recognition that an enablement model is a better way to describe the full continuum of the human health experience. The most recent revision of the ICF (Fig. 1-3) has now been widely accepted across the world as a system for classifying the relationship of health, function, and disability.^9,10

The ICF is based on the premise that disability is actually a continuum that coexists with health. In other words, everyone is in some state of health and some state of disability. The ICF consists of three parts: the health status (top row of the model); three components of functioning—body structure, activities, and participation (second row of the model); and contextual factors of environment and person (third row of the model). It clarifies that factors related to the person and to the environment can modulate the affect of health status on body structures and functions, on activity, and on participation in one's role. It also provides a model that supports the concept of prevention as a means to maintain the maximum level of health possible. Physical therapists have begun using this system to describe patient care, especially when discussing patients with long-term disabilities.^10,11,12

While the Nagi-based models and the ICF use somewhat different language, they have in common a view that pathology is a starting point for a cascade of events and that these events can be modulated by many factors. Table 1.1 provides the definitions of the terms used in the two models, Nagi (and its derivatives) and ICF, which are most familiar to physical therapists.

These models have given us language which allows us to explore all the sequelae of disease and illness, be they acute or chronic, and which describes prevention activities. Moreover, these models firmly recognize that health is a complex phenomenon that goes well beyond the biological basis of pathology to include the psychological and social aspects of human life.

DESCRIBING THE PROCESS AND OUTCOMES OF OUR CARE

The Guide's Patient Care Management Model

The American Physical Therapy Association's (APTA) Guide to Physical Therapist Practice was developed to provide a common language of physical therapy care that could be used by a wide variety of audiences, including physical therapist clinicians, educators, and researchers, and also external groups such as referring sources and payers. One of the most important parts of the Guide is the patient care management model (Fig. 1-4).^8,13 The process described by this model uses decision making that is very similar to the process used by other health care practitioners, including physicians. It does this through both its organization and its language, as will be pointed out as the model is described.

Examination

This first step in the model uses the term examination. This is an important language change from the terms evaluation or assessment that have been used traditionally in physical therapy. The process of examination is both sequential and iterative and consists of three parts: history, systems review, and specific tests and measures. An examination begins with the history. By learning about the patient's past and current status, as reported by the patient and others, the clinician can begin to focus on the specific tests to perform. The history includes not only the typical chief complaint, medical history, and medication history but also demographic data; social history, including employment; growth and development; living environment; general heath status; social and health habits; family history; functional status; and a review of clinical tests performed by other practitioners. (Chapter 7 provides valuable information on conducting the patient interview as the major source of the patient's history.)

Systems Review: All practitioners do a quick systems review each time they see a patient. For physical therapists this is a review of the four systems that are in their primary domain: musculoskeletal, neuromuscular, cardiovascular/pulmonary, and integumentary, as well as the patient's communication abilities. A review of a system consists of simple measures that help to decide if further specific testing is needed. For example, if we are seeing a patient who has come to us with an upper-extremity problem and we watch that person walk into the treatment booth with no apparent difficulty, we have just done a systems review for ambulation problems and do not do any specific lower-extremity testing. Most practitioners do this as an almost automatic part of practice. For physical therapists, adopting the patient care management model means documenting the systems review as a formal part of a patient's care.

Specific Tests and Measures: The history and the systems review are used to guide the practitioner in selecting from among the available specific tests and measures that seem best indicated for the individual patient. Box 1.1 shows the list of specific tests and measures used by physical therapists. The Guide provides a detailed analysis of each area of testing, including clinical indications, the types of tests available, the tools used in gathering data, and the data generated.⁸

The decision about which tests to choose demonstrates one of the interrelationships between the concept of the enablement/disablement model and the patient care management model. We will analyze one of the areas of special tests and measures, flexibility, as an example. A physical therapist would make the decision to use a measure of flexibility when finding clinical indicators from the history and systems review that supported that decision. These indicators could be based in pathology, such as the presence of rheumatoid arthritis. They could be based in impairments, such as reports of decreased range of motion. They could be based in functional limitations, such as reports of inability to dress. They could be based in disability, such as the report that the patient is no longer able to work. Whatever the indication, and there may be many, the clinician would then need to decide which specific tests of flexibility would be used, choosing from among regional and specific joint measures. Depending on the tests chosen, the therapist would need access to instruments such as goniometers and also skill in correct measurement techniques. Finally, the tests would give the therapist data that specify those joints and regions where range of motion is not normal.

Evaluation

The next step of the patient care management model is evaluation. In the patient care management model, evaluation is not the physical act of collecting data, but rather the judgment which the physical therapist uses to transform the data from the examination into clinical decisions about the patient. These decisions include diagnosis and prognosis.

Diagnosis: One decision made through evaluating the examination data is the patient diagnosis. The issue of diagnosis is the primary locus of the interrelationship between managing patient care and the concept of disablement. The most common way that patients are classified is by the pathology that begins or has the potential to begin the patient's problems. But a diagnosis is simply assigning a patient to a particular classification. It is this diagnosis, or classification, that helps guide all of the rest of the treatment decisions. So a diagnosis, or classification label, can describe the patient's impairments, limitations, functional limitations, disabilities, or handicaps (using the language of the Nagi model) or describe the impairments, activity limitations, or participation restriction (using the language of the ICF).

There are two parts to this decision that bear analysis. The first part of the equation is the decision to make a classification at all. Why do individual patients need to be assigned to a particular classification? This is the very heart of the diagnostic process. By recognizing that an individual patient is similar to other patients, we can begin to predict with some certainty the likelihood of recovery, the likelihood of various interventions being successful, and the most likely time frame for that level of recovery. If patients are considered only one at a time, we can never develop a sense of these probabilities.

The second part of this equation is the type of diagnosis. As we've discussed, diagnoses of pathology alone are not always useful for physical therapists in designing a plan of care. For example, knowing that a patient has had a cerebral vascular accident (a diagnosis of pathology) tells us very little about this patient's impairments or functional limitations. But it is those very impairments and functional limitations that the therapist wants to address through physical therapy interventions. classification by impairments and functional limitations allows physical therapists to appropriately choose interventions with the best likelihood of success in ameliorating those impairments or functional limitations.

Prognosis: Prognosis is another of the decisions made based on the examination. A prognosis is a prediction of how long it will take to reach certain goals in plans of care. The data from the examination lead not only to a classification but also to a classification that has known probabilities of recovery. The prognosis becomes part of a plan of care that includes achievable goals and outcomes. Goals refer to removing or reducing impairments. Outcomes refer to removing or reducing functional limitations and disability.

Intervention

The final decision made based on the examination is the choice of interventions. The Guide describes three areas of interventions: documentation and communication, patient/client-related instruction, and procedural interventions. The Guide notes that both documentation and communication and patient/client instruction are an integral part of each and every visit, thereby recognizing the importance of these activities for the success of patient care.⁸

Procedural Interventions

Procedural interventions (Box 1-2), the methods used to reach goals and outcomes, vary in each episode of care based on the goals and outcomes established for the patient.⁸ The description of each intervention category in the Guide includes clinical considerations, again divided by pathology, impairments, functional limitations and disability, specific interventions, and anticipated goals and expected outcomes.

Outcomes

The last step in the model is identifying the outcomes anticipated for the individual patient as well as collecting data that help predict outcomes for groups of patients with similar classifications or diagnoses. There are a variety of outcomes of interest to physical therapists, including reduction of pathology, impairments, and functional limitations and disability, as well as improving quality of life, patient satisfaction, and cost efficiency.

CONCLUSION

Both the disablement models and the patient care management model provide language to describe care, to communicate with patients and with each other, to facilitate good clinical thinking (see Chapters 4 and 5), and to frame scholarship. Mr. Ketterman's case (see Appendix) is written using the ICF language and the terminology from the patient care management model. This choice of language allows us to organize a complex health history for a patient with multiple problems in a way that allows for a clear understanding of his needs, the therapist's plans for intervention, and the results of those interventions.

SELF-ASSESSMENT

1. Can you describe the cascade of events for one of your patients, or a patient you have observed, using either the Nagi model of disablement or the International classification of Functioning, Disability and Health?

2. Using this same patient, can you identify diagnoses across the disablement model you are using? Are you using more than a diagnosis of pathology? How do each of the diagnoses across the disablement model affect your clinical decisions and your expectations for the patient's success?

CONTINUED LEARNING

• The Guide to Physical Therapist Practice

In addition to the printed version of the Guide, which originally appeared as an article in Physical Therapy, it is now available in a Web-based format at http://guidetoptpractice.apta.org/. In addition to much more detail about the patient care management model, the Guide has several sections with background information on the tests, measures, and interventions used in physical therapy. It also has a section, "Preferred Practice Patterns" which integrates the information on tests, measures, and interventions into an organized set of expectations for over 30 of the most common groups of patients seen in physical therapy. All of this information can be useful to both students and practicing clinicians in reviewing and expanding their own knowledge base and in assessing the care they plan for particular patients.

If you have not used the patient care management model to document your care, try reorganizing one of your current patient cases into the patient care management model. Did you find that some data might be missing? What advantages and disadvantages do you see to using this format for documentation?

• International classification of Functioning, Disability, and Health (ICF)

In addition to the texts that are available from the WHO, a very useful Web site about the ICF is http://www.who.int/classifcations/icf/en/. The site includes several useful tools, including a checklist for use in documentation. The World Confederation for Physical Therapy also has a variety of resources about the ICF at its Web site, http://www.wcpt.org/icf. Both the WHO and the APTA offer many opportunities to become better trained in using ICF language to describe and document care. The WHO materials are available for download and use at its Web site. The APTA has developed training modules at its Learning Center, http://learningcenter.apta.org/, to assist in understanding and applying the ICF; one module can be used to earn continuing education credits.

REFERENCES

INTRODUCTION

Physical therapists are responsible for making many types of decisions with and for their patients. These decisions are elementary components of every aspect of the patient management model discussed in Chapter 1; for example, physical therapists are required to develop a diagnosis and prognosis and to select interventions. The decisions range from thoughtfully considered decisions developed over time, such as an appropriate discharge destination for a patient following an inpatient course of care, to moment-by-moment decisions, such as movements required during a treatment session. Over the century of our profession's history, physical therapists have grown both in the capability and desire to make these clinical decisions. Early in the profession, physicians wrote detailed treatment prescriptions that covered all but the most mundane clinical decisions required of physical therapists. We were viewed as technical assistants who could neither diagnose a patient's impairments or functional limitations nor develop a plan of care. In the 21st century, we view ourselves as autonomous practitioners with significant responsibility for the judicious care and successful outcomes for our patients. Thus, the types of clinical decisions made and the characteristics of the clinical reasoning processes are of great interest to practitioners who wish to improve their approach to practice.

The science that supports theories of clinical reasoning in physical therapy is built upon studies of cognitive processing in nonmedical fields,¹ medical problem solving,^2,3,4,5,6 and expert practitioners in nursing,^7,8,9 occupational therapy,¹⁰ and physical therapy.^{11,12,13,14,15,16,17,18,19,20,21,22} Expert practitioners provide researchers with the opportunity to study the combination of knowledge and skill acquisition along with the cognitive and metacognitive processes of clinical reasoning.

The most recent attempts to understand the clinical reasoning of experts in contrast to novices have been conducted in the practice environment as the importance of the contextual nature of clinical reasoning has emerged.²³ Grounded theory qualitative case studies have also emerged as the research method of choice for studying the expertise of health care practitioners^8,13,24 as contrasted with earlier research on physician decision making that used standardized stimuli with research subjects in an artificial setting. These innovations in research approaches have provided health care professionals with a richer body of theory for designing strategies with which to increase the clinical reasoning abilities of students and professionals. For example, the majority of studies of decision making by physicians have been conducted to understand diagnostic decisions about the pathology causing the patient's symptoms.^6,25,26 The Guide to Physical Therapist Practice, ²⁷ introduced in Chapter 1, has been important in expanding the work on clinical reasoning in physical therapy.^28,29,30

This chapter will provide an overview of the types of decisions required of physical therapists and the clinical reasoning strategies that have been identified. In the following two chapters, we will discuss errors or biases that could cause clinical judgment to be flawed. Case vignettes will be presented to encourage you to engage actively in a reasoning exercise and then attempt to identify which of several strategies you find typical of your own thinking.

THE CONTEXT OF CLINICAL REASONING

First Things First: Is This My Patient?

Ken's Experience, Part One: Ken is participating in a weekend neighborhood social event that includes a picnic and games. His neighbors know that Ken is a physical therapist in private practice in the community, and some have sought his services in the past. During a game of volleyball, Michelle, one of Ken's neighbors, falls to the ground and grabs her ankle after jumping to block a ball. She cries out in pain for help, and the neighbors gather around her, several looking at Ken, expecting that he will help.

Ken's experience highlights many decisions that a physical therapist must address before making any further clinical decisions. At this point in the case, Ken has made no decision, but with his first step toward the potential patient, he will have made a plethora of decisions.

*Pause now and list as many decisions as you can that you would make in the first few seconds of your response.

Now review Box 2.1 to see the many questions that have to be answered as Ken prepares to decide about intervening in the situation. Notice that several of the questions that are included are designed to help the therapist refine his decision making: legal, ethical, competence-based, and status-based. This tool can be helpful in thinking broadly about all the types of decisions required when you are in a similar situation. Ken's experience illustrates the importance of making good judgments before initiating a relationship with a patient; if a poor judgment is made at this point, later decisions will likely become more complex. For example, if Ken's specialized expertise and practice are in cardiac rehabilitation and he decides to help Michelle, he may enter into the diagnostic reasoning process without sufficient knowledge to avoid causing harm. These unexpected, pre-relationship types of decisions have not been studied in the literature, but they share some characteristics with ethical reasoning—the type of reasoning used by physical therapists when ethical and practical dilemmas arise.³¹ Experienced clinicians often reflect that the decisions related to whom to treat and when to treat become some of the most challenging clinical decisions in their practice. We all would benefit from more evidence in order to answer questions concerning many of the patients with whom we have the opportunity to interact. In this case, a potential future patient needs emergency care, and the decision to intervene or not becomes one of the most important decisions of the entire case.

DIAGNOSTIC CLINICAL REASONING

Shoe On, Shoe Off: What Has Happened to Your Ankle?

Ken's Experience, Part Two: Ken has decided that he will help Michelle by examining her ankle, with her permission. He wants to help Michelle understand what has happened to her ankle, ease her pain, and determine what level of care she needs immediately and in the next 24 hours. He wants to make helpful recommendations to Michelle; he is not at this moment interested in soliciting her as a patient.

Ken has decided to enter into a relationship with his neighbor, and although preliminary, the relationship has the elements of a patient–physical therapist relationship, even though the parties have been cast into these roles suddenly The next set of decisions calls on Ken to bring his clinical expertise, knowledge, and reasoning together to make correct clinical decisions and recommendations for Michelle.

*Pause now and note the clinical decisions that you would make if you were Ken, their purpose, and the order in which you would pursue them.

Now review Box 2.2, and determine if you have made similar choices. Ken has several types of decisions to make as this list indicates, but his primary concern might be to diagnose the patient's pathology. However, doing this prior to discussing with Michelle what he can offer her and gaining her consent would be unnecessarily risky in the context of this case. Some of Ken's options might raise ethical dilemmas. One such dilemma might be captured in his decision to take her shoe off or not for the purposes of examining her ankle. Taking the shoe off might cause more pain or injury, and once off, increased edema is a potential complication unless alternative strategies to control edema are available. But if Ken chooses not to take Michelle's shoe off, he may limit his ability to make an accurate diagnosis of the presence of a fracture.

Once Ken makes a diagnosis, he will be simultaneously assessing his confidence in the diagnosis, given any limitations on his examination or other data-gathering abilities. The greater the limitations, the lower Ken's confidence will be that he has made a correct decision, and this uncertainty will flow into his next set of decisions. Ken must make a prognosis as to whether or not Michelle requires medical care at a hospital emergency room, an urgent medical care center, or in a physical therapist's office on Monday morning.

Sometimes, when we are uncertain of our diagnosis of a patient's pathology and make either overly risky or overly cautious decisions, every other component of care can be affected, resulting in harm to the patient or unnecessary expense and use of resources (Fig. 2-1). Other times, a less accurate diagnosis of the pathology of an injury may not have negative consequences. For example, if you are incorrect in isolating the specific muscle in the group of wrist extensors that may be experiencing an inflammatory response in a patient with forearm pain, you are still able to proceed with your intervention decisions with some confidence that they will be successful regardless of which muscle is most affected, as your intervention will address the entire group.

We discuss more about possible mistakes in diagnosis and other aspects of care in Chapter 3 and possible reasons for these mistakes in Chapter 4. In Ken's case, if he is uncertain of his diagnosis, this will not deter him from making intervention decisions because the motivation to do something for Michelle is strong, as she has an acute injury. Groopman,³² in his text entitled How Doctors Think, suggests that emergency room physicians, when most uncertain, ask themselves the following types of questions, which might be helpful to Ken right now:

• Is this patient seriously injured?

• What is the worst outcome here?

• How likely is it?

Types of Diagnostic Decision Making

Diagnostic decision making has been studied in physical therapy over the past 20 years, and researchers have found that it shares many similarities to the processes used by physicians, although the nature of the diagnoses made by each professional are different. The research has identified two major processes in decision making: hypothetico-deductive reasoning and pattern reasoning.

Elstein et al³ identified a model of medical decision making, termed the hypothetico-deductive model of reasoning, as the dominant strategy used by physicians to cluster patient symptoms into diagnostic categories of pathology. Often termed backward reasoning, this process consists of gathering cues, generating hypotheses that fit one or more of the cues, and then testing the hypotheses to determine the best fit for the diagnosis. This process is used to turn an unstructured, unfamiliar, or complicated patient problem into one that can be addressed with the use of one or more hypotheses. In physical therapy this process is commonly seen in the practice of reevaluation, where the prevailing hypothesis can be continually confirmed or disconfirmed.¹³

Pattern Recognition or Forward Reasoning

For experienced clinicians, researchers also identified the importance of contextually developed clinical knowledge in reasoning. They found that specialists combined highly organized and accessible knowledge with mental processing during the diagnostic process. In situations where the patient's symptoms were familiar to the physician, reasoning was more rapid or automatic. The specialists utilized knowledge developed from their experiences to build recognizable patterns of patient symptoms.³³ This type of diagnostic reasoning is termed pattern recognition, or forward reasoning, and is typical of more experienced clinicians whereas the hypothetico-deductive, or backward, reasoning is more common amongst novices.

Choosing Between These Processes

Student physical therapists are educated with an emphasis on mastering basic and applied research knowledge so that they might identify in their patients the likely pathology, functional limitations, and impairments that should be the focus of their physical therapy care. This type of knowledge may, however, fail a novice clinician when faced with a complex clinical scenario that was never discussed while acquiring technical knowledge. With increasing experience, expert practitioners in physical therapy were found to have combined their basic science knowledge with a growing rich resource of practical knowledge. Using techniques called reflection in action,³⁴ they then transformed their knowledge into a rich pool of information to inform clinical decisions. Reflection in action requires the practitioner to analyze cases that caused doubt and to rethink the choices made to determine what worked and why it worked. Schon³⁴ suggests that without reflective practice, logical or rational problem solving alone will never be enough to enhance clinical reasoning.

These two primary types of diagnostic reasoning (backward and forward reasoning), as well as reflection, have been identified in studies of novice as well as expert physical therapists.^13,18,35 All agree that physical therapists are not as concerned with a correct diagnosis of pathology as they are with a correct understanding and labeling of the significant impairments and functional limitations the patient is experiencing. Research by Jones et al³⁶ suggests that physical therapists use a hypothesis-oriented approach while working collaboratively with the patient to understand the functional losses most important to the patient. Edwards et al³¹ studied the clinical reasoning strategies of expert physiotherapists in Australia and identified two types of diagnostic reasoning strategies. Diagnostic reasoning using either forward or backward strategies resulted in the identification of physical disabilities and impairments important for guiding the selection of interventions. Narrative reasoning was defined as important in the diagnostic role of the PT as the physical therapist attempted to understand the patient's story, illness experience, context, beliefs, and culture.

If Ken happens to specialize in orthopedic or sports physical therapy, he may move quickly to help Michelle and, with a careful examination of her ankle and well-honed questions about her symptoms, be able to use a pattern recognition clinical reasoning strategy to form his diagnosis. If he specializes in cardiac rehabilitation, he may take a more deliberate route in both his examination and questioning, testing his hypothesis about the likelihood of a lower limb fracture against each finding. In either case, Ken will be able to understand the impact of his diagnosis on Michelle's pain and loss of function over the days to come.

PATIENT MANAGEMENT REASONING

Lisa's Experience: Lisa, a certified clinical specialist in neurological physical therapy, practices in an integrated service delivery facility that provides acute inpatient rehabilitation services, subacute inpatient rehabilitation services, and home health physical therapy services to patients who have significant rehabilitation needs. The philosophy of this health care center is to provide patients with continuity of care from the same practitioners as they progress to lower levels of care requirements and finally to their homes. Lisa is fortunate to be able to follow her caseload of patients who have experienced stroke and traumatic brain injury for a long time and across settings. Lisa believes that this extended contact with her patients helps her clinical reasoning skills as she is able to assess the outcomes of her decisions directly and for a longer-than-average period. She practices with an interdisciplinary team of rehabilitation providers who respect each other's expertise and collaborate on many clinical decisions. For example, Lisa will be presenting Mr. Simon's case to the team today as she has identified a complication of pusher syndrome in his recovery from a right cerebral artery stroke. Lisa has developed a set of intervention techniques but she is reevaluating her decision for discharging Mr. Simon to his home given this complication and the difficulty of teaching the patient and his family safe transfer techniques.

Diagnostic reasoning has received much attention in the medical literature, but more recent research in health professions such as nursing, occupational therapy, and physical therapy has unearthed a wealth of perspectives on the complex decisions required of practitioners who spend a significant amount of time with a patient across episodes of care. A physician may repeat his or her diagnostic reasoning skills every 15 minutes throughout the day with different patients. However, physical therapists and other rehabilitation personnel will practice other types of clinical reasoning during their extended encounters with each patient, often over frequent visits. Table 2.1 contrasts some of the aspects of medical decision making with those found in nursing and occupational therapy.

Lisa's experience with Mr. Simon provides a stimulus for reflecting on the many areas of clinical reasoning in physical therapy patient management.

*Pause now and reflect on Lisa's clinical decisions in Mr. Simon's case. How many different types of decisions for a physical therapist can you identify in this scenario?

Now review Table 2.2, which shows several categories of clinical reasoning related to patient management identified from research on nurses, occupational therapists, and physical therapists. Many of these categories will sound familiar to physical therapists because they describe the types of thinking that is required of them each day. Table 2.2 defines each type of clinical reasoning and gives examples of each type that could be drawn from Mr. Simon's case. The breadth of the categories of patient management reasoning identified speaks to the range of reasoning skills required of a physical therapist.

Continued study of patient management reasoning is crucial to improving understanding of clinical reasoning strategies. As we will discuss in the next two chapters, we are prone to clinical reasoning errors in both diagnostic and patient management. For example, evidence from actual patient experience provides the basis for studying the typical errors physical therapists make in predictive reasoning. Are we overly optimistic in predictive reasoning, perhaps ignoring evidence about typical recovery patterns? The identified categories of patient management reasoning can help us understand which types of reasoning are most easily learned, which types are most valuable to the patient experience, and which types of reasoning are most challenging when practicing outside our area of expertise.

Developing or Improving Clinical Reasoning Skills

How does this research help you develop or improve your clinical reasoning skills? First, these descriptions of clinical reasoning strategies identified in studies of physical therapists can facilitate discussions between novice and experienced clinicians. Questions such as the ones in Table 2.2 can help physical therapists reflect on their clinical reasoning abilities and identify areas in which they would like to improve. Ask yourself which types of decisions come easily, and which types of decisions are typically more difficult? Second, describing these types of reasoning allows us to articulate the complexity of prolonged interactions with patients. Most researchers found that physical therapists, as well as nurses and occupational therapists, are adept at thinking or reasoning along multiple lines of thought simultaneously. Mattingly and Fleming identified this behavior in occupational therapists and termed it "the therapist with a three-track mind" (p. 133). Such therapists can monitor their procedural reasoning as they select treatment procedures and their interactive reasoning as they assess their communication with the patient and, using conditional reasoning, they can pull it all together to create an individualized treatment session for the patient.¹⁰ Jensen et al,¹³ studying 12 physical therapist experts across four specialty areas, also found this fluidity during treatment sessions with patients. To the observer it is often difficult to determine which type of reasoning is occurring at any particular moment; this makes mentoring students in the clinical setting more challenging. Students can see what the clinician is doing, but they cannot see what the clinician is thinking. The categories can help mentor and student alike make clinical reasoning more explicit and thus allow both to improve their thinking.

CONCLUSION

To understand clinical reasoning in physical therapy requires one to consider the context in which physical therapists practice, the types of decisions that they make, the type of reasoning they use, as well as the development of these skills across a professional career from student to novice to expert practitioner. We have illustrated in this chapter that our limited research suggests that physical therapists share some common thinking processes with other health care practitioners. Jensen et al^13,14,15,16 were the first to propose a model of expert practice in physical therapy by contrasting the thinking of novice and expert physical therapists. Clinical reasoning was identified as a key component of the model of expertise. The physical therapist who achieves this level of practice has developed a context-rich, multidimensional base of knowledge through reflective practice, paying close attention to the patient as a source of knowledge. Their clinical reasoning processes are characterized by a strong value for and use of collaborative reasoning with patients for management decisions. They can efficiently recognize patterns in diagnostic reasoning by matching their patient's examination findings to their mental library of patient illness scripts. What is more difficult for expert and novice clinicians alike is describing their thinking processes, or being aware of the nuanced changes that accompany their experience. Research in this area is vital to the profession so that clinicians can develop a lifelong practice of enhancing clinical reasoning abilities.

Physical therapy experts also confirm that patients' values as they relate to their own health care should constitute the primary focus of the clinician's reasoning processes. The second section of this text will delve further into strategies to uncover and value patient preferences in all clinical reasoning.

SELF-ASSESSMENT

1. Identify some experiences that you have had that are similar to the cases presented in this chapter and then use the questions in Tables 2.1 and 2.2 to reflect on your decision making.

2. Think about a patient you have treated recently and identify three decisions you made about that patient's care. If you haven't treated any patients, think about a patient whose care you have observed. Then use the categories in Table 2.2 to think about what kind of reasoning you used in making these decisions.

3. After these reflections, try to describe your own personal decision making process. Some of us are more graphically oriented than others, so perhaps you would prefer to draw this process rather than write about it.

CONTINUED LEARNING

Choose one of the articles in the reference list, read it carefully, and share what you have learned with your colleagues.

REFERENCES

INTRODUCTION

We all hope we have chosen health care practitioners who have a good amount of expertise, who can get things right. The issue of a practitioner's clinical expertise has been reinforced as crucial to good decision making by authors throughout the development and implementation of the evidence-based medicine movement. Straus et al¹ define the clinical expertise element of evidence-based medicine as "the ability to use our clinical skills and past experience to rapidly identify each patient's unique health state and diagnosis, their individual risks and benefits of potential interventions, and their personal circumstances and expectations."^{1, p1} Similarly, physical therapists must make important choices in practice not only in determining the correct diagnosis and intervention but in performing many hands-on interventions, predicting outcomes of care, setting goals with patients and families, and coordinating resources and personnel to maximize the functional outcomes for the patient. Expertise in all these elements of practice reassures us that our physical therapist will also make good decisions regarding our care and get things right.

However, in clinical practice we all acknowledge that sometimes things don't go right. Sometimes cognitive skills, manual skills, and past experiences are insufficient for the task at hand. In these situations, clinical practice errors of varying magnitudes can occur. Of course, poor outcomes in clinical practice can also occur in a random fashion, as biological events are probabilistic in nature. More often, errors occur when the current level of scientific knowledge used to inform choices is inadequate, or when the clinician lacks knowledge, violates the patient trust in a purposeful manner, or uses faulty logic. Whatever the cause, an examination of clinical practice errors provides a source of knowledge to help us improve our level of practice.

Providing physical therapy care in a manner that reduces errors and increases patient safety should be the goal of every clinician, but it is not an easy goal to attain. Weingart² states that the 1999 Institute of Medicine report To Err is Human³ provided evidence that medical errors pose daily risks throughout the health care system in this country but also elucidated the difficulty in estimating the magnitude and attribution of these risks. While injury of various types may occur to patients under our care, determining that these injuries are a result of clinician error is a complicated process. Weingart² suggests that harm can result from many complex factors, including: the natural history of the patient's disease, comorbidities, the risks or negative effects of selected interventions, clinical misjudgments, technical performance errors, and bad luck. Given this complexity, it is not surprising that independent reviewers of patient safety events are able to identify when a harmful event has occurred but are not good at analyzing the degree of harm caused in the event nor if the event is a result the presence of a medical error.⁴

In this chapter we will provide an overview of research on errors in medicine, physical therapy, and occupational therapy. An understanding of the type, frequency, causes, and harm associated with errors is a good first step in our attempts to increase patient safety. We will examine errors and the causes of errors, for example, faulty clinical reasoning. In the previous chapter we identified the typical reasoning strategies used by physical therapists in making many types of decisions. Here, we will focus on learning how we might categorize and learn from errors of clinical reasoning. Our increasingly complex health care system can contribute to a clinician's sense of uncertainty when the clinician is making difficult decisions. The ability to recognize those situations in which our uncertainty is heightened should be an important component of reflection in our clinical reasoning process.

DESCRIBING MISSES AND MISTAKES IN HEALTH CARE

Although bad medical outcomes that happen to famous people garner much public attention, researchers find that good epidemiological data on errors is limited. Weingart et al⁵ report that the majority of information on medical errors is derived from a few academic medical center studies, and in general the literature suffers from a lack of standardized research methods, definitions, and systems for monitoring and reporting errors. Their review of this literature does reflect increased risk for those patients who are sicker, over 64 years of age, undergo multiple interventions, remain in a hospital longer, and are treated through an emergency department or in an intensive care unit. The estimates of medical errors outside of the hospital setting is low, with prevalence rates for medical errors in physicians' offices found to be 8% to 9% of all errors.

Methods for monitoring and analyzing medical errors in these studies range from medical chart reviews, adverse drug event records, clinician self-reports, and observational studies. Adverse events are defined differently, allowing underestimation of the error rate. For example, when a mistake is made but caught prior to injury to the patient, how should it be labeled? In many settings these events are termed near misses, rather than mistakes or errors, an event that was recognized and recovered in time to prevent an adverse event. Much less is known about near misses than about true adverse events. In 2000, Weingart et al⁵ found several additional terms for categorizing medical errors. Some examples are:

• Surgical adverse events versus medication errors

• Preventable versus unpreventable errors

• Diagnostic errors versus therapeutic mishaps

• Fatal versus life-threatening versus serious errors

• Errors of omission or commission

Clearly any efforts to improve the rates of medical errors are dependent upon a common language for describing and reporting them.^4,6

Creating a Taxonomy

In 2005, researchers with the Joint Commission, previously known as the Joint Commission on Accreditation of Health Organizations (JCAHO), conducted a systematic review of the literature on patient safety terminologies and classifications of errors.⁷ Their work resulted in a proposal for a standardized patient safety event taxonomy. The goal was to create a multidimensional taxonomy that could be used in various settings and by various practitioners to report incidents of patient safety. The resulting taxonomy includes five primary classification groups, categorizing data that will describe the following aspects of a patient safety event:

• Impact

• Type

• Domain

• Cause

• Prevention and mitigation of the event

Impact

The impact schema describes the degree of harm caused to the patient in medical and non-medical contexts. The levels in this taxonomy, shown in Table 3.1, include psychological and physical medical injury, and progress from no harm, to temporary harm at various levels, to permanent harm, to death or profound psychological harm.

Type

The taxonomy for type of error includes three categories:

1. Communication

2. Patient management

3. Clinical performance

The inclusion of these three categories illustrates the comprehensive nature of the delivery of patient care and the wide scope of activities in which errors can occur. Communication errors could result from inaccurate and incomplete information, questionable advice or interpretation, questionable consent or disclosure processes, or questionable documentation. Patient management errors can be categorized as emanating from questionable delegation, tracking or follow-up, referral or consultation, or use of resources. The clinical performance errors are grouped into three temporal categories, pre-intervention, intervention, and post-intervention (Table 3.2). Table 3.2 also illustrates the various levels of clinical performance errors, including determining incorrect diagnoses and prognoses.

Domain

The domain taxonomy provides a description of the setting of the event, the staff involved, certain patient characteristics, and the targeted type of patient care intervention In this JCAHO taxonomy, descriptions are included that will provide information on patient safety events occurring in several locations in which physical therapists practice, including the rehabilitation component of a hospital, nursing home, rehabilitation facility, and practitioner office. The listing of categories for staff includes physical therapists and occupational therapists. Intervention types include therapeutic, diagnostic, rehabilitation, preventive, or palliative.

Cause

The cause taxonomy has two main categories: systems (organization and technical) and human. Systems errors are often not proximate to the patient-clinician relationship but can play a significant role in the context of the safety event. Examples of organizational system causes are the culture—for example, the value for patient safety—or organizational management—for example, the monitoring of unsafe practices. Organizational processes can also affect patient safety, including productivity demands and safety oversight procedures. Technical systems errors typically are found in equipment or materials design, malfunction, or availability. Human errors are categorized as those attributable to the patient, the clinician, or to individuals external to the patient-clinician relationship. Patient errors result from behaviors that are not controllable by the clinician. External human failures are behaviors that are beyond the control of the entire organization. Practitioner errors can be skill-based, rule-based, knowledge-based, or unclassifiable. This classification scheme can be used to identify prevention and mitigation actions that can decrease the rate of medical error by selecting from a variety of procedures that will trigger alerts or alarms for practitioners prior to implementation of services. Techniques are suggested, for example, to improve the accuracy of patient identification, reduce acquired infections, and prevent medication errors.

The authors of the JCAHO patient safety event taxonomy recommend its usage in a wide variety of health care systems, including electronic medical records, to enhance the quality of data on medical errors available to researchers on health care quality and to policy makers. However, they recognize the limitations of such a system to fully capture the nuances of accurately describing medical errors in the following statement:

Weingart⁴ recommends that such a taxonomy will enhance our detection of patterns of medical mistakes; he also acknowledges that "[i]This not obvious, however, that a classification scheme can reliably capture the 'story' of the event and how it unfolded."^{4, p.93}

COGNITIVE ERRORS IN PHYSICIAN DECISION MAKING

Several studies in medicine attempt to focus specifically on the cognitive errors made by physicians as they practice clinical reasoning. Such studies have created categories of cognitive errors by attributing incorrect choices during simulated patient encounters to the patterns of thinking observed in subjects. Young et al⁸ studied the cognitive processing abilities of 15 residents as they responded to a recall task and two clinical decision scenarios presented as sample cases by a researcher. A variety of cognitive processing errors could be identified, including:

• Knowledge factors:

  • Inaccurate knowledge or incomplete knowledge about some aspect of a complex system, defined as buggy knowledge

  • Knowledge calibration errors—seen when the residents could not identify gaps in their own knowledge;

• Simplifications—using rules of thumb in complex situations;

• Attentional dynamics such as situational awareness—the inability to track processes and data in time or across more than one topic

• Fixations—inability to reassess a situation in light of new information

Typically, residents with less experience make a greater number of cognitive errors than those with more experience. Studies of cognitive reasoning such as this are often conducted in a laboratory setting, not a clinical setting. (See Chapter 4 for a more detailed discussion of cognitive errors in decision making.)

Studying cognitive processing skills in decision making is one approach to avoiding reasoning errors. Students who struggle with a repeated deficiency in one aspect of knowledge transfer or attentional dynamics may seek help with specifically planned learning experiences to enhance their thinking skills.⁹ A student or novice clinician might master these cognitive processes in the classroom but have difficulty demonstrating sufficient performance in a busy clinical environment. The guidance of a more experienced clinician can assist the novice to identify errors such as his or her gap in knowledge or the tendency to ignore new exam findings. However, these analytical experiences are often difficult to incorporate into a busy clinician's schedule.

Kempainen et al¹⁰ suggest that improvements in clinical reasoning will only happen if physicians have knowledge of cognitive processing and use it to internally monitor their thinking. This requires a commitment of time for reflection on the quality of one's thinking in situations where no known errors have been committed as well as when they have. Borrell-Carrio and Epstein¹¹ have called for physicians to increase self-awareness of their errors in practice through reflection on their abilities to tune or calibrate both their thinking and their emotional responses to the decision confronting them. The habit of self-questioning during clinical work is recommended to increase one's awareness of the personal conditions under which they function during difficult clinical decisions. Mamede et al¹² suggest that reflective physicians are those who can examine a decision from a broader perspective and can tolerate or even enjoy the internal tension created when working with complex cases.¹² Box 3.1 presents some provocative self-reflective questions drawn from the work of these authors.

ERRORS IN PHYSICAL THERAPY AND OCCUPATIONAL THERAPY PRACTICE

The best description of practice errors has been provided by the medicine, nursing, and pharmacy professions, and most of this work describes medication errors.¹³ While not as comprehensive, a few authors have reported useful information about the types of errors made by occupational therapists and physical therapists. Scheirton and Mu^14,15 have studied errors in occupational therapy practice in physical rehabilitation and geriatric settings, using focus groups and survey research. The majority of practice errors identified by occupational therapists were preventable errors that occurred during patient interventions. Misjudgment and insufficient experience and preparation were identified as potential causes of errors made by the occupational therapists in these studies. These researchers have further investigated the response to errors among these therapists, finding that they reported a significant amount of emotional distress. Coping strategies were used to help the therapist recover from the effects of the error, including: a corrective action plan for the therapist, volunteering to spend extra time with the injured patient, and making constructive practice changes to prevent a similar error in the future. These authors have included physical therapists in some of their focus group research on patient safety and reported an ethical analysis of six cases of error committed by physical and occupational therapists in a recent report.¹⁶ They identified that the cases illustrated what they termed technical and moral errors; analysis of moral errors focused on the ethical responsibilities of the clinicians in each case.

Much less research has been conducted on the types of errors that characterize physical therapist practice. In 1992, Deusinger¹⁷ reported the results of telephone interviews with 119 physical therapists in Missouri who agreed to discuss clinical errors in their practice careers. The author attempted to develop a typology to categorize the reported errors as to:

• Type—defined as an error of omission or commission

• Action—defined as decision making, interpersonal communications, technical or psychomotor skills, or cognitive factors related to knowledge

• Consequences—defined as potential versus actual, and as social or physical

The typology demonstrated reliability only in categorizing the error type and the consequences. Sixty-one different tasks were described by the subjects as associated with errors in practice, 13 of which accounted for 65% of the errors. The majority of errors occurred during intervention tasks (29%) such as gait, modalities, and exercise. Eighteen percent of the errors were attributed to mistakes in planning the patient's program or delegation of care. Only 7% of errors were reported during the action of examination or evaluation, perhaps attributable to the time frame of data collection (1986). Subjects were asked to report the likely cause of the error, and these data appear to reflect findings similar to those in medicine, with the most frequent causes reported including inadequate knowledge or clinical experience and cognitive failures including inaccurate assumptions and failure to consider consequences of the decision. Environmental factors were also identified, including time pressures and inadequate staffing or supervision.

Errors and Error Prevention Across Patient/Client Care Management Model

As seen in the research efforts in medicine, the types of errors and the consequences of these errors should be of specific interest to physical therapists who wish to improve their clinical expertise. One perspective for studying the type of errors physical therapists make is to consider the management model developed by the APTA,¹⁸ discussed in Chapter 1. Using this model to evaluate errors in practice may focus attention on the elements of patient management activities that place patients at the most risk. This perspective may also be helpful for analyzing cognitive errors or problems with thinking, as the patient management model represents actions performed with the patient in a temporal fashion. For example, what are the consequences for clinical decisions about interventions, if diagnosis is incorrect?

Examination

The physical therapist patient management model begins with clinical decisions about how to examine the patient. The therapist must conduct an appropriate history interview, select which systems to screen and examine, and then select the correct tests and measures to perform with the patient. Each test must be performed accurately and interpreted correctly to inform the evaluation of the patient's condition. This is not only the first interaction between patient and therapist but, some would say, the most important element of the model in which to strive for clinical expertise. Groopman¹⁹ suggests many cognitive errors that happen during the medical interview derive from poor physician communication skills. A few of these include failing to actively listen to the patient, ask open-ended questions in a manner that communicates interest in the patient, and follow up on aspects of the patient's story. Roter et al^20,21 found that physician's use of good communication skills can enhance the patient's involvement in problem posing and problem solving collaborations during the interview. In Section Two Coulehan and Block discuss ways to improve the interview process to overcome these types of errors and Gordon, Nixon-Cave, and Johnson all provide information about exploring patients' values as a component of the therapeutic alliance.

Tests and Measures As clinicians with over 800 physical performance tests and measures at our disposal, physical therapists face a challenge in selecting a test with the best diagnostic properties for each patient problem reported. Evidence regarding the diagnostic accuracy of many of these tests has increased significantly over the past decade.^22,23,24,25 Physical therapists who increase their knowledge of these studies, especially for the most frequently used tests in their practice, will enhance their ability to avoid the errors of selecting inappropriate tests, in applying and interpreting the tests, and in over- or underutilization of tests. The consequences of errors in decision making during the examination phase include overtesting that can cause increased pain to the patient, undertesting that can cause the physical therapist to make an incorrect diagnosis, and poor communication that can create an adversarial patient-therapist relationship.

Evaluation

Once the examination data have been collected, the physical therapist evaluates the data to determine the patient's diagnosis. These data will be used in developing a prognosis and a plan of care. Diagnostic errors may result in incorrect or incomplete diagnoses, such as when one aspect of the diagnosis is missed completely. How might these errors occur? Perhaps the physical therapist misinterprets the data from the examination due to not being able to conduct the test accurately; for example, a test appears to produce normal results but the clinician may not have had sufficient strength to administer the test on a particularly large patient. Other sources of cognitive diagnostic errors include incorrectly identifying a pathological structure or movement pattern or miscategorizing the nature of the patient's movement impairments. Diagnostic errors also might be a result of cognitive, communication, or technical errors occurring during the examination phase of patient management. Thus an error occurring earlier in the patient management process will likely influence the next phase of decision making. For novices, diagnostic errors may occur due to the inability to put all the examination data together into a cogent hypothesis that fits.²⁶

The ability to see patterns emerging from examination data is developed through reflective practice by expert physical therapists.²⁷ Experts are skilled evaluators, as they combine their content knowledge, their knowledge of the patient and of self, and their practice knowledge to make diagnostic decisions. The consequences of a misdiagnosis in physical therapy are not precisely known, but may include inappropriate selection of interventions that may cause injury to the patient or extend the time needed to reach a successful conclusion of care, delayed referral of the patient to an appropriate practitioner for further diagnostic evaluation, or labeling and treating patients unnecessarily when misidentifying their movement patterns as abnormal.

Prognosis

Prognostic errors occur in practice when the physical therapist does not accurately predict the natural course or progress of the patient's pathology or impairments, or the recovery of functional performance, or future disability level. Perhaps a client might be advised to return to work or sports activity at too early a point in his or her recovery. Prognostic errors might also result in either over- or underutilization of health care resources of all types, including durable medical equipment or disposition placement for continuing care. Avoiding prognostic errors is difficult for physical therapists because our health care system partitions the continuous care of our patient among more than one physical therapist, passing the patient along to the physical therapist practicing in the acute care setting, the home health setting, or the outpatient setting. These disconnected care patterns make it difficult for the physical therapists to gain information on patient outcomes that can inform their decisions during their interactions with the patient. Physical therapists must work to discover methods to monitor the long-term outcomes of their patients in order to build the experiential knowledge base that will enhance future prognostic decisions.

Intervention

Errors that occur during physical therapist interventions are perhaps the most obvious clinical practice mistakes because they may risk the physical safety of the patient. In addition to the psychomotor skills required of many clinical interventions, the physical therapist must know which intervention to choose, when and where to apply them, and with what force. Any misjudgment of one of these aspects of the intervention can change the right treatment into a dangerous treatment rather quickly. An accessible store of knowledge of past patients' reactions to interventions, coupled with close monitoring of the patient during treatment, may reduce the occurrence of intervention errors in clinical practice.

Outcomes

Errors that may accompany the outcomes of physical therapy care can best be described as a mismatch between the outcome measures selected by the physical therapist and the patient's goals or circumstances. The physical therapist may believe the care delivered was successful based upon an improvement in the patient's pain scores, but the patient may have functional goals that were never identified or addressed. The assumption that a patient's function will automatically return once pain is reduced can represent an error of faulty logic on the therapist's part. If the therapist does not fully consider the patient's circumstances, this may also lead to a mismatch in treatment outcomes. For example, the therapist may underestimate the aerobic or muscular endurance required for mobility by a patient living in a very rural environment.

Implications of errors in Physical Therapy Practice

We are just beginning to understand the importance of learning about practice errors in the physical therapy profession. It can be a difficult topic to discuss, as physical therapists cannot avoid their natural human tendency to deny fault, blame others including the patient, and refuse to acknowledge that an error has been made, whether or not injury to the patient has occurred. Ethicist Ruth Purtillo²⁸ reminds us that despite the best training physical therapists are not immune to situations where harm occurs. She then challenges the profession and individual clinicians to confront tendencies to deny the errors we make, to acknowledge them to ourselves and to the affected patient, and further to offer an apology and request forgiveness. Only when these steps in response to an error are taken, suggests Purtilo, can the therapeutic relationship between patient and therapist be restored.

The benefits of studying errors in clinical practice are significant for both clinicians and the profession. Knowing more about the effects of errors or near misses in physical therapy can enhance our understanding of practice in ways we do not often pursue. For example, if a clinician gives a patient a diagnosis of a labral tear at the shoulder, without the assistance of diagnostic imaging, and treats her conservatively for instability, pain, and decreased function, but the patient actually has a bicepital tendonosis, will the error in diagnosis make any difference in the short- and long-term outcomes for this patient? Likely not, as the conservative treatment approach for both of these pathologies is similar. How precise are our diagnostic categories in physical therapy, and how precise do they need to be, given the interventional options? To move our science and practice to a higher level, we can study our diagnostic errors and identify which diagnostic mistakes carry the most risk for our patients. In some diagnostic groups, it is the missed diagnosis that presents the most risk, for example, failing to refer a patient with a suspicious skin lesion for further testing. In some diagnostic groups, it is the misdiagnosis that places the patient at risk, for example, diagnosing an ankle sprain versus a fracture of the talus.

Occasionally a clinician will conclude a course of physical therapy with a patient, fighting a nagging concern that neither the patient's diagnosis nor the intervention was quite right, and that the patient got better anyway. This might be referred to as a good outcome that is robust to both diagnostic and interventional errors. It also might be considered inappropriate treatment of a pathology that has a natural history of improvement without care. Or the good outcome might be the result of the placebo effect that results from the attention and care of the physical therapist and has little to do with the direct effects of the interventions chosen. Our practice would be enhanced if we knew which of the explanations was accurate.

DECISION MAKING UNDER UNCERTAINTY

Many of the errors that we have discussed can find their basis in the uncertainty that surrounds the data we use to make decisions. Those who write about decision making under uncertainty advocate improving our knowledge by learning more about a few key variables.

Expected Value of Alternatives

One variable is the expected value of alternatives. For example, knowing whether a patient prefers more normality by learning ambulation after a serious stroke or faster functionality by using an electric wheelchair can help us decide more accurately what our course of intervention should be. Also, knowing the level of risk aversion or risk seeking our patients and we are comfortable with can also reduce errors. It is especially important to know if there is a disparity between our patient's preferences and our own.³⁰ As stated by an expert pediatric physical therapist, "I think one of the things I have learned over the past couple of years… is that the parents' goals are the goals, and the child's goals are the goals. It's not that the therapist's goals are the goals."²⁷

Concept of Probability

But perhaps the most important variable we must understand to reduce error is the concept of probability, along with the actual probabilities associated with the options in each clinical decision. The concept of probability helps us recognize that our decisions are not based on certainty. This by itself improves decision making. And identifying specific probabilities improves it even more.

Base Rate

There are a few pieces of data that help us understand probability. For example, we first need to know the base rate of the event with which we are concerned: just how many people have this problem? Base rate is measured by both the prevalence (how many people have this problem now) and incidence of the problem (how many people get this problem in any given time period).

In addition to base rate, we also need to know the likelihood that an intervention will lead to either an improvement or decrement in the patient's status. These measures include things such as absolute and relative risk. These types of data and the statistics that analyze them are discussed in the epidemiology literature.^31,32 We discuss more detailed ways to understand the research on specific probabilities in Chapter 12. Chapter 4 contains many examples of the faults in decision making that we can all experience. Understanding the concept of probability and knowing actual probabilities can assist us in avoiding these faults.

CONCLUSION

We all make mistakes! As we have discussed in this chapter, there are many sources for the mistakes that we make.

• First, we know that biologically based events are inherently probabilistic; almost nothing about illness and disease can be known with absolute certainty. The best way to deal with this source of error is to understand what is known about probabilities and take that into account in decision making.

• Second, the level of knowledge about the topic may not be currently sufficient; science may not yet have discovered the information necessary to understand what is happening. The only way to deal with this source of error is to remain vigilant about new knowledge as it appears.

• Third, knowledge might exist about the topic but we, personally, have not yet learned it. Again, vigilance about identifying knowledge new to us is essential.

• Fourth, we know what the best option for the patient's problem is but choose not to take that option because it is not in our best interest; we violate the patient's trust in us. The remedy for this source of error is to always make the ethical choice.

• Fifth, we do not think in a logical way about the information we have available to us. A remedy for this is to understand the errors we humans typically make in our logic.

The first two causes are beyond our control, as we cannot change the nature of biological events, nor can we know something not yet known. But we can work to reduce the impact of the latter three causes of errors. Section Three of this book provides many tools to overcome deficiencies in our own knowledge. Section Two provides ways to reflect on actions to overcome the temptation to violate the patient's trust. The remaining chapters in Section I identify ways to improve our logic in our patient care–related decision making.

SELF-ASSESSMENT

1. Think about a mistake that you have made in patient care. Apply the categories from the JCAHO taxonomy and then determine if this can lead you to identify ways to prevent similar mistakes in the future.

2. Use the questions listed in Box 3.1 to reflect on the characteristics of your own thought processes as you approach patient care.

3. Identify a specific patient whom you have treated lately and follow the patient across the patient/client management model. At each phase identify a potential error and then determine what the results of that error would have been.

CONTINUED LEARNING

Choose one of the many nonfiction books written about medical errors. After reading it, and after reading this section of the book, reflect on what can reasonably be learned from these books. Do they simply tell scary or interesting stories, or do they teach lessons that can be applied to your patients? The list below is just a small sample of these books.

• Banja J, Medical Errors and Medical Narcissism

• Geller ES, Johnson D, The Anatomy of Medical Error: Preventing Harm with People-Based Patient Safety

• Groopman J, How Doctors Think

• Nuland S, The Soul of Medicine

REFERENCES

INTRODUCTION

In this chapter, we will examine some important questions that health care practitioners should consider before stating a diagnosis or prognosis. Embedded in each diagnosis or prognosis are a number of assumptions, possibly shared, between the clinician and patient. In addition, there may well be biases that serve to undermine the accuracy of the prognosis, biases that may work entirely subconsciously on the clinician. In Chapter 2 we described two ways in which clinicians typically make decisions:

• the classic differential diagnosis known as hypothetico-deductive reasoning, or

• the expert's ability to use pattern recognition as a shortcut to a diagnostic or intervention decision.

Neither one of these decision making styles or other types of clinical reasoning is wrong in and of themselves, but each can lead to faulty reasoning and errors in judgment. Biases, defined as inaccurate beliefs that affect decision making,¹ coupled with any clinical reasoning strategy can result in errors. Can we ever free our clinical reasoning of biases? While the answer to that question is unavailable, we do know that if we do not recognize our biases, we cannot work to avoid their negative impact on our decision making. Let's examine the case example of Jack Watson.

In essence, each of Jack Watson's health care practitioners is saying, "Most of my patients are helped by this treatment." This is surely offered as an honest assessment of treatment efficacy, based on the practitioners' recall of personal experience. In turn, patients are no doubt encouraged by such a positive endorsement. They will have every reason to think that they are more likely than not to benefit from the treatment.

BIASES IN DECISION MAKING

In this chapter, we will discuss five types of biases. Any one of these biases can influence the clinician's assessment of the efficacy of treatment. In each case, we start by showing how it applies to the Hyalgan therapy decision. We will also cite other specific clinical examples, mostly from the practice of physical therapy.

Confirmation Bias

Suppose you believe Hyalgan works really well, and you want to learn more about your beliefs by talking to patients and former patients with joint problems. If some of these patients are doing better, you may be inclined to ask them if they took Hyalgan (since it fits with your model of what works). This is confirmation-seeking behavior, and we are all naturally inclined to engage in it. If, on the other hand, other patients are not doing better, you may not be so inclined, again because it doesn't fit with your model. You therefore assume they didn't try Hyalgan, and therefore never ask if they did. If you did ask, it would be an example of disconfirmation seeking. Disconfirmation seeking is not as natural as confirmation seeking.^3,4,5,6

But notice that you can learn just as much about the effectiveness of Hyalgan by asking for disconfirmation as by asking for confirmation. Suppose those with injections and those without had the same high success rate? If you only seek to confirm, you will conclude that Hyalgan was the cause. Table 4.1 summarizes this tendency. Patients in the first row, who are feeling better, are more likely to be asked if they have tried injections than are patients in the second row. The selective questioning serves to undermine the possibility for true learning, because the results from patients who are not doing better, or are doing even worse, are not taken into account. But if you seek to disconfirm as well, you might soon discover that there is no relative contribution being made to patients' well-being from the treatment you so earnestly believe in.

Consider again the statement, "Most of my patients are helped by this treatment." The next time you find yourself about to say this, ask how much you really know about patient outcomes when the treatment is not given. Is it possible that you are attributing to the treatment something that can just as easily be attributed to something else, such as nature's natural healing process?

Confirmation Bias in the Selection of Interventions

Confirmation bias can also affect a physical therapist's selection of interventions. Physical therapists who are strong proponents of a certain type of physical therapist intervention, for example, joint mobilization for musculoskeletal pain relief, may be susceptible to taking note of the times this treatment decreases a patient's pain, while paying less attention to the frequency with which pain is reduced when another treatment is used. When we have invested our time and resources in developing certain aspects of our practice, we are less likely to ask ourselves, "Should I take another approach to the care of this type of patient?" Perhaps our most difficult problem with confirmation bias is our belief that doing something is better than doing nothing. After all, our patient has often come to us because doing nothing has not worked. We are well advised to be aware of our own decisions that are most vulnerable to our avoidance of disconfirming evidence.

Follow-up Bias for Patients You Treat

There is usually incomplete information from patients who received our treatment. As a clinician, you should ask if you observe all the outcomes from all your patients and, if not, from whom you are more likely to learn the outcome. Consider for now just those patients who get the intervention. If you had a chance to learn the impressions of all of these patients, you would have an unbiased sample, but you may not hear from many of your patients. This lack of complete information could work either way. In saying "Most of my patients are helped by this treatment," you must consider the patients who have been treated who don't return for further evaluation or therapy. If you assume that all of these patients have not returned because they were helped by the Hyalgan and are now with reduced pain, then this will lead you to think the intervention is more beneficial than it really is. If, on the other hand, the patients who are not helped by the intervention are more likely to return, your impressions of the intervention could be the opposite.

Consider Table 4.2. This table classifies the n patients for whom the Hyalgan injections have been given. Your estimate of effectiveness may well be based on the ratio of (a/n_learn), which is the percentage of all patients from whom you have feedback, who have improved. The problem is that patients you have observed may have very different results, on average, from the patients you don't observe. You are estimating the ratio (n_improve/n), the overall improvement rate, with a biased sample of patients you have learned about, n_learn. In giving advice to a new patient, you must recognize this follow-up bias that derives from your own experience.⁸

In Table 4.3, we show how this might work out for four hypothetical physical therapists. Each one has seen 300 patients over the past 3 years who have had injections. Each one has also learned that 65% of those patients from whom they have learned the outcome report improvement.

The first two hypothetical practices shown in Tables 4.3a and 4.3b, have low learning (100 patients out of 300), while the second two hypothetical practices shown in Tables 4.3c and 4.3d, have high learning (200 patients out of 300). Furthermore, the improvement rate for those patients not seen varies across these four hypothetical practices. It is low in the first and third practice (30%), and high in the second and fourth (80%). Note that these variations lead to wide variations in the true overall improvement rate. The lowest overall improvement rate is 42%, in Table 4.3a, while the highest is 75%, in Table 4.3b. The point is that distortions in estimates of success rate can come from both incomplete samples and differences in improvement between those patients who give feedback and those who do not.

Finally, recall the physician who told Jack Watson that 75% of the patients improved with injections. The physician could be more susceptible to a bias caused by incomplete information than is the physical therapist. The physical therapist will typically have more ongoing contact with the patient and can learn over time whether an intervention seems to be helping. However, the physician's plan is to reconsider arthroscopic surgery in the event of treatment failure and no surgery if the treatment succeeds. If a patient does not return, the physician (possibly unconsciously) may count this treatment as a success. Yet many factors besides treatment success could lead patients not to return. Patients might reconsider the idea of having surgery and simply live with the pain. They might be upset that the physician recommended this treatment and go elsewhere for follow-up. They might move. The point is to think about follow-up bias whenever relying on your own experiences with patients. Those whose outcomes you don't actually learn about could have either better or worse outcomes than do those you do learn about. Multiyear clinical studies are often confronted with follow-up bias. One can imagine a successful intervention for which the patients in the control group, seeing no improvement, are less motivated to continue in the study.

Because physical therapy care is segmented in our health care system, patients are often passed from an inpatient physical therapist to an outpatient or home health physical therapist as the course of their injury or illness progresses. All physical therapists that transfer care of their patients to another practitioner are likely to be exposed to follow-up bias for many of the decisions they make for their patients. Imagine a physical therapist practicing in a hospital that makes an error in predicting a safe discharge location for a patient. Unless the patient is readmitted to the same hospital and the care system allows the PT to learn of this, the therapist has no ability to learn how to avoid that error of judgment in the future. In outpatient physical therapy care, we could benefit from structured follow-up of patients who cancel appointments and all patients who conclude a course of treatment so that we may learn the long-term outcomes of our care decisions. In the absence of real outcome information, our natural tendencies to fill the void, combined with our tendencies toward optimism or pessimism, may take over. At the least, we should be honest with ourselves and our patients in recalling what we know and what remains unknown about a specific clinical decision.

Spectrum Bias

Is this patient typical of previous patients who have tried the treatment? Are they more or less likely to be helped? If they are different in any important way, this is known as "spectrum bias," because it refers to how a given patient presents relative to the entire spectrum of patients who have tried the treatment.^9,10

Again consider the statement, "Most of my patients are helped by this treatment." Think about all the patients who have received the treatment. How does your patient compare with this entire spectrum? If some kinds of conditions, or clinical presentations, are more amenable to treatment success than others, where does your patient fit in? There may be good reason to think the response of your patient is unlike the response of the average patient. For example:

• You may think your patient is more or less likely to be compliant with all the other parts of the treatment plan.

• You may think your patient's condition is more or less likely to be helped. After all, Jack has both a chronically deficient meniscus and severe chondromalacia, and the injections may work best for those with only chondromalacia.

• You may even be less sure of the underlying problem for your patient than for the average patient who receives the treatment.

• You might have a practice that sees lots of difficult cases, so difficult cases are more likely to be referred to you.

Perhaps you are fortunate enough to have good evidence from a controlled clinical trial. Even here, you must be careful to see how your patient compares with those in the trial. Consider the evidence from an investigation of Hyalgan described in the Hyalgan labeling.² This study was a double-masked, placebo- and naproxen-controlled, multicenter prospective clinical trial of Hyalgan as a treatment for pain in osteoarthritis of the knee. One issue is whether your patient would meet the entry requirements. But another is where your patient fits into the baseline level of pain spectrum. Hyalgan showed some improvement relative to the controls, but the relative improvement might have been greater for some baseline pain levels than for others. The published results of the study make it impossible to see where these relative differences are, so, as a clinician, you are left only with a sense of caution about how your patient will fare.

Trying to avoid spectrum bias can paralyze a clinician's thinking. While each patient is unique, we are serving that individual patient well when we consider how similar her or his case is to the evidence available to us for our choices. If we limit ourselves to consider the most likely predictive factors that recommend one treatment over another for a group of patients similar to the patient we are currently treating, we will avoid the errors that accompany spectrum bias.

Spectrum bias should also be considered as we sort among the various diagnostic tests available to us. The dichotomization of outcomes and groups that is typically used in sensitivity and specificity analysis for diagnostic test accuracy promotes a problem of spectrum bias in that the intermediate-range subjects are often not included in the analysis. Section Three provides more information about many of these issues.

Wishful Thinking or Value-induced Bias

We all, at one time or another, engage in wishful thinking. And our hopes can sometimes influence our beliefs. We want to believe that our diagnoses are accurate and that our treatment recommendations will work, and this may color our interpretation of what we see and what our patients tell us.

It is useful to distinguish between two kinds of wishful thinking: self-centered events and events in which the world at large is seen optimistically.¹¹ In the former case, one may think, "If I try this intervention, it will work well, even though the evidence shows it doesn't work so well when others try it." There is an illusion of control, not unlike the lottery ticket owners who place higher value on tickets they have personally selected than on tickets they have not selected. In the latter case, many people are simply too inclined to be optimistic generally, rather like Dr. Pangloss, the self-proclaimed optimist in Candide.

In applying the idea of wishful thinking to medical decision making, some researchers have labeled this phenomenon as "value-induced bias."^12,13,14,15 With value-induced bias, the value or importance of a medical outcome can influence judgments of its likelihood of occurring. For example, smokers may convince themselves they are especially immune to the harmful effects of smoking (self-centered wishful thinking) or that the overall data on smoking is inconclusive (general optimism).

This can apply to our patient contemplating taking Hyalgan. Because we want it to work, we may unconsciously exaggerate our view of its success rate in the overall population (general optimism). Alternatively, we may unconsciously feel that we somehow have a better success rate when we support it than when it is recommended by others (self-centered wishful thinking). Value-induced bias could even lead us to overestimate the likelihood of success for treatments with high transaction costs. For example, a treatment that is expensive, and has significant side effects, or that involves a cross-country trip, might be judged to have a higher probability of success. In the present case, if you know you have a patient for whom the $400 monetary cost plus the time taken from work to get the treatment represents a high burden, you might place greater faith in the treatment than you would for a patient for whom this is not a heavy burden.^14,15

The next time you think, "Most of my patients are helped by this treatment," ask yourself if you might be influenced, at least in part, by wishful thinking. Yes, we really want our patients to be helped by the intervention. Yes, we don't want patients to pay for unnecessary treatment. But we should not let overall optimism or self-centered wishful thinking color our beliefs. Expert clinicians know that the value or importance of finding the right intervention for a given patient cannot, by itself, influence its likelihood of success.

Physical therapists may find themselves vulnerable to this bias when successive treatment attempts have consistently failed. Each subsequent treatment decision, prefaced by "let's try this," carries an increased hope for a successful outcome. Value-induced bias may be introduced in the clinical decision process when someone other than the PT or the patient recommends the course of action. If you find yourself thinking, "I'm doing this against my better judgment," ask yourself whose judgment has influenced your decision and what real evidence is being presented to support the decision.

Ease of Recall

Our estimates of the likelihood of a given event are often based, at least in part, on how easily we can recall instances of that event or similar events. Events that are easy to recall are judged to be more common than those that are hard to recall. Often, this is not a bad rule of thumb. After all, commonly occurring events are the ones that are more likely to be recalled and hence more likely to be judged common.

The problem occurs when ease of recall doesn't line up with how common the event is. Some highly unusual events might get undue media coverage, leading one to believe they are more common than they are. A recent experience may be more likely to be remembered and thus weighed more heavily. Vivid, or highly salient, events are harder to forget. The availability bias^16,17,18 is the human tendency to overestimate probabilities of events associated with these overexposed, recent, vivid, or salient events. Perhaps the most famous example is mortality from shark attacks. The United States averages only about one such death per year,¹⁹ yet many of us might guess a much larger number based on coverage in the media and the image presented by a shark attack.

The same goes for cause-and-effect relationships. For example, our recommendation about Hyalgan to a patient might be unduly influenced by a particularly dramatic or recent success (or failure) that actually represents a low-probability association. If our 400-pound patient with 20 years of osteoarthritis of the knees found complete pain relief after these injections, this unlikely outcome could make an impression on us. Because it is at the forefront of our memory, we place undue weight on it in estimating treatment success for our next patient.

Ease of recall does not necessarily mean that a probability estimate is biased. In many cases, common events are experienced more frequently and make a greater impression, and this can be helpful in judging the likelihood of future such events. But it is when the availability of information is not related to actual frequency that we must be most cautious. Detmer et al²⁰ asked surgeons to estimate the in-hospital mortality rate for their entire service. Surgeons from high-mortality specialties gave estimates that were more than double those given by low-mortality specialties. The point is that we must be careful not to rely on our own memory or experiences in making judgments about treatment efficacy, side effects, or risks.

CONCLUSION

The five biases described in this chapter are summarized in Table 4.4. In any given clinical situation, we might see more than one of these operating. In fact, in our clinical example in which the patient and his health care providers are trying to decide whether to try Hyalgan for an arthritic knee, all five could potentially be at work. As a result, it is virtually impossible to make definitive statements about the exact probability of success of this intervention.

The important point is to recognize that at least one of these biases can be at work in every clinical diagnosis or treatment decision and to be wary of their perverse effect on judgment and decision making. Simply knowing these biases, thinking of how they can influence your thinking whenever there is uncertainty, considering their implications at least qualitatively, and sharing their consequences with your patients can go a long way to improving your practice of physical therapy.

SELF-ASSESSMENT

1. Consider some advice you have given or seen given to patients over the past few weeks. Can you identify any examples of these biases in your decisions about the advice? If you have not recently been involved in patient care, reflect instead on advice you or a family member may have received from a health care practitioner.

2. Identify three changes you might recommend in a physical therapist's practice to help you and your colleagues reduce the likelihood of these biases.

CONTINUED LEARNING

Select and read one of the books described below. These texts also describe other biases at work in human decision making. Using this information, continue the process of assessing your practice to see how you can change it to reduce the likelihood of allowing these biases to negatively influence your decision making.

• Baron, J. Thinking and Deciding, 4th ed. Cambridge, UK: Cambridge University Press; 2007.

This is a very impressive textbook covering the full range of psychological research into the nature of thinking and decision making. The presentation is remarkably thorough and clear.

• Hoch, SJ, Kunreuther, HK. Wharton on Making Decisions. Hoboken, NJ: John Wiley & Sons; 2004.

This collection of articles written by professors at The University of Pennsylvania's Wharton School addresses both intuitive and analytical approaches to decision making. The major topic headings are Personal Decision Making (Chapters 2,3,4), Managerial Decision Making (Chapters 5, 6, 7, 8), Multiparty Decision Making (Chapters 9, 10, 11, 12), and Impact of Decision Making on Society (Chapters 13, 14, 15, 16, 17).

• Kahneman, D, Tversky, A. Choices, Values, and Frames. Cambridge, UK: Cambridge University Press; 2000.

This is another collection of readings, representing both experimental and empirical studies of human judgment and choice. Many of these readings have become classics.

REFERENCES

INTRODUCTION

Early physicians learned the art of medicine by tutorial, observing their mentors and discussing the decisions they made.¹ In this manner the rich wisdom of practice was passed from one generation to the next. Of course, errors in thinking were also passed on. Our current health care system has formalized this process of passing on knowledge through organized education that includes both formal academic preparation and well-defined clinical education requirements. During this formal education students are often asked to explain their clinical decision making, but experienced teachers and mentors seldom explicate their own. In addition, explicit processes to explain or improve clinical decision making are considerably diminished once the practitioner has met entry-level requirements. Licensed practitioners, be they physicians or physical therapists, usually make their clinical decisions in isolation, often with only the patient as a fellow participant in the process. This means that there are few opportunities to articulate, examine, and validate the clinical decision making process. We continue to support behaviors that often are just as likely to lead to the continuation of poor decision making as to better decision making. Some would even say that our current health care system forces us to work against the process of making clinical decisions explicit for practitioners, thus forcing isolated and isolating practice environments on young practitioners.

Many systems have been developed in response to this problem. They are primarily divided into two approaches. The first is a process approach that provides templates that encourage good decision making—helping us with the way we think. The second is a content approach that provides guidelines about what the decisions should be—suggesting to us what to think. When used together these two approaches can be very successful in helping us reflect on and improve our decisions.

PROCESS APPROACHES TO THE WAY WE THINK

The linguist B. L. Whorf has said, "Language shapes the way we think, and determines what we can think about."² In Chapter 1 we discussed the language of physical therapy practice, focusing on enablement/disablement models and on the patient care management model as sources of the language we use to communicate. We can also view these models as guides to our thinking processes, leading us to organize our clinical decisions in certain ways, to adopt new ways to think.

Enablement/Disablement Models

The use of a disablement model as a basis for our thinking leads us first to recognize that we should be thinking about our patients on multiple levels of pathology and all of its sequellae. If we consider only the pathology or etiology-related diagnosis, we will not consider the need to assess the patient's impairments or functional limitations. The promulgation of the Nagi model and its related forms made possible the development of the Guide to Physical Therapist Practice (which we will return to below) by allowing its authors the ability to think about the physical therapist's care in that organized framework.³

The International classification of Functioning, Disability, and Health (ICF; see Chapter 1) extends the somewhat linear thinking of the Nagi-related models to the more complex interrelationships among function, disability, and health, recognizing the impact of environmental and personal factors.^4,5,6 In particular, the World Health Organization (WHO) states that the ICF was designed to help reformulate the concepts of health and disability by making it clear that all people have some elements of disability, no matter how healthy, and that all people have some elements of health, no matter how disabled. This shift is accomplished by giving us a common metric for measuring both health and disability. This certainly provides a new way of thinking about describing, improving, and measuring health status.⁷

Patient/Client Management Model

We also have the patient/client management model to help us think about patient care.³ In Chapter 1 we discussed the common language the patient/client management model provides us to document what we do with patients. But the model also provides us a framework for the process of our patient care. The very presentation of the model (see Fig. 1-4) shows that this is a model of a sequence of events and thought processes. There is an order to each part of the model that represents the process that is used in interacting with our patients. For example, examination is generally the first step in the process. The logic of examination means we start with a history; we augment the history with the systems review to identify possible problem areas; we then use both the history and the systems review to focus the selection of specific tests and measures. All these data lead us to the diagnosis and prognosis, which in turn lead us to choose the appropriate interventions designed to achieve certain outcomes. The model includes a step labeled evaluation that is clearly not a behavior but a cognitive process where integration of data to derive conclusions takes place. Therefore, the patient/client management model is clearly a process of thinking that we can use in our patient interactions.

The Hypothesis-Oriented Algorithm for Clinicians

Rothstein and Echternach proposed the Hypothesis-Oriented Algorithm for Clinicians (HOAC) in 1986 to explicate the hypothetico-deductive process in physical therapy.⁸ As we have discussed (see Chapter 2), the hypothetico-deductive process consists of gathering cues, generating hypotheses that fit one or more of the cues, and then testing the hypotheses to determine the best fit or the diagnosis. This process is used to turn an unstructured, unfamiliar, or complicated patient problem into one that can be addressed with the use of one or more hypotheses. The HOAC was a branching process that led clinicians from initial data gathering to problem statements to hypotheses that could be tested by further data collection or treatment. This model became widely adopted in teaching the clinical reasoning in physical therapy education.

The HOAC has been recently modified as the HOAC II,⁹ which adds the language of the patient care management model and also recognizes the role of physical therapy in prevention of impairments and disabilities, thereby updating to the current language of our care. As with the original HOAC, the HOAC II offers a series of steps or an algorithm that starts with history and examination, moves through identification of problems, to generation of hypotheses, and to development of a plan of care. It culminates in implementation and then reassessment. Figure 5-1 demonstrates the detail inherent in the HOAC II. By using this algorithm, therapists, especially when dealing with patient problems with which they are unfamiliar, can be more efficient in their clinical care.

The utility of all of these models (the ICF, the patient/client management model, and the HOAC) for guiding our clinical thinking can be seen in the integration of them into a framework for decision making with neurological patients (Fig. 5-2).¹⁰ The authors intend the framework to apply to both the hypothetico-deductive process and the forward reasoning or pattern recognition seen in more experienced clinicians. Shenkman et al state that "by building from simple principles to the entire framework, experienced clinicians and educators can use this framework to assist students and novice clinicians to better understand and incorporate all aspects of decision making necessary for effective care of patients. This framework also provides the experienced clinician with a structure for articulating intuitive decisions."¹⁰

As can be seen by reviewing each of these models, they do not prescribe what to think, but offer us ways to go about the clinical decision making process. The patient/client management model and the HOAC II come from physical therapy. The first was designed by a large group of physical therapists working under the auspices of the American Physical Therapy Association to articulate the process of good clinical care.³ The second is an important contribution from its authors to very explicitly describe the use of the hypothetico-deductive process in physical therapy.^8,9 The ICF also has its roots in the field of rehabilitation, with widespread input from member countries of the World Health Organization over a number of years.⁷ The most recent framework, the integration of the other three into one model,¹⁰ was done as part of the III Step Conference that was conducted by the Neurology Section and the Section on Pediatrics of the American Physical Therapy Association in 2005 to integrate the science of motor control into the selection and application of interventions in physical therapy.¹¹

Although each process is unique in its origins, they have much in common. All are based on review of the literature in decision making, especially by health care practitioners. All speak directly to the decisions that physical therapists need to make each and every day with each and every patient. As we have discussed in Chapters 2 through 4, clinical decision making is a complex topic. Good clinical decision making is also very difficult. These tools assist all clinicians, be they novice or experienced, by providing useful and valid frameworks. As we go about our clinical care, we can incorporate these processes into our thought processes. This can be done informally and personally by choices made by each clinician in his or her own thinking. They can also be included in more formal ways into our practices through documentation systems (see Chapter 18), which also serve as a means of communicating with other practitioners, and in the organization of in-service presentations and other forms of professional development.

As we said earlier, the language we use and the way we think are inexorably linked. By adopting the language of these models we will change our thinking patterns. By these useful paradigms, we will change our language. Of course, the desired outcome of both is better patient care.

CONTENT APPROACHES TO WHAT WE THINK

If our goal is to improve patient care, then finding recommendations for our behavior that assure we are doing the right thing would be very welcome. After all, if we know that one option is clearly superior to another, we would certainly take that option. The literature is full of recommendations to health care practitioners about what decisions they should make about every aspect of care from the smallest of care issues to decisions that can have immediate effects on mortality. Recommendations exist across the range of decisions we need to make, from screening to diagnosis to intervention. But not all of these recommendations are created equal! Some may come from the makers of various medications or devices, some may come from professional groups with particular political perspectives, some may come from consumer groups with a strong advocacy agenda. And some may come after years of rigorous examination that supports the conclusions drawn as having a positive effect on our patients' health.

It is important that clinicians choose from among these many recommendations wisely or the very goal of improved care will be lost. Therefore we should use an explicit framework to guide our selection of recommendations. Such a framework can help us to be sure that when we are accepting a recommendation about our choices in patient care we have made the right decision.

Eddy has developed just such a framework.¹² In it he describes these types of patterns to guide practice. They include:

• standards—when there is clear agreement across the profession that everyone should, even must, make particular clinical decisions;

• guidelines—behaviors that are recommended as the best choice; and

• options—a range of acceptable behaviors, all equally acceptable, from which to choose.¹²

Note that this discussion refers to standards of specific clinical care, not to standards of general professional practice, such as codes of ethics.

Guidelines

There are no known published standards of care in physical therapy. As is true in most health professions there is little if any agreement about clinical choices that can meet the level of scrutiny that defines a standard of care. However, many types of guidelines have been developed that are available and sometimes used by physical therapists. The agencies that have developed these guidelines range from large national organizations to individual practices. The guidelines also vary greatly in their reliance on evidence from the research as opposed to practitioner opinion. There are many examples of guidelines that have been developed by both governmental and professional advisory groups.^13,14,15 Many of these guidelines are written with the patient diagnosis as the focus, so they apply broadly to many practitioners, regardless of discipline. Guidelines have also been developed by many facilities and other clinical entities, often in the form of critical or clinical paths.^16,17 Some of these types of guidelines are intended to apply across a wide spectrum of practices, but some, such as guidelines for postsurgical care, are idiosyncratic to the preferences of specific practitioners.

Options

There is a growing body of literature about certain guidelines that are termed clinical decision rules or clinical prediction rules.^18,19 A clinical decision rule is a tool the clinician can use to increase the accuracy of diagnoses and prognoses and therefore improve the selection of specific interventions. These rules are developed and published in the literature as options for clinical practice. The rules are developed by testing the results from previously published evidence with specific patient populations. We discuss the development of clinical decision rules, as well as the development of other types of guidelines, in Chapter 15.

In addition, we have a major source for acceptable options in physical therapy, the Guide to Physical Therapist Practice.³ The Guide contains over 30 descriptions of preferred practice patterns for patients with dysfunction or the potential for dysfunction in the musculoskeletal, the neuromuscular, the cardiovascular and pulmonary, and the integumentary systems (Table 5.1). These patterns represent the profession's first attempt to identify broad diagnostic categories that group together patients with similar impairments, in order to describe options for care for these groups of patients. Each practice pattern is comprised of specific content in seven general areas, which are derived from the patient/client management model (Table 5.2).

1. The Patient/Client Diagnostic classification area provides inclusion and exclusion criteria for that particular pattern and some related diagnoses of pathology.

2. The Examination area lists the range of items a clinician should consider in conducting the history, systems review, and specific tests and measures; this is not meant to be a list of what should be done, but rather a fairly complete list of what could be done to provide the information to arrive at diagnoses and prognoses for a given patient who fits this pattern.

3. The Evaluation/Diagnosis/Prognosis area provides expected outcomes for patients in this pattern, length of care, and reasons why a particular patient may depart from these expected norms.

4. The Intervention area lists a range of interventions that are appropriate for a patient who fits this pattern; again, it is not intended that any patient receive all of these interventions, but rather they represent a fairly complete list of what could be done to help achieve the patient's intended outcomes.

5. The Reexamination area provides the reasons for and timing of reexamination.

6. The Global Outcomes area lists outcomes that are applicable to all patients.

7. The Criteria for Termination of PT Services area describes the reasons why patients should be discharged or care should be discontinued.

The elements of these practice patterns can help clinicians evaluate the comprehensiveness of their clinical thinking.

Clinical Guidance

Eddy points out the necessity to move cautiously from options to guidelines to standards. Since standards can be used to hold practitioners accountable for various specific behaviors by forming the basis for malpractice claims or loss of licensure, they require very strong evidence and widespread professional and patient acceptance. However, another important reason to move slowly is to preserve what is termed "warranted variation." When we are not certain of the best course of events, some variety across practitioners and across regions gives us a natural laboratory to explore the outcomes of these various options for our patients. Moving too rapidly to reach a conclusion can result in adoption of a test or intervention that has a less than optimal outcome.¹²

Another benefit of providing options is that they allow the possibility of mistakes. Although we do not want to make mistakes that have lasting harm to patients, mistakes are a powerful way of learning. It has been shown that recognizing a mistake, analyzing it, and adapting behavior accordingly are all hallmarks of the clinical expert.²⁰

CONCLUSION

By understanding the typical processes of decision making, recognizing the result of our decisions, and using some tools to improve our decision making process, we are in a better position to understand and use the disablement model, the patient care management model, and evidence based practice to improve patient care.

These descriptions of best clinical practice go by a variety of names, serve a variety of purposes, and have a variety of formats. Some are descriptions, at varying levels of urgency, of what to do with certain patients, and others are descriptions of processes of thinking, without obvious plans of care. When selected wisely, these tools can help clinicians make better clinical decisions. In later chapters we will explore more deeply what a clinician needs to know to make wise choices.

SELF-ASSESSMENT

1. Identify a patient whom you have seen lately. Determine which practice pattern the patient represents, and then compare the care you provided with the care described in the pattern. Why was your care different from that in the pattern? After reviewing the pattern, would you change your care?

2. Identify a patient whom you have seen lately. Outline the decision making process for this patient using the format of the HOAC. How was your process different from the HOAC process? Would you change your process with similar patients in the future? Why or why not?

CONTINUED LEARNING

Identify some practice recommendations that influence your care. Determine if they are standards, guidelines, or options. Identify the background information that was used to develop them.

1. Can you find expert opinion and evidence from the literature, or are they based on one person's preferences?

2. As you move through this text, come back to these recommendations to evaluate their quality.

REFERENCES