reasoning comes from empirical studies of nonmedical problem- 21

4 Decision-Making in
Clinical Medicine
solving behavior. Because of the diverse perspectives contributing to
this area, with important contributions from cognitive psychology,
medical education, behavioral economics, sociology, informatics, and
decision sciences, no single integrated model of clinical reasoning

CHAPTER 4
Daniel B. Mark, John B. Wong exists, and not infrequently, different terms and reasoning models
describe similar phenomena.
Intuitive Versus Analytic Reasoning A useful contemporary
Practicing medicine at its core requires making decisions. What makes model of reasoning, the dual-process theory distinguishes two general
medical practice so difficult is not only the specialized technical conceptual modes of thinking as fast or slow. Intuition (System 1)

Decision-Making in Clinical Medicine

knowledge required but also the intrinsic uncertainty that surrounds provides rapid effortless judgments from memorized associations
each decision. Mastering the technical aspects of medicine alone, using pattern recognition and other simplifying “rules of thumb” (i.e.,
unfortunately, does not ensure a mastery of the practice of medicine. heuristics). For example, a very simple pattern that could be useful
Sir William Osler’s familiar quote “Medicine is a science of uncertainty in certain situations is “black woman plus hilar adenopathy equals
and an art of probability” captures well this complex duality. Although sarcoid.” Because no effort is involved in recalling the pattern, the
the science of medicine is often taught as if the mechanisms of the clinician is often unable to say how those judgments were formulated.
human body operate with Newtonian predictability, every aspect of In contrast, Analysis (System 2), the other form of reasoning in the
medical practice is infused with an element of irreducible uncertainty dual-process model, is slow, methodical, deliberative, and effortful. A
that the clinician ignores at her peril. Although deeply rooted in student might read about causes of hilar adenopathy and from that list
science, more than 100 years after the practice of medicine took its (e.g., Chap. 66), identify diseases more common in black women or
modern form, it remains at its core a craft, to which individual doctors examine the patient for skin or eye findings that occur with sarcoid.
bring varying levels of skill and understanding. With the exponential These dual processes, of course, represent two exemplars taken from
growth in medical literature and other technical information and an the cognitive continuum. They provide helpful descriptive insights but
ever-increasing number of testing and treatment options, twenty-first very little guidance in how to develop expertise in clinical reasoning.
century physicians who seek excellence in their craft must master a How these idealized systems interact in different decision problems,
more diverse and complex set of skills than any of the generations that how experts use them differently from novices, and when their use can
preceded them. This chapter provides an introduction to three of the lead to errors in judgment remain the subject of study and considerable
pillars upon which the craft of modern medicine rests: (1) expertise in debate.
clinical reasoning (what it is and how it can be developed); (2) rational Pattern recognition, an important part of System 1 reasoning, is
diagnostic test use and interpretation; and (3) integration of the best a complex cognitive process that appears largely effortless. One can
available research evidence with clinical judgment in the care of indi- recognize people’s faces, the breed of a dog, an automobile model, or
vidual patients (evidence-based medicine [EBM]). a piece of music from just a few notes within milliseconds without
necessarily being able to articulate the specific features that prompted
■ BRIEF INTRODUCTION TO CLINICAL REASONING the recognition. Analogously, experienced clinicians often recognize
Clinical Expertise Defining “clinical expertise” remains surpris- familiar diagnostic patterns very quickly. The key here is having a large
ingly difficult. Chess has an objective ranking system based on skill library of stored patterns that can be rapidly accessed. In the absence
and performance criteria. Athletics, similarly, have ranking systems of an extensive stored repertoire of diagnostic patterns, students (as
to distinguish novices from Olympians. But in medicine, after phy- well as experienced clinicians operating outside their area of expertise
sicians complete training and pass the boards (or get recertified), no and familiarity) often must use the more laborious System 2 analytic
tests or benchmarks are used to identify those who have attained the approach along with more intensive and comprehensive data collection
highest levels of clinical performance. At each institution, there are to reach the diagnosis.
often a few “elite” clinicians who are known for their “special problem- The following brief patient scenarios illustrate three distinct pat-
solving prowess” when particularly difficult or obscure cases have baf- terns associated with hemoptysis that experienced clinicians recognize
fled everyone else. Yet despite their skill, even such master clinicians without effort:
typically cannot explain their exact processes and methods, thereby • A 46-year-old man presents to his internist with a chief complaint
limiting the acquisition and dissemination of the expertise used of hemoptysis. An otherwise healthy, nonsmoker, he is recovering
to achieve their impressive results. Furthermore, clinical virtuosity from an apparent viral bronchitis. This presentation pattern suggests
appears not to be generalizable, e.g., an expert on hypertrophic cardio- that the small amount of blood-streaked sputum is due to acute
myopathy may be no better (and possibly worse) than a first-year med- bronchitis, so that a chest x-ray provides sufficient reassurance that
ical resident at diagnosing and managing a patient with neutropenia, a more serious disorder is absent.
fever, and hypotension. • In the second scenario, a 46-year-old patient who has the same chief
Broadly construed, clinical expertise encompasses not only cogni- complaint but with a 100-pack-year smoking history, a productive
tive dimensions involving the integration of disease knowledge with morning cough with blood-streaked sputum, and weight loss fits
verbal and visual cues and test interpretation but also potentially the the pattern of carcinoma of the lung. Consequently, along with the
complex fine-motor skills necessary for invasive procedures and tests. chest x-ray, the clinician obtains a sputum cytology examination and
In addition, “the complete package” of expertise in medicine requires refers this patient for a chest CT scan.
effective communication and care coordination with patients and • In the third scenario, the clinician hears a soft diastolic rumbling
members of the medical team. Research on medical expertise remains murmur at the apex on cardiac auscultation in a 46-year-old patient
sparse overall and mostly centered on diagnostic reasoning, so in with hemoptysis who immigrated from a developing country and
this chapter, we focus primarily on the cognitive elements of clinical orders an echocardiogram as well, because of possible pulmonary
reasoning. hypertension from suspected rheumatic mitral stenosis.
Because clinical reasoning occurs in the heads of clinicians, objec-
tive study of the process is difficult. One research method used for Pattern recognition by itself is not, however, sufficient for secure
this area asks clinicians to “think out loud” as they receive increments diagnosis. Without deliberative systematic reflection, undisciplined
of clinical information in a manner meant to simulate a clinical pattern recognition can result in premature closure: mistakenly jump-
encounter. Another research approach focuses on how doctors should ing to the conclusion that one has the correct diagnosis before all the
reason diagnostically, to identify remediable “errors,” rather than on relevant data are in. A critical second step, therefore, even when the
how they actually do reason. Much of what is known about clinical diagnosis seems obvious, is diagnostic verification: considering whether

HPIM21e_Part1_p001-p090.indd 21 19/01/22 1:49 PM

 22 the diagnosis adequately accounts for the presenting symptoms and a case from a morbidity and mortality conference in which an elderly
signs and can explain all the ancillary findings. The following case patient presented with painless dyspnea of acute onset and was eval-
based on a real clinical encounter provides an example of premature uated for a pulmonary cause but was eventually found to have acute
closure. A 45-year-old man presents with a 3-week history of a “flulike” MI, with the diagnostic delay likely contributing to the development of
upper respiratory infection (URI) including dyspnea and a productive ischemic cardiomyopathy. If the case was associated with a malpractice
cough. The emergency department (ED) clinician pulled out a “URI accusation, such examples may be even more memorable. Errors with
PART 1

assessment form,” which defines and standardizes the information the availability heuristic arise from several sources of recall bias. Rare
gathered. After quickly acquiring the requisite structured examination catastrophic outcomes become memorable cases with a clarity and
components and noting in particular the absence of fever and a clear force disproportionate to their likelihood for future diagnosis—for
chest examination, the physician prescribed a cough suppressant for example, a patient with a sore throat eventually found to have leuke-
The Profession of Medicine

acute bronchitis and reassured the patient that his illness was not seri- mia or a young athlete with leg pain subsequently found to have an
ous. Following a sleepless night at home with significant dyspnea, the osteosarcoma—and those publicized in the media or recently experi-
patient developed nausea and vomiting and collapsed. He was brought enced are, of course, easier to recall and therefore more influential on
back to the ED in cardiac arrest and was unable to be resuscitated. clinical judgments.
His autopsy showed a posterior wall myocardial infarction (MI) and The third commonly used cognitive shortcut, the anchoring heu-
a fresh thrombus in an atherosclerotic right coronary artery. What ristic (also called conservatism or stickiness), involves insufficiently
went wrong? Presumably, the ED clinician felt that the patient was adjusting the initial probability of disease up (or down) following a
basically healthy (one can be misled by the way the patient appears on positive (or negative test) when compared with Bayes’ theorem, i.e.,
examination—a patient that does not appear “sick” may be incorrectly sticking to the initial diagnosis. For example, a clinician may still judge
assumed to have an innocuous illness). So, in this case, the physician, the probability of coronary artery disease (CAD) to be high despite a
upon hearing the overview of the patient from the triage nurse, elected negative exercise perfusion test and go on to cardiac catheterization
to use the URI assessment protocol even before starting the history, (see “Measures of Disease Probability and Bayes’ Rule,” below).
closing consideration of the broader range of possibilities and associ- The fourth heuristic states that clinicians should use the simplest
ated tests required to confirm or refute these possibilities. In particular, explanation possible that will adequately account for the patient’s
by concentrating on the abbreviated and focused URI protocol, the symptoms and findings (Occam’s razor or, alternatively, the simplicity
clinician failed to elicit the full dyspnea history, which was precipitated heuristic). Although this is an attractive and often used principle, it
by exertion and accompanied by chest heaviness and relieved by rest, is important to remember that no biologic basis for it exists. Errors
suggesting a far more serious disorder. from the simplicity heuristic include premature closure leading to the
Heuristics or rules of thumb are a part of the intuitive system. neglect of unexplained significant symptoms or findings.
These cognitive shortcuts provide a quick and easy path to reaching For complex or unfamiliar diagnostic problems, clinicians typically
conclusions and making choices, but when used improperly, they can resort to analytic reasoning processes (System 2) and proceed method-
lead to errors. Two major research programs have studied heuristics ically using the hypothetico-deductive model of reasoning. Based on
in a mostly nonmedical context and have reached very different con- the patient’s stated reasons for seeking medical attention, clinicians
clusions about the value of these cognitive tools. The “heuristics and develop an initial list of diagnostic possibilities in hypothesis generation.
biases” program focuses on how these mental shortcuts can lead to During the history of the present illness, the initial hypotheses evolve
incorrect judgments. So far, however, little evidence exists that educat- in diagnostic refinement as emerging information is tested against
ing physicians and other decision makers to watch for the >100 cogni- the mental models of the diseases being considered with diagnoses
tive biases identified to date has had any effect on the rate of diagnostic increasing and decreasing in likelihood or even being dropped from
errors. In contrast, the “fast and frugal heuristics” research program consideration as the working hypotheses of the moment. These mental
explores how and when relying on simple heuristics can produce good models often generate additional questions that distinguish the diag-
decisions. Although many heuristics have relevance to clinical reason- nostic possibilities from one another. The focused physical examina-
ing, only four will be mentioned here. tion contributes to further distinguishing the working hypotheses. Is
When diagnosing patients, clinicians usually develop diagnostic the spleen enlarged? How big is the liver? Is it tender? Are there any
hypotheses based on the similarity of that patient’s symptoms, signs, palpable masses or nodules? Diagnostic verification involves testing the
and other data to their mental representations (memorized patterns) adequacy (whether the diagnosis accounts for all symptoms and signs)
of the disease possibilities. In other words, clinicians pattern match and coherency (whether the signs and symptoms are consistent with
to identify the diagnoses that share the most similar findings to the the underlying pathophysiologic causal mechanism) of the working
patient at hand. This cognitive shortcut is called the representative- diagnosis. For example, if the enlarged and quite tender liver felt on
ness heuristic. Consider a patient with hypertension who has head- physical examination is due to acute hepatitis (the hypothesis), then
ache, palpitations, and diaphoresis. Based on the representativeness certain specific liver function tests will be markedly elevated (the pre-
heuristic, clinicians might judge pheochromocytoma to be quite diction). Should the tests come back normal, the hypothesis may have
likely given this classic presenting symptom triad suggesting pheo- to be discarded and others reconsidered.
chromocytoma. Doing so, however, would be incorrect given that Although often neglected, negative findings are as important as
other causes of hypertension are much more common than pheo- positive ones because they reduce the likelihood of the diagnostic
chromocytoma and this triad of symptoms can occur in patients who hypotheses under consideration. Chest discomfort that is not provoked
do not have it. Thus, clinicians using the representativeness heuristic or worsened by exertion and not relieved by rest in an active patient
may overestimate the likelihood of a particular disease based on the lowers the likelihood that chronic ischemic heart disease is the under-
presence of representative symptoms and signs, failing to account for lying cause. The absence of a resting tachycardia and thyroid gland
its low underlying prevalence (i.e., the prior, or pretest, probabilities). enlargement reduces the likelihood of hyperthyroidism in a patient
Conversely, atypical presentations of common diseases may lead to with paroxysmal atrial fibrillation.
underestimating the likelihood of a particular disease. Thus, inexpe- The acuity of a patient’s illness may override considerations of prev-
rience with a specific disease and with the breadth of its presentations alence and the other issues described above. “Diagnostic imperatives”
may also lead to diagnostic delays or errors, e.g., diseases that affect recognize the significance of relatively rare but potentially catastrophic
multiple organ systems, such as sarcoid or tuberculosis, may be partic- conditions if undiagnosed and untreated. For example, clinicians should
ularly challenging to diagnose because of the many different patterns consider aortic dissection routinely as a possible cause of acute severe
they may manifest. chest discomfort. Although the typical presenting symptoms of dissec-
A second commonly used cognitive shortcut, the availability heu- tion differ from those of MI, dissection may mimic MI, and because
ristic, involves judgments based on how easily prior similar cases or it is far less prevalent and potentially fatal if mistreated, diagnosing
outcomes can be brought to mind. For example, a clinician may recall dissection remains a challenging diagnostic imperative (Chap. 280).

HPIM21e_Part1_p001-p090.indd 22 19/01/22 1:49 PM

 Clinicians taking care of acute, severe chest pain patients should expertise. Their use in developing medical expertise and maintaining 23
explicitly and routinely inquire about symptoms suggestive of dis- or enhancing it has not yet been adequately explored. Some studies
section, measure blood pressures in both arms for discrepancies, and in medicine suggest that the most beneficial approach to education
examine for pulse deficits. When these are all negative, clinicians exposes students to both the signs and symptoms of specific diseases
may feel sufficiently reassured to discard the aortic dissection hypothe- (disease pattern recognition) and, in addition, the lists of diseases that

CHAPTER 4
sis. If, however, the chest x-ray shows a possible widened mediastinum, can present with specific symptoms and signs (differential diagnosis).
the hypothesis should be reinstated and an appropriate imaging test Active learning opportunities useful for those in training include
ordered (e.g., thoracic computed tomography [CT] scan or transesoph- developing a personal learning system, e.g., systematically reflecting
ageal echocardiogram). In nonacute situations, the prevalence of on diagnostic processes used (metacognition) and following-up to
potential alternative diagnoses should play a much more prominent identify diagnoses and treatments for patients in their care.

Decision-Making in Clinical Medicine

role in diagnostic hypothesis generation.
Cognitive scientists studying the thought processes of expert clini- ■ DIAGNOSTIC VERSUS THERAPEUTIC
cians have observed that clinicians group data into packets, or “chunks,” DECISION-MAKING
that are stored in short-term or “working memory” and manipulated to The modern ideal of medical therapeutic decision-making is to “per-
generate diagnostic hypotheses. Because short-term memory is limited sonalize” treatment recommendations. In the abstract, personalizing
(classically humans can accurately repeat a list of 7 ± 2 numbers read treatment involves combining the best available evidence about what
to them), the number of diagnoses that can be actively considered in works with an individual patient’s unique features (e.g., risk factors,
hypothesis-generating activities is similarly limited. For this reason, the genomics, and comorbidities) and his or her preferences and health
cognitive shortcuts discussed above play a key role in the generation goals to craft an optimal treatment recommendation with the patient.
of diagnostic hypotheses, many of which are discarded as rapidly as Operationally, two different and complementary levels of personaliza-
they are formed, thereby demonstrating that the distinction between tion are possible: individualizing the risk of harm and benefit for the
analytic and intuitive reasoning is an arbitrary and simplistic, but options being considered based on the specific patient characteristics
nonetheless useful, representation of cognition. (precision medicine), and personalizing the therapeutic decision
Research into the hypothetico-deductive model of reasoning has had process by incorporating the patient’s preferences and values for the
difficulty identifying the elements of the reasoning process that distin- possible health outcomes. This latter process is sometimes referred
guish experts from novices. This has led to a shift from examining the to as shared decision-making and typically involves clinicians sharing
problem-solving process of experts to analyzing the organization of their their knowledge about the options and the associated consequences
knowledge for pattern matching as exemplars, prototypes, and illness and trade-offs and patients sharing their health goals (e.g., avoiding a
scripts. For example, diagnosis may be based on the resemblance of a new short-term risk of dying from coronary artery bypass grafting to see
case to patients seen previously (exemplars). As abstract mental models their grandchild get married in a few months).
of disease, prototypes incorporate the likelihood of various disease fea- Individualizing the evidence about therapy does not mean relying
tures. Illness scripts include risk factors, pathophysiology, and symptoms on physician impressions of benefit and harm from their personal
and signs. Experts have a much larger store of exemplar and prototype experience. Because of small sample sizes and rare events, the chance
cases, an example of which is the visual long-term memory of experi- of drawing erroneous causal inferences from one’s own clinical experi-
enced radiologists. However, clinicians do not simply rely on literal recall ence is very high. For most chronic diseases, therapeutic effectiveness
of specific cases but have constructed elaborate conceptual networks of is only demonstrable statistically in large patient populations. It would
memorized information or models of disease to aid in arriving at their be incorrect to infer with any certainty, for example, that treating
conclusions (illness scripts). That is, expertise involves an enhanced a hypertensive patient with angiotensin-converting enzyme (ACE)
ability to connect symptoms, signs, and risk factors to one another in inhibitors necessarily prevented a stroke from occurring during treat-
meaningful ways; relate those findings to possible diagnoses; and identify ment, or that an untreated patient would definitely have avoided their
the additional information necessary to confirm the diagnosis. stroke had they been treated. For many chronic diseases, a majority
No single theory accounts for all the key features of expertise in of patients will remain event free regardless of treatment choices;
medical diagnosis. Experts have more knowledge about presenting some will have events regardless of which treatment is selected; and
symptoms of diseases and a larger repertoire of cognitive tools to those who avoided having an event through treatment cannot be
employ in problem solving than nonexperts. One definition of exper- individually identified. Blood pressure lowering, a readily observable
tise highlights the ability to make powerful distinctions. In this sense, surrogate endpoint, does not have a tightly coupled relationship with
expertise involves a working knowledge of the diagnostic possibilities strokes prevented. Consequently, in most situations, demonstrating
and those features that distinguish one disease from another. Memo- therapeutic effectiveness cannot rely simply on observing the outcome
rization alone is insufficient, e.g., photographic memory of a medical of an individual patient but should instead be based on large groups of
textbook would not make one an expert. But having access to detailed patients carefully studied and properly analyzed.
case-specific relevant information is critically important. In the past, Therapeutic decision-making, therefore, should be based on the best
clinicians primarily acquired clinical knowledge through their patient available evidence from clinical trials and well-done outcome studies.
experiences, but now clinicians have access to a plethora of information Trustworthy clinical practice guidelines that synthesize such evidence
sources. Clinicians of the future will be able to leverage the experiences offer normative guidance for many testing and treatment decisions.
of large numbers of other clinicians using electronic tools, but, as However, all guidelines recognize that “one size fits all” recommen-
with the memorized textbook, the data alone will be insufficient for dations may not apply to individual patients. Increased research into
becoming an expert. Nonetheless, availability of these data removes the heterogeneity of treatment effects seeks to understand how best to
one barrier for acquiring experience with connecting symptoms, signs, adjust group-level clinical evidence of treatment harms and benefits to
and risk factors to the possible diagnoses and identifying the additional account for the absolute level of risks faced by subgroups and even by
distinguishing information necessary to confirm the diagnosis, thereby individual patients, using, for example, validated clinical risk scores.
potentially facilitating the development of the working knowledge nec-
essary for becoming an expert. ■ NONCLINICAL INFLUENCES ON CLINICAL
Despite all of the research seeking to understand expertise in med- DECISION-MAKING
icine and other disciplines, it remains uncertain whether any didactic More than three decades of research on variations in clinician practice
program can actually accelerate the progression from novice to expert patterns has identified important nonclinical forces that shape clin-
or from experienced clinician to master clinician. Deliberate effortful ical decisions. These factors can be grouped conceptually into three
practice (over an extended period of time, sometimes said to be 10 years overlapping categories: (1) factors related to an individual physician’s
or 10,000 practice hours) and personal coaching are two strategies practice, (2) factors related to practice setting, and (3) factors related
often used outside medicine (e.g., music, athletics, chess) to cultivate to payment systems.

HPIM21e_Part1_p001-p090.indd 23 19/01/22 1:49 PM