Clinical decision-making

System 1 thinking is intuitive, fast, and almost unconscious.

System 1 processing makes a direct association between new information and a similar example stored in a practitioner’s memory. It is based on pattern recognition, and the accumulation of many automatic responses. For example, when an experienced physician sees a “typical” skin rash and identifies symptoms and signs and recognizes a clinical syndrome, they instinctively arrive at a diagnosis by “unconscious pattern matching" to clinical templates acquired through experience. Arriving at a diagnosis through pattern recognition is quick, often very effective and usually correct. However, it is prone to interference by cognitive and affective biases as well as attentional drift (i.e. decreasing attention paid during routine practices while using system 1 thinking) that may occasionally mislead even the most experienced physicians.

System 2 thinking is conscious, analytical, and effortful.

System 2 is slower and more cognitively demanding than system 1 thinking. When no match for a given clinical presentation is found in a practitioner’s memory or when the situation is ambiguous, the physician employs analytical reasoning, which requires a more deliberate, methodical approach.

Medical students and residents make greater use of system 2 thinking, as they acquire the knowledge required to guide practice. As health professionals gain expertise, knowledge and experience, system 2 thinking increasingly yields to system 1 thinking. This allows busy providers to be more efficient and effectively manage heavy workloads.

Both approaches to diagnostic reasoning are essential to the successful practice of medicine and both bear their pitfalls. While system 1 thinking is prone to cognitive and affective biases, system 2 thinking requires a sound knowledge base as an anchor. Experienced or expert providers learn to leverage both modes of thinking and to call upon the skills inherent to each, based on the situation. Strategies to promote sound clinical reasoning should focus on both systems.

Collapse section

Clinical decision-making is a function of complex interactions between system 1 and system 2 thinking in the context of personal performance-modifying factors (anxiety, fatigue, stress, hunger, etc.) and system performance-modifying factors (bullying, hierarchy, resource limitations, etc.). Overreliance on intuitive thinking and automaticity as well as the failure to employ cognitive forcing strategies to leverage analytical thinking and minimize inattention can lead to suboptimal clinical decisions.

Experienced physicians have knowledge of disease patterns that were learned over time. They recognize the subtle differences between their knowledge of the disease—“illness scripts”—and the specific presentation of the patient, thus distinguishing between typical and atypical presentations of the disease. The early stage of diagnosis extracts this knowledge from memory.

A methodical approach to collecting and evaluating information and a thorough knowledge base of disease presentation is crucial to successfully arrive at a hypothesis. This can occur only by obtaining sufficient clinical information from the patient encounter (via the history, physical examination, and investigations) to inform and verify the initial “gut” diagnostic hypothesis.

Among the most prevalent causes of diagnostic delay is a physician’s failure to capture sufficiently thorough information while taking the history, performing the physical exam, ordering investigations, and documenting the care. When asked to collect information, physicians tend to favour facts that confirm their beliefs. Once the initial hypothesis or favoured diagnosis is made, physicians tend to suppress incongruences and ignore missing data. Being aware of this tendency and taking the time to consciously consider its impact (a process called cognitive forcing) may help you avoid cognitive pitfalls that can lead to diagnostic delay.

Diagnostic delays arising from cognition arise in experts and novices alike. While experts are more likely to overlook incongruences or missing information, they are less likely to make a mistake in generating a hypothesis based on their knowledge of patterns of disease. Novices, requiring time to learn the patterns of illness, are more likely to generate a faulty hypothesis but are less likely to overlook the data.³ In addition, novices may pay more attention to faulty or irrelevant data and hence generate a faulty hypothesis.⁴ Consequently, a multi-pronged strategy is key to minimizing diagnostic delay.

Strategies to improve diagnostic accuracy are based on:

1. Acquiring and retaining knowledge
2. Retrieving that knowledge from memory
3. Performing sufficiently thorough clinical encounters
4. Employing cognitive forcing strategies to leverage system 2 thinking to complement system 1 thinking

Decision support tools, directed reflective practice, and diagnostic checklists are helpful in acquiring knowledge of typical and atypical presentations of disease as well as in helping less experienced physicians recognize inconsistent or atypical features.⁴ For experts, these aids are most effective when applied in the more methodical “verification” or “second stage” of diagnosis, after arriving at an initial hypothesis.

Collapse section

A number of strategies help to improve diagnostic accuracy. One such approach is cognitive debiasing: the conscious consideration of cognitive biases and of their potential impact on a decision by considering questions such as “What am I missing?” or “Could I be wrong?”

Diagnostic accuracy may be enhanced through:

• increasing experience to sharpen intuitive thinking (recognizing patterns)
• complementing existing knowledge and skills by obtaining additional information, especially for unusual presentations or challenging cases
• • consulting with experts
  • obtaining a second opinion
  • seeking additional information from the patient
• reflective practice to trigger and elevate cognitive awareness

Reflection may be an effective strategy to promote cognitive refocusing and mitigate errors in thinking. “Slowing down when you should” is a mark of expert judgment. "Slowing down" does not refer to thinking slowly or employing system 2 thinking but rather to the conscious effort to consider the possibility of diagnostic error, to focus on maintaining situational awareness, and to consider other possibilities before making a final decision.

Reflecting “in action” (thinking about an activity or reflecting on it while carrying it out) and reflecting “on action” (reflecting on an activity after carrying it out) may mitigate for deficiencies in thinking and promote safe care. Reflecting in action occurs during the course of a patient’s care. The pause or slowdown allows providers to reflect on the evolution of a patient’s clinical situation. “Slowing down when you should” allows providers to shift their thinking from a primarily intuitive and automatic approach toward a more analytical one, in an attempt to mitigate cognitive and affective biases.⁵ By actively maintaining their attention and switching from automatic decision-making to analytical decision-making, expert diagnosticians may combat distraction, fatigue, or bias, and remain flexible to nuances in the clinical presentation and evolution of the patient’s course. Listening up is a key skill in slowing down.

Situational awareness is critical for effective clinical decision-making. It can be achieved through collaborative decision-making during team rounds, huddles, time outs, and briefings/debriefings. These team activities allow insights from others who can provide alternative suggestions and solutions.

Teams must similarly engage in cognitive forcing strategies to mitigate the risk of group-think bias. Group-think bias can be mitigated by:

• challenging consensus opinions
• alerting teams to red flags
• suggesting potential alternative courses or decisions
• contingency planning

A culture of psychological safety and speaking up is critical to reaching better decisions by the team. The process begins with appropriate preparation and includes:

• briefings
• reviewing the medical record
• handovers
• rounds
• planned time-outs (i.e. surgical safety checklist)

These structured team processes force the team to “slow down when it should” at the preparation phase of clinical decision-making. They also allow healthcare providers to confirm information about the specific patient, the environment, the tasks, and the time or urgency.

Collapse section

While cognitive debiasing and “slowing down when you should” hold promise as ways to minimize the risks associated with diagnostic uncertainty, they are not panaceas: both require the health provider to make the effort to consciously keep automaticity in check. As well, neither can compensate for deficits in knowledge or critical reasoning skills.⁶ Consequently, physicians must make continuing medical education and professional development a centerpiece of their professional practice.

While some diagnoses can be made quickly, others can be elusive. At times, symptoms or signs can be non-specific and require time to evolve to reveal their source. In such cases, it can be helpful to remember that diagnosis is a process, where several cycles of communicating with the patient, testing, reviewing results, consulting with colleagues, observing, and further patient interactions can help narrow the diagnosis over time. Physicians sometimes wrongly believe that ordering a panoply of unnecessary or non-indicated tests will help compensate for diagnostic uncertainty. Excessive investigation, however, is generally unnecessary, and falsely reassures physicians that they have “covered all bases.” Further, over-testing may have unintended consequences and result in harm through unnecessary investigations, each with their own risks.

Rather than relying on a shotgun approach to ordering tests, safe care can more likely be achieved through competent, thorough assessments and good communication and documentation. Not only does documentation leave an “intellectual footprint” of a health provider’s reasoning and demonstrate their diligence, it can help a clinician identify potential gaps in their assessment. In addition, when the diagnosis is difficult to establish or is uncertain, communicating that fact to the patient and including a plan for next steps such as consultations, reassessment visits or testing may be helpful in strengthening the therapeutic relationship and managing expectations. The process of communicating and documenting clinical reasoning and the rationale for decisions can also help trigger “slowing down when you should,” and as such, is an important component of safe care.

Collapse section

Many factors can impact a physician’s ability to perform an adequate assessment. Challenges can be related to:

Patient factors:

• Underlying medical conditions
• Complexity of presentation
• Complexity of family history
• Language barriers
• Cultural/ethnic/gender issues
• Health literacy
• Difficult patient behaviours
• Misaligned expectations

Environment factors:

Physical environment:

• Noise
• Lack of privacy
• Interruptions
• Distractions
• Overcrowding
• Virtual visits

Human environment:

• Knowledge, skills, and experience
• Communication skills
• Time and urgency
• Lack of resources including human resources (translators)
• Multitasking
• Stress
• Fatigue
• Illness
• Cognitive and affective biases
• Insight into one’s limitations

In addition to the above factors, the gradual and usually unconscious drifting away from methodical approaches to evaluation may contribute to less thorough assessments (i.e. omitting to inquire about family history) and can contribute to misdiagnosis and diagnostic delay. It is important for faculty to teach learners how to be thorough and reliable in assessing patients, as expected for their own specialty, and to identify behavioural drift in order to enable a return to sound practices. To this end, individual practitioners must be aware of, and adhere to, the standard of care—the level of care that would reasonably be expected of a normal prudent physician with similar training and experience in a similar circumstance.

The CMPA’s experience with matters involving diagnostic delay indicates that paying attention to certain aspects of practice may help improve diagnostic accuracy.

To reduce diagnostic delay:

• Avoid multitasking and focus exclusively on the patient in the room.
• Be methodical and have a reliable approach to the history and physical examination, for example:
• • determine the past history of similar problems
  • ask about the family history
  • explore what has already been done to manage the symptoms
  • be mindful of the possibility of misdiagnosis when a patient makes recurrent visits for the same clinical problem
• Take an appropriately thorough history of the patient’s health condition.
• • Telephone or hallway advice to colleagues are two situations where lack of details about the patient’s condition are common and can lead to decisions based on insufficient information gathering.
• With consent, seek help from those who may be able to assist in obtaining the history (e.g. substitute decision-maker, parent, friend, or interpreter), if the patient is unable to provide a history due to a language barrier or capacity issue.
• Adequately assess relevant risk factors, including family history.
• Inquire about "red flag" symptoms.
• Read the notes taken by other healthcare professionals (e.g. nurses, paramedics).
• • These frequently contain ancillary details to support or refute a hypothesis.
• Make an effort to explore rather than to dismiss patient concerns. The following questions may help providers notice the importance of symptoms or signs that may have been unconsciously overlooked:
• • What do you feel is your biggest challenge dealing with this problem?
  • How are you coping with this?
  • What are your goals for your care?
• Review pertinent medical records, test results, and consultation reports.

Providing advice to colleagues in “hallway consultations”

It is not uncommon for practitioners to seek support and guidance from colleagues. Whether it be due to lack of resources, a desire not to disturb a consultant, or a belief that the issue at hand does not appear to warrant a formal consultation, the practice of seeking and providing advice informally in casual encounters, is ever-present.

The physician who provides advice in such informal contexts may be engaging their duty of care to the patient – even when they have not formally assessed the patient—and can contribute to diagnostic delay and patient harm. Telephone or hallway conversations may not allow practitioners to engage fully, mistakenly believing they do not need to, and thus lead to insufficient assessments.

To reduce the risk associated with engaging in informal consultations:

As a physician asking for advice:

• Be clear with your colleague whether you seek generic management advice, or have questions about a specific patient.
• Ensure that you share all relevant information.
• Offer to formally consult your colleague before that colleague forms an opinion, if necessary and possible.
• Document the advice received and the name of the colleague involved.

As a physician being asked for an opinion:

• Engage fully with your colleagues.
• Clarify early whether you are discussing an issue generally or with a specific patient in mind
• • Recognize the possibility that you may engage a duty of care to the patient
• Resist the urge to simply validate an initial impression without truly engaging.
• Remain methodical in your approach to assessment.
• Question sufficiently to gather the necessary information to make a recommendation and communicate clearly with your colleague on the limitations, if any, of the advice you are providing.
• Be alert to the complexity of care and offer to see the patient in a formal consultation if you feel the need.
• Document the encounter. (This may serve to trigger reconsideration, as described below.)

Collapse section

Peer experts are often critical of cursory physical examinations and of the documentation in cases of diagnostic delay. Time constraints, lack of a private physical space for the examination, patient reluctance to be examined, or cognitive biases are often the reasons behind incomplete physical exams.

To minimize misdiagnosis and diagnostic delay:

• Properly expose the area to be examined.
• Perform an appropriately thorough physical examination (i.e. multisystem exam).
• Perform a complete examination of a given system.

• In CMPA files involving diagnostic delay, neurological examinations, for example, are frequently omitted or documented as “grossly normal”, which yields little information about what was specifically tested.

• Examine all the areas that the history indicates should be examined.

• In CMPA files involving diagnostic delay, clinically indicated breast, rectal, or genital examinations are often inappropriately omitted.

• Take, review, and repeat vital signs as appropriate.

• In CMPA files involving diagnostic delay, abnormalities in vital signs are often unaccounted for or unacknowledged.

• Perform examinations aimed at ruling out specific differential diagnoses.
• Perform examinations to rule out the worst-case scenario diagnosis.
• Re-examine the patient if the clinical condition of the patient has changed or prior to discharge after an observation period.
• Avoid indiscriminately relying on examinations performed by others or at another time.
• If an important examination cannot be performed due to lack of patient consent, document your discussion with the patient about the implications of their choice.

As is the case with history taking, the gradual and usually unconscious drifting away from methodical approaches to the physical examination may contribute to less thorough assessments (i.e. omitting to examine the abdomen because of a pending ultrasound, for instance) and can contribute to misdiagnosis and diagnostic delay. It is important for faculty to teach learners how to be thorough and reliable in examining patients, as expected for their own specialty, and to identify behavioural drift in order to enable a return to sound practices.

Collapse section

Generating a differential diagnosis is an important component of clinical reasoning. Often, especially as the result of increasing experience, differential diagnoses may be considered unconsciously and go undocumented. Expertise develops from cognitive restructuring of knowledge upon which providers act and the formal generation of a differential diagnosis allows the information that has been gathered from the history and physical exam to consciously synthesize.

Taking time to formally consider a reasonable differential diagnosis, particularly for complex or unexpected problems or uncertain diagnoses,

• forces the provider to slow down their thinking, allowing them to consider other possibilities
• helps the provider identify information gaps
• helps direct deliberate and appropriate investigations aimed at ruling out possibilities and confirming a final diagnosis
• helps choose investigations wisely and avoid the practice of “defensive medicine”

The documentation of a differential diagnosis helps demonstrate a physician's competence, prudence, and thoughtfulness. To minimize diagnostic delay:

1. Prioritize the list of differential diagnoses by likelihood, urgency and potential severity of the proposed conditions.
2. Actively seek additional information to rule in or out reasonable alternative diagnoses.
3. Actively and consciously consider what information supports the favoured diagnosis and what information doesn’t fit, and thus minimize cognitive biases.
4. Actively seek to rule out the worst case scenario or “must not miss” diagnosis.

Such an approach forms the basis of a reflective, deliberate practice, which is a key component of safe practice and learning. Furthermore, communicating the differential diagnosis process—the process of ruling things in or out—can be very helpful in forming strong patient partnerships.

Collapse section

When deciding whether to use an available—although limited—healthcare resource, it is important to use sound medical judgment and to act in the patient’s best interests.

• When ordering investigations, using the “shotgun approach” (where one orders many non-indicated tests out of a perceived abundance of caution) may lead to patient harm through false-positives and incidental discovery of insignificant issues.
• When completing requisitions, it is important to communicate sufficient pertinent information to allow the next care provider to understand the situation, orient their decisions, and facilitate triage.
• • Taking the time to include enough pertinent clinical and specimen information and verifying the correctness of patient identifiers fosters optimal inter-professional collaboration and timely diagnoses.
• Being familiar with current clinical practice guidelines for the investigation of a suspected condition can assist in requesting the appropriate investigation to make the diagnosis.
• Seeking a patient’s informed consent for any investigations or procedures is paramount, especially when those investigations entail significant risks or side effects.
• Explaining the reasons behind testing and the potential diagnosis it could uncover may enhance the patient’s engagement with the testing process and promote follow-through with testing and increase follow-up intentions.
• Putting protocols in place to enable the timely receipt, effective review, and appropriate management and follow-up of investigative tests helps decrease the risk of diagnostic delay.

Often, investigations are deferred pending the initial test results. To mitigate the risk of diagnostic delays, schedule re-assessments and document a plan for follow-up. This plan should include the conditions that might trigger further investigations or the changes in signs or symptoms that should trigger a re-assessment. 

Pregnant patients

Pregnant patients may require modifications to the diagnostic approach. Ruling out the possibility of pregnancy and documenting having done so are important aspects of providing safe medical care when testing could potentially harm a mother or fetus. If testing is indicated the consent discussion should include the special risks to the fetus and mother, as appropriate, and this discussion should be documented.

Choosing wisely

The Choosing Wisely Canada (CWC) campaign aims to help doctors and patients engage in conversations about unnecessary tests and treatments and make smart and effective choices to ensure high-quality care. In collaboration with national medical professional societies, it identifies evidence-based recommendations for tests, treatments, and procedures that support value-added care.⁷

• Unnecessary tests and treatments do not add value to care. In fact, they take away from care by contributing to the scarcity of resources, potentially exposing patients to harm, leading to more testing to investigate false positives, and contributing to stress for patients.
• As with any clinical practice guideline, physicians are not obligated to follow the CWC recommendations but should consider what is best for their individual patients.

Collapse section

Relying solely on the classic features of a disease may be misleading in some cases. The clinical presentation of a disease often varies and the non-specific nature of initial symptoms and signs may play a role in diagnostic delay. Often, symptoms and signs may require hours, days, or even months to develop.

Experiencing distress or anxiety about a particular case may cloud judgment and inhibit an accurate diagnosis.

To minimize the risk of misdiagnosis or diagnostic delay, ask yourself:

• Do I have sufficient understanding of the clinical presentation to offer an opinion on the diagnosis?
• Have I assessed what information supports my favoured diagnosis?
• Have I considered what information supports other diagnoses?
• Have I ruled out a potentially serious diagnosis?
• How might the treatment to date have altered the clinical picture?

This can be a valuable approach when assuming the care of a patient. Reformulating the differential diagnosis may be prudent, especially if the clinical picture is evolving, if the diagnosis is not yet firmly established, or if the clinical care to date has not resolved the concern.

One way to improve diagnostic reasoning is to leverage the process of documenting to force a “cognitive pause” to consider other alternatives. Many practitioners use the SOAP (Subjective, Objective, Assessment, Plan) format for writing a clinical note. Using the FOAM approach to populate the “assessment” portion of the note can help broaden factors under consideration, identify information gaps, lead to the consideration of other diagnostic possibilities and help demonstrate your diligent reasoning.

To use the FOAM approach, ask yourself:

• What information FAVOURS the most likely diagnosis?
• What information supports OTHER diagnoses?
• What ADDITIONAL information would help narrow down the diagnosis?
• What information would support or suggest a MUST-NOT-MISS diagnosis?

Multiple return visits

Evolving or missed diagnoses often lead to repeat visits. When faced with a patient returning several times for the same symptoms, physicians may fall victim to cognitive or affective biases. Rather than dismissing patients who are returning for the same symptoms, using the situation as a trigger to recognize the likelihood of a diagnostic error may help avoid prolonging a diagnostic delay. Performing a new evaluation and searching for alternative diagnoses may lead to the correct diagnosis.

Acknowledging uncertainty and mitigating for the possibility of diagnostic error through the provision of individualized discharge instruction help establish an important safety net.

In addition, the likelihood of misdiagnosis and diagnostic delay may be decreased by seeking advice from others, engaging in diagnostic huddles, and implementing a strong safety culture that encourages speaking up.

Collapse section

Artificial intelligence (AI) is broadly defined as “machine learning,” referring to automation as a means to process information and to learn patterns that can be linked to big data analytics.⁸,⁹ AI is not intended to replace the physician in establishing diagnoses but rather to support the physician’s thinking and reasoning in light of increasingly complex practices.

It is anticipated that AI will soon become a crucial component of medical decision-making. Many EMRs incorporate clinical decision support tools using AI. When such tools are enabled, it is important to take the time to consider the algorithm’s suggestions and to critically assess them. Audit trail functions of EMRs can track the amount of time a provider spends viewing the recommendations of a clinical decision support tool and, while not obligated to follow its recommendations, a provider should be able to justify their decisions to ignore an AI-generated recommendation.

Consider applicable policies or guidelines of your College or health institution before using AI decision support tools. Remember:

• These tools are intended to support your clinical decisions and may help to broaden your differential diagnosis.
• The actual decisions and care you provide should be based on objective evidence and sound medical judgment.

Collapse section

A complete medical record facilitates further care and provides the best evidence of the encounter.

A good clinical note should enable a future reader to understand the details of the patient’s presentation as well as the previous provider’s clinical reasoning in reaching a diagnosis and treatment plan. As such, they should be able to provide care with an understanding of what has already been done and how that might influence their future decisions.

It is important to document the following:

• pertinent positive and negative historical and physical examination findings
• differential diagnoses
• diagnostic investigations and results
• treatment plan
• consent discussions
• patient reassessments including time and date
• specific patient discharge and follow-up instructions
• no-show patients and re-bookings as appropriate
• handouts given and discussed with patients
• case-specific discussions with consultants (including name, time, and advice received)

Collapse section