Legal

Medical Cognitive Error

search

Medical Cognitive Error, Cognitive Bias, Decision Making Strategy, Cognitive Mistakes in Medicine, Anchoring Bias, Confirmation Bias, Diagnosis Momentum, Availability Bias, Representative Restraint, Visceral Bias, Blois' Funnel, Diagnostic Inference, Value-Induced Bias, Loss Aversion, Non-linear Probability Weighting, Base Rate Fallacy, Base Rate Neglect, Base Rate Bias, Diagnostic Time-Out, Diagnosis Premature Closure, Cognitive Dispositions to Respond in Medicine, SPIT Mnemonic, Thin Slicing, Thick Slicing, Self-Talk Incrementalization, Risk Shifting

  • Background
  1. Diagnostic Inference
    1. Revise opinion based on imperfect information
    2. Associated with predictable patterns of bias (see cognitive mistakes below)
  2. Blois' Funnel
    1. Differential diagnosis is honed and refined during the course of patient evaluation
  1. Energy and focus diminish over the course of a work shift
  2. End of shift Fatigue is common and may result in less ideal management decisions
  3. Corrective Strategies
    1. Take a 5-10 min break to recharge, walk, eat, drink every 3-4 hours
    2. Remove yourself from distracting conversations or activities
    3. Reassess your patient list every 2-3 hours
      1. Write down each patient, their acuity, and remaining barriers to disposition
      2. Starting with highest acuity patient, complete next task
    4. Make an exit plan in the final 1-2 hours of a shift
      1. Which patients will need admission, transfer to another facility, or signout to oncoming provider?
      2. Which patients need additional information or Consultation to make a disposition decision?
  4. Examples
    1. Antibiotics are more often prescribed for likely Viral Infections towards shift end
      1. Linder (2014) JAMA Intern Med 174(12): 2029-31 [PubMed]
  5. References
    1. Orman in Herbert (2017) EM:Rap 17(3): 7-8
  • Precautions
  • Interruptions are frequent in the emergency department (6-7/hour)
  1. Multi-tasking is a misnomer, and instead tasks are switched
    1. Task displaced by interruption is returned to after a mean delay of 23 minutes
    2. High cognitive load and frequent task switching is a risk for errors
  2. Corrective Strategies
    1. Offload tasks to other members of the team or to a list
      1. Example: Observation of monitor while performing high cognitive load tasks
      2. Example: Write down tasks that need to be completed when at a computer
    2. Delay or defer interruptions during important tasks
      1. Critical patient decision making
      2. Patient sign-out
  3. References
    1. Lin and Skaugset in Herbert (2017) EM:Rap 17(3): 5-6
    2. Westbrook (2010) BMJ Qual Saf 19(4)
    3. Skaugset (2016) Ann Emerg Med 68(2): 189-95 +PMID:26585046 [PubMed]
  • Causes
  • Cognitive Mistakes
  1. Anchoring Bias
    1. Basing decisions on a single piece or cluster of initial information
    2. Focusing on a single diagnosis early in the decision making process
      1. May result in mis-interpretation or ignoring conflicting later data
  2. Confirmation Bias
    1. Data acquisition and interpretation is used to confirm rather than refute a single hypothesis
    2. Exacerbates Anchoring Bias
  3. Diagnosis Premature Closure (Search Satisfaction)
    1. Differential diagnosis evaluation stops after one diagnosis is found or ruled-out
    2. Alternative diagnoses are not considered and additional data is not pursued
  4. Diagnosis Momentum
    1. Previous treating physician's working diagnosis is carried forward
    2. May be exacerbated by labels or stereotypes (e.g. intoxicated or mentally ill)
  5. Availability Bias (Availability Heuristic)
    1. Likelihood of diagnosis is based on the ease of recall of similar cases or episodes
    2. May be influenced by dramatic cases (e.g. litigated cases)
  6. Representative Restraint
    1. Clinicians focus on classic, but rare disease presentations
      1. At the expense of missing atypical, but common disease presentations
    2. Representative heuristics have an insensitivity to pretest probability
    3. Countered by the mantra, "When you hear hoof beats, think horses, not zebras"
  7. Visceral Bias
    1. Clinicians may be misguided by their own emotions and state of mind
    2. Medical decision making may be impacted by negative impressions of the mentally ill or intoxicated
  8. Value-Induced Bias
    1. Overestimation of disease outcome probability based on an outcome value
    2. Example: Imaging is ordered for a new onset Headache without red flags
      1. Ordered due to the value of missing the rare Brain Tumor
  9. Environment (example Emergency Department)
    1. High levels of diagnostic uncertainty
    2. Decision making at a rapid pace and a high volume
    3. Frequent interruptions
  10. Loss Aversion
    1. Humans are risk averse and often over emphasize risks or losses, and under emphasize benefits or gains
    2. This is typically applied to financial decisions (e.g. losing $20 has a greater impact than finding $20)
    3. Offer equivalent patient options (e.g. aggressive Resuscitation versus aggressive comfort measures)
  11. Diminishing Sensitivity to Losses and Gains
    1. Small additive risks are underestimated (e.g. CT radiation)
    2. Outweighed by greater theoretical risk concern (e.g. Abdominal Pain of unclear cause)
  12. Non-linear Probability Weighting
    1. We over-estimate the risk of rare events (plane crash)
    2. We under-estimate the risk of common events (Motor Vehicle Accidents)
  13. Base Rate Fallacy (Base Rate Neglect, Base Rate Bias)
    1. When presented with both general information as well as specific information
      1. People are guided by the specific information
    2. Specific information may lead to assume an alternative, rare diagnosis
      1. Whereas the most likely diagnosis is the much more common one
    3. For a dry cough of 1-2 weeks, Bronchitis would be much more likely than Coccidioidomycosis
      1. Even if you are told they just traveled to Arizona
  14. Risk Shifting
    1. Medical decision making requires some amount of shifting a patients risk to the provider
    2. Providers assume a risk of missed diagnosis by avoiding additional testing (e.g. CT Imaging)
    3. Evaluation is a balance between a patient's condition risk and a provider's level of risk acceptance
      1. Shared Decision Making is an educated negotiation of what level of investigation is sufficient
  15. Diagnostic Overshadowing
    1. Known underlying condition results in medical provider mistaken assumptions and misdiagnosis
    2. New and serious condition is missed, confused for an exacerbation of the underlying condition
    3. Iezzoni (2019) N Engl J Med 380(22): 2092-3 [PubMed]
    4. Lazris (2023) Am Fam Physician 108(3): 292-4 [PubMed]
  • Prevention
  • Mnemonic - "When U RACE, tie your LACES"
  1. When U RACE
    1. Unexplained complaint
      1. Initial hypothesis-driven data does not result in a diagnosis to explain symptom presentation
    2. Return Visit
      1. Risk of Anchoring Bias and Diagnosis Momentum based on the evaluation already undertaken
      2. Return visit is an opportunity to take a fresh look and possibly catch prior diagnostic error
    3. At-Risk patient
      1. Very young or very old patients
      2. Mentally ill patients
      3. Intoxicated patients
      4. Immunocompromised patients
    4. Critical Condition
      1. Time sensitive diagnosis and management (e.g. MI or CVA)
      2. Results in harmful intervention (e.g. TPA in Aortic Dissection)
    5. End of Shift
      1. Fatigue is a ripe milieu for diagnostic error
      2. End-of-shift hand-offs are complicated by incomplete communication
      3. Risk of Anchoring Bias and Diagnosis Momentum
  2. Tie your LACES
    1. Life-threatening diagnoses fully considered?
      1. Consider worst-case scenarios for a given presentation
    2. Anything else possible on the differential diagnosis?
      1. Avoid premature closure by adequately considering alternative diagnoses
    3. Coherent explanation?
      1. Consistency between diagnosis and clinical findings
    4. Everything explained (Adequacy)?
      1. Normal and abnormal findings are ALL explained by the final diagnosis
    5. Second problem present?
      1. Could more than one diagnosis better account for the current presentation
  3. References
    1. Gordon and Kemnitz (2013) Crit Dec Emerg Med 27(12): 11-18
  • Approach
  • Decision Making Strategy
  1. Effective emergency decision making is a combination of thin and Thick Slicing
  2. Thin Slicing (fast and intuitive)
    1. Formulate initial patient plan based on limited information, gestalt, pattern recognition and intuition
    2. Allows for rapid emergency decision making
    3. Often accurate, especially in more expert clinicians (but risky in medical trainees)
    4. Pitfalls
      1. Requires experience, training and practice, with constant exposure to cases, simulations, peer review
      2. Risk of anchor bias and Confirmation Bias
      3. Processing occurs in the prefrontal cortex (hidden brain)
      4. Prefrontal cortex is easily overwhelmed by large amounts of incoming information
    5. Checklists help to avoid major pitfalls and anchoring
      1. Clinical Decision Rules
      2. Review of Systems (clusters of outlier symptoms may redirect diagnosis)
      3. Life-threatening differential diagnosis list (e.g. PE, Aortic Dissection, Mesenteric Ischemia)
  3. Thick Slicing (slow and logical)
    1. Deeper evaluation and analysis of clinical data including subtle findings
    2. Allows for reconsideration of data, differential diagnosis before disposition
    3. May involve checklists and reference review
    4. Consider pre-rehearsed scripts for the safe evaluation and management of complex presentations
    5. Slower than Thin Slicing and requires setting aside time to collect and review available data
  4. Resources
    1. Gladwell (2007) The Power of Thinking Without Thinking, Back Bay Books
      1. Paid link to Amazon.com (ISBN 9780316010665)
    2. Gawande (2011) Checklist Manifesto: How to Get Things Right, Picador
      1. Paid link to Amazon.com (ISBN 0312430000)
  5. References
    1. Herbert and McCollum in Herbert (2016) EM:Rap 16(9): 4-6
    2. Jaban in Hewrbert (2017) EM:Rap 17(5): 5-6
  1. At the outset of visit, explain that evaluation and testing is imperfect
    1. No diagnostic test, exam, imaging is certain
    2. Miss rates and False Negatives are common
  2. In the pursuit of a diagnosis, there are risks to testing
    1. False Positives (e.g. Low Risk Chest Pain)
    2. CT-associated Radiation Exposure
  3. Likelihood of a diagnosis for a specific patient and their presentation should guide testing
    1. Balance the risks and benefits of testing, the likelihood and risk of a missed diagnosis
  4. Medications and other treatment are imperfect
    1. Pain will not be completely relieved
    2. Medications have adverse effects (e.g. Opioid Abuse, Constipation)
    3. Medications are often prescribed when not indicated (e.g. antibiotics) with secondary harm (e.g. c. diff)
  • Approach
  • Self-Talk Incrementalization
  1. Break down complicated and high risk situations into manageable tasks
    1. Concentrate on each single task, with multiple adjustment strategies available if difficult
    2. Slow down, and perform each task with adequate precision to ensure success
  2. Self-talk yourself into success at each step
    1. Perform each task with confidence of success
  3. Related measures
    1. Maintain a sense of self efficacy and self confidence
    2. Control your breathing (employed in yoga, stress reduction, Mindfulness)
    3. Practice for stressful situations
      1. Simulations
      2. Mental rehearsal
  4. References
    1. Mason and Levitan in Herbert (2017) EM:Rap 17(12): 19-20
    2. Swaminathan and Hicks in Herbert (2018) EM:Rap 18(9): 12-4
  • Approach
  • Develop a thorough differential diagnosis (SPIT Mnemonic)
  1. Serious possible diagnoses
    1. Exclude life threatening diagnoses (where appropriate)
  2. Probable diagnoses
    1. Evaluate and treat likely diagnoses
  3. Interesting diagnoses
    1. Consider uncommon, complex diagnoses (where appropriate)
  4. Treatable diagnoses
    1. Consider diagnoses with specific, effective treatment regimens (and empiric therapy trial)
  5. References
    1. Bukata (2013) EM Bootcamp, Approach to the ED Patient
  • References
  1. Gordon and Kemnitz (2013) Crit Dec Emerg Med 27(12): 11-18
  2. McCollum in Herbert (2018) EM:Rap 18(2): 11-3
  3. Menchine in Majoewsky (2012) EM:RAP 12(2): 1-2