Skip to content
Download PDF

Decision-Making in Clinical Medicine

Chapter 4 | Part 1: The Profession of Medicine · Part 1 – The Profession of Medicine · Chapter 4

Key Clinical Points

  1. Clinical expertise integrates disease knowledge, pattern recognition, and communication skills.
  2. Dual-process theory distinguishes Intuitive Reasoning (System 1) from Analytic Reasoning (System 2).
  3. Heuristics like representativeness, availability, and anchoring can lead to diagnostic errors.
  4. Premature closure occurs when a clinician stops the diagnostic process before all data are considered.
  5. Diagnostic imperatives require ruling out rare but catastrophic conditions regardless of low prevalence.
  6. Sensitivity (SnNout) is used to rule out disease; Specificity (SpPin) is used to rule in disease.
  7. Bayes' Rule quantifies posttest probability by combining pretest probability with test sensitivity and specificity.
  8. Likelihood Ratio (LR) measures the strength of a test's evidence for a diagnosis.
  9. ROC curves illustrate the trade-off between sensitivity and specificity; Area Under Curve (AUC) measures information content.
  10. Personalized medicine combines evidence, patient-specific features, and shared decision-making.

DEFINITION & OVERVIEW

• Clinical Expertise: Definition (Harrison's 22e): "Clinical expertise encompasses not only cognitive dimensions involving the integration of disease knowledge with verbal and visual cues and test interpretation but also potentially the complex fine-motor skills necessary for invasive procedures and tests." • Components of Expertise: - Integration of disease knowledge with verbal/visual cues - Test interpretation - Fine-motor skills (for invasive procedures) - Communication and care coordination • Clinical Reasoning: Process of making decisions in the face of intrinsic uncertainty. • Evidence-Based Medicine (EBM): Integration of best available research evidence with clinical judgment for individual patients.

Dual-Process Theory

• Intuitive Reasoning (System 1): - Characteristics: Rapid, effortless, based on memorized associations. - Mechanism: Pattern recognition and "rules of thumb" (heuristics). • Analytic Reasoning (System 2): - Characteristics: Slow, methodical, deliberative, effortful. - Requirement: Necessary when pattern recognition is insufficient or outside the clinician's expertise. • Pattern Recognition: - Requires a large library of stored patterns. - Risk: Without deliberate systematic reflection, can lead to premature closure. • Diagnostic Verification: - Testing adequacy (does it explain all symptoms/signs?) - Testing coherency (are signs/symptoms consistent with pathophysiology?)

ETIOLOGY & PATHOPHYSIOLOGY

• Cognitive Processing: - Clinicians group data into "chunks" for working memory. - Working memory is limited (traditionally 7 ± 2 items). • Expertise Advantages: - Larger store of exemplar/prototype cases. - Elaborate conceptual networks (illness scripts). - More knowledge of presenting symptoms and larger repertoire of cognitive tools.

Heuristics and Biases

• Heuristics: Rules of thumb used in the intuitive system. - "Heuristics and biases" program: Focuses on how shortcuts lead to incorrect judgments. - "Fast and frugal heuristics" program: Explores when simple heuristics produce good decisions. • Representativeness Heuristic: - Mechanism: Judging likelihood based on similarity to a stored pattern (mental representation). - Risk: Overestimating rare diseases with classic symptoms; underestimating common diseases with atypical presentations. • Availability Heuristic: - Mechanism: Judging likelihood based on ease of recall. - Sources of Error: - Rare catastrophic outcomes are more memorable/distinct. - Media coverage or recent experiences increase recall ease. • Anchoring Heuristic (Conservatism): - Mechanism: Insufficiently adjusting initial probability after a test result. - Risk: Sticking to an initial diagnosis despite new data. • Simplicity Heuristic (Occam's Razor): - Principle: Use the simplest explanation that accounts for all findings. - Risk: Premature closure; neglecting unexplained significant symptoms.

CLINICAL FEATURES

• Diagnostic Imperatives: - Definition: Recognition of rare but potentially catastrophic conditions that must be ruled out regardless of low prevalence. - Example: Aortic dissection (must be considered in all cases of acute severe chest discomfort). - Clinical Action: Routine inquiry into symptoms/signs of dissection, measurement of blood pressures in both arms, and assessment for pulse deficits.

Clinical Scenarios

• Hemoptysis Examples: - Case 1 (Healthy non-smoker): Pattern suggests acute bronchitis. - Case 2 (100-pack-year smoker, weight loss): Pattern suggests lung cancer. - Case 3 (Migrant with apical murmur): Pattern suggests pulmonary hypertension from rheumatic mitral stenosis. • Premature Closure Example: - Patient with "flulike" URI and dyspnea was treated for bronchitis because of a standardized assessment form. - Failure: Clinician failed to elicit full history; patient died of myocardial infarction.

INVESTIGATIONS & DIAGNOSIS

• Purpose of Testing: Reduce uncertainty about diagnosis or prognosis to facilitate management. • Test Accuracy Metrics: - Sensitivity (True-positive rate): Proportion of patients with disease who have a positive test. - Formula: TP / (TP + FN) - Rule: High sensitivity when negative helps rule out disease (SnNout). - Specificity (True-negative rate): Proportion of patients without disease who have a negative test. - Formula: TN / (TN + FP) - Rule: High specificity when positive helps rule in disease (SpPin). • ROC Curves (Figure 4-1): - Plot: Sensitivity vs. 1 - Specificity. - Area Under Curve (AUC): Measure of information content (0.5 = no info; 1.0 = perfect test). - Selection of Cut Point: Depends on the relative harms/benefits of treatment for those with vs. without disease.

Bayes' Rule and Likelihood Ratios

• Bayes' Rule: Quantifies revised uncertainty after a test. - Formula: Posttest probability = (Pretest probability × Test sensitivity) / [(Pretest probability × Test sensitivity) + ((1 - Pretest probability) × False-positive test rate)]. • Likelihood Ratio (LR): - Definition: Ratio of the probability of a given test result in a patient with disease to the probability of that result in a patient without disease. - Interpretation: - LR 2–5: Modest discriminatory ability. - LR > 10: Stronger evidence for diagnosis. • Clinical Examples (CAD): - Exercise Treadmill (LR 2.4): Low pretest probability (10%) → Posttest probability ~30%. - Exercise SPECT (LR 9.0): Low pretest probability (10%) → Posttest probability ~90%. • Table 4-1: Measures of Diagnostic Test Accuracy - Positive/Disease: True positives (TP) - Positive/No Disease: False positives (FP) - Negative/Disease: False negatives (FN) - Negative/No Disease: True negatives (TN) - Sensitivity: TP / (TP + FN) - Specificity: TN / (TN + FP) - False-negative rate: 1 – sensitivity - False-positive rate: 1 – specificity

MANAGEMENT & TREATMENT

• Personalized Medicine: - Goal: Combine best evidence with individual patient features (genomics, comorbidities) and preferences. • Two Levels of Personalization: 1. Precision Medicine: Individualizing risk of harm/benefit based on specific patient characteristics. 2. Shared Decision-Making: Incorporating patient's values and goals into the decision process.

Decision Making Process

• Evidence vs. Experience: - Avoid relying on personal experience for rare events or small sample sizes due to high risk of erroneous inference. • Clinical Scenarios (Table 4-2): - Table 4-2: Wells Clinical Prediction Rule for Pulmonary Embolism (PE) - Clinical signs of DVT: 3 points - Heart rate >100 bpm: 1.5 points - History of DVT/PE: 1.5 points - Malignancy: 1 point - Interpretation: Score >6.0 (High), 2.0–6.0 (Moderate), <2.0 (Low).

KEY PEARLS & HIGH-YIELD POINTS

• SnNout / SpPin: - High Sensitivity → Negative result → Rule Out. - High Specificity → Positive result → Rule In. • Bayes' Rule Application: - Use to avoid anchoring bias by calculating posttest probability based on pretest probability and LR. • Diagnostic Imperatives: - Always rule out high-stakes conditions (e.g., aortic dissection) even if they are statistically unlikely.

Reference Tables

TABLE 4-1 Measures of Diagnostic Test Accuracy

Harrison's 22e, p.24

DISEASE STATUS
TEST RESULT PRESENT ABSENT
Positive True positives (TP) False positives (FP)
False negatives (FN)
Test Characteristics in Patients with Disease
True-positive rate (sensitivity) = TP/(TP + FN)
False-negative rate = FN/(TP + FN) = 1 – true-positive rate
Test Characteristics in Patients without Disease

TABLE 4-2 Wells Clinical Prediction Rule for Pulmonary Embolism (PE)

Harrison's 22e, p.26

CLINICAL FEATURE POINTS
Clinical signs of deep-vein thrombosis 3
Heart rate >100 beats/min 1.5
History of deep-vein thrombosis or PE 1.5
Malignancy (with treatment within 6 months) or palliative 1
INTERPRETATION
Score >6.0 High
Score <2.0 Low