Clinical Problem-Solving and Applied Knowledge
What the domain tests
Vignette-based items test whether you can move from a problem representation to a prioritised differential, choose discriminating features and act on red flags. The exam rewards Bayesian reasoning: post-test probability depends on pre-test probability and the likelihood ratio of the finding, not on the test result alone.
Core rules
- Represent the problem in one sentence. "A 62-year-old smoker with acute tearing chest pain and a pulse deficit" is a stronger representation than "chest pain".
- Use discriminating features. A pulse deficit and mediastinal widening on chest radiograph raise the probability of aortic dissection; they do not prove it.
- Act when unstable. In an unstable patient, treat the lethal diagnosis first and investigate in parallel.
- Reassess after every intervention. Response to treatment is diagnostic information.
Worked example
A 55-year-old hypertensive man has sudden severe anterior chest pain radiating to the back. Blood pressure is 190/100 mmHg in the right arm and 150/90 mmHg in the left, and the chest radiograph shows a widened mediastinum. Reasoning: tearing pain, an inter-arm systolic difference and mediastinal widening make aortic dissection the leading diagnosis. Do not give thrombolysis for presumed myocardial infarction. If stable, obtain urgent CT angiography; start intravenous beta-blockade to reduce dP/dt and control heart rate, then give analgesia. Type A dissection requires cardiothoracic surgery.
Traps
- Anchoring on the common diagnosis and stopping at "chest pain equals acute coronary syndrome".
- Premature closure after one normal test.
- Ignoring red flags because the patient looks well.
- Confusing the sensitivity of a finding with its specificity.