Clinical Assessment, Diagnosis, and Diagnostic Reasoning
Assessment is a probability exercise
A focused history (onset, provocation, quality, radiation, severity, timing, associated symptoms, risk factors, medications) plus a targeted exam should generate a ranked differential within minutes. Expert reasoning is Bayesian: estimate pretest probability from age, comorbidity, and vital signs; choose the test that will move you across a management threshold; then act on post-test probability. A highly sensitive test is best for ruling out when negative, a highly specific test for ruling in when positive, and likelihood ratios are more informative than either alone.
Know the validated rules - and their populations
Ottawa ankle and knee rules, Canadian CT Head and C-Spine rules, NEXUS, PERC and Wells for PE, HEART, NIHSS, PECARN. A rule applied outside its derivation population (pregnancy, anticoagulation, extreme age) is a classic distractor.
Traps
- Atypical presentations: women, older adults, and patients with diabetes or immunosuppression may lack classic symptoms. "Indigestion" plus diaphoresis in a 58-year-old woman still deserves an ECG and troponin.
- Anchoring and premature closure after the first abnormal finding; satisfaction of search after one fracture is found.
- Ordering imaging because a score was "positive" in a patient whose pretest probability was already very low.
Worked example
A 28-year-old with pleuritic chest pain, HR 88, SpO2 98%, no estrogen use, no prior VTE, no unilateral leg swelling, no hemoptysis, no recent surgery: PERC is negative, so no D-dimer and no CT. If she were 32 weeks pregnant, PERC would not apply - you would move to age-adjusted D-dimer and pregnancy-specific imaging algorithms instead.
Rule of thumb
Before every test ask two questions: will this result change management, and what harm follows a false positive?