Subinternship Medicine · Year 4 · from Subinternship Medicine
Case 2: Bayesian Reasoning and Test Interpretation
Clinical Image
Source: Wikimedia Commons - Laboratory Medicine - Public Domain
Case Presentation
A 25-year-old healthy woman presents with 2 days of left-sided pleuritic chest pain after a 4-hour car trip. She has no leg swelling, no prior VTE, no recent surgery, and is not on estrogen-containing contraceptives. Vital signs are normal with oxygen saturation of 99% on room air. Her Wells score for PE is 0 (low probability). The emergency physician orders a D-dimer, which returns elevated at 620 ng/mL (normal less than 500 ng/mL). The EP plans to order a CT pulmonary angiography. The medicine sub-intern asks about the pre-test probability. The supervising resident explains that with a Wells score of 0 and normal vital signs, her pre-test probability is approximately 3-5%. A D-dimer has sensitivity of approximately 95% and specificity of approximately 40% in this population. Using the Fagan nomogram or likelihood ratio calculation, even with a positive D-dimer (LR+ approximately 1.6), her post-test probability only rises to approximately 5-8%. The team considers whether the low positive D-dimer could be a false positive. Upon further history, the patient mentions she has had a cold for the past week with mild productive cough. Inflammation from viral infection commonly elevates D-dimer. After shared decision-making about the low probability of PE versus radiation exposure from CT, the patient opts for close follow-up rather than CT-PA. Her symptoms resolve over 5 days without treatment.
Key Learning Points
- Pre-test probability should be estimated before ordering tests; when pre-test probability is very low, even positive results may represent false positives
- Likelihood ratios quantify how much a test result changes probability: LR+ greater than 10 strongly increases probability, while LR+ of 1-2 provides minimal diagnostic value
- D-dimer has high sensitivity but low specificity, making it useful for ruling out PE in low-probability patients but not for ruling in PE
- Clinical decision rules (Wells, PERC) should guide test ordering; ordering tests indiscriminately leads to false positive cascades