Skip to content

Zebra Hunting vs. Common Things

Why It Matters

Medical education emphasizes rare, interesting conditions. This creates a bias toward exotic diagnoses that are intellectually satisfying but statistically unlikely. Conversely, dismissing all uncommon presentations as common diseases creates missed diagnoses. The tool corrects both errors by providing a framework for when to pursue rare diagnoses and when to stick with common ones.

The Key Move

Start with common explanations for any presentation. Only pursue rare diagnoses when:

  1. common diagnoses have been adequately excluded,
  2. the presentation has features that are atypical for common diseases, or
  3. treatment for common diseases has failed. The threshold for investigating rare diseases should be proportional to their rarity - you need stronger evidence to justify evaluating a 1-in-10,000 condition than a 1-in-100 condition.

Classic Application

Headache evaluation. Common causes: tension headache, migraine, sinus infection. Rare causes: brain tumor, temporal arteritis, subarachnoid hemorrhage. A young patient with gradual headache onset and stress triggers gets treated for tension/migraine without extensive imaging. An older patient with sudden severe headache ("worst of my life") and neck stiffness gets immediate CT/LP for subarachnoid hemorrhage despite its rarity. The presentation characteristics justify pursuing the zebra.

Surprising Application

Performance regression debugging. Application suddenly slows down. Common causes: increased load, database query regression, memory leak. Rare causes: cosmic ray bit flip, hardware failure, malicious actor. Start with common causes (check metrics, review recent deployments). Only investigate rare causes if common ones are excluded or if presentation is atypical (slowdown correlates with specific hardware node, suggesting hardware failure).

Failure Modes

Premature closure on common diseases without adequate evaluation - "probably just anxiety" for chest pain misses actual cardiac disease. The tool isn't "always assume common diseases;" it's "require progressively stronger evidence for progressively rarer diagnoses." Also fails when base rates are genuinely uncertain (novel contexts) or when the patient population is pre-selected for rare conditions (specialty clinic vs. primary care). The worst failure is missing a zebra because you refused to look for it despite accumulating evidence.

Go Deeper

Sackett, D.L. (1992). "A Primer on the Precision and Accuracy of the Clinical Examination," JAMA, 267(19), 2638-2644; Sotos, J.G. (2006). "Zebra Cards: An Aid to Obscure Diagnoses," on when rare diagnoses deserve consideration.