How a Test Result Shifts Your Probability Estimate

Here is something clinicians do every single time they order a test — they start with a guess. That guess is called **pre-test probability**: your best estimate of how likely a patient has a disease *before* any test result comes in. You build it from the patient's age, symptoms, risk factors, and how common the disease is in people like them. If a 45-year-old smoker comes in with chest pain, your pre-test probability for coronary artery disease might be 60%. If a healthy 17-year-old with the same chief complaint walks in, it might be 5%. Now you order a test. Here is the key idea: **the test result does not give you a yes-or-no answer. It shifts your probability.** A positive result pushes your estimate upward; a negative result pushes it downward. How far it moves depends on two properties of the test: - **Sensitivity**: how good the test is at detecting true disease (a highly sensitive test that comes back negative powerfully rules the disease out — remembered by the mnemonic SnNout: **S**e**n**sitive, **N**egative, rules **out**). - **Specificity**: how good the test is at excluding disease in healthy people (a highly specific test that comes back positive powerfully rules the disease in). Clinicians combine these into a single number called the **likelihood ratio (LR)**. A positive result uses LR+ = sensitivity ÷ (1 − specificity). A negative result uses LR− = (1 − sensitivity) ÷ specificity. **The three-step calculation:** 1. Convert pre-test probability to odds: **odds = p ÷ (1 − p)** 2. Multiply by the likelihood ratio: **post-test odds = pre-test odds × LR** 3. Convert back to probability: **post-test probability = odds ÷ (1 + odds)** **Worked example:** A patient has a 10% pre-test probability for a disease. The test is positive, and LR+ = 10. - Step 1: Pre-test odds = 0.10 ÷ 0.90 = 0.111 - Step 2: Post-test odds = 0.111 × 10 = 1.11 - Step 3: Post-test probability = 1.11 ÷ 2.11 ≈ **53%** Notice how a seemingly low 10% starting point jumps to 53% with a strong positive result. Now flip it: if the pre-test probability had been only 2%, the same LR+ = 10 gives post-test odds of 0.020 × 10 = 0.204, and post-test probability = 0.204 ÷ 1.204 ≈ **17%** — still probably not the disease despite a positive test. The crucial insight: **a moderately or even strongly positive test on a very-low-probability patient may still leave post-test probability below a treatment threshold.** This is why good clinicians think in probabilities, not just positive/negative — and why mastering this concept prevents both over-diagnosis and missed diagnoses.