A diagnostic test for tuberculosis has a sensitivity of 95% and specificity of 85%. A competing test has a sensitivity of 80% and specificity of 98%. In a low-prevalence setting (1% disease prevalence), which feature best discriminates the clinical utility of these two tests?

## Comparing Diagnostic Tests in Low-Prevalence Settings ### Understanding the Scenario We have two tests with inverse sensitivity–specificity profiles: - **Test 1:** Sensitivity 95%, Specificity 85% - **Test 2:** Sensitivity 80%, Specificity 98% - **Disease prevalence:** 1% (low) ### Calculating Positive Predictive Value (PPV) Using the formula: $PPV = \frac{TP}{TP+FP} = \frac{Sensitivity \times Prevalence}{Sensitivity \times Prevalence + (1-Specificity) \times (1-Prevalence)}$ **Test 1 (Sensitivity 95%, Specificity 85%):** $$PPV_1 = \frac{0.95 \times 0.01}{0.95 \times 0.01 + 0.15 \times 0.99} = \frac{0.0095}{0.0095 + 0.1485} = \frac{0.0095}{0.158} \approx 6\%$$ **Test 2 (Sensitivity 80%, Specificity 98%):** $$PPV_2 = \frac{0.80 \times 0.01}{0.80 \times 0.01 + 0.02 \times 0.99} = \frac{0.008}{0.008 + 0.0198} = \frac{0.008}{0.0278} \approx 29\%$$ ### Key Insight: Specificity Dominates in Low-Prevalence Settings **High-Yield:** In low-prevalence populations, **specificity is the dominant determinant of PPV**. A highly specific test (Test 2) will have substantially better PPV despite lower sensitivity, because false positives are minimized when disease is rare. **Key Point:** Test 2's PPV (~29%) is nearly **5 times higher** than Test 1's PPV (~6%), making it clinically superior for confirming diagnosis in this setting. ### Negative Predictive Value (NPV) Comparison Using: $NPV = \frac{Specificity \times (1-Prevalence)}{Specificity \times (1-Prevalence) + (1-Sensitivity) \times Prevalence}$ **Test 1:** $NPV_1 = \frac{0.85 \times 0.99}{0.85 \times 0.99 + 0.05 \times 0.01} = \frac{0.8415}{0.8420} \approx 99.9\%$ **Test 2:** $NPV_2 = \frac{0.98 \times 0.99}{0.98 \times 0.99 + 0.20 \times 0.01} = \frac{0.9702}{0.9722} \approx 99.8\%$ NPVs are similar (both very high) because disease is rare — both tests rule out disease effectively. ### Clinical Application | Feature | Test 1 | Test 2 | Winner | |---------|--------|--------|--------| | **Sensitivity** | 95% | 80% | Test 1 | | **Specificity** | 85% | 98% | Test 2 | | **PPV (at 1% prev.)** | ~6% | ~29% | **Test 2** | | **NPV (at 1% prev.)** | ~99.9% | ~99.8% | Test 1 | | **Use case** | Screening (rule-out) | Confirmation (rule-in) | **Test 2 for diagnosis** | **Clinical Pearl:** Test 2 is superior for **confirming TB** in low-prevalence settings (e.g., non-endemic regions). A positive result is much more likely to represent true disease. Test 1 would be better for screening, where high sensitivity is needed to avoid missing cases. **Mnemonic:** **"High Spec = High PPV in Low Prev"** — When disease is rare, specificity drives the positive predictive value.