Sensitivity, Specificity, PPV, NPV

2x2 table: Disease+ Disease- Test+ TP FP Test- FN TN SENSITIVITY = TP / (TP + FN) = "% of diseased correctly identified" = SnNout — high Sn rules OUT disease (negative result reliable) = use for screening SPECIFICITY = TN / (TN + FP) = "% of healthy correctly identified" = SpPin — high Sp rules IN disease (positive result reliable) = use for confirmatory testing PPV = TP / (TP + FP) = "% of positives that have disease" = depends on prevalence (↓ prevalence → ↓ PPV) NPV = TN / (TN + FN) = "% of negatives that don't have disease" PREVALENCE = (TP + FN) / Total LR+ = Sn / (1 − Sp) — likelihood ratio for positive test LR− = (1 − Sn) / Sp — likelihood ratio for negative test NNT = 1 / Absolute Risk Reduction (number needed to treat) NNH = 1 / Attributable Risk (number needed to harm) STUDY DESIGNS (descending evidence): 1. Systematic review / meta-analysis of RCTs 2. RCT 3. Cohort 4. Case-control 5. Cross-sectional 6. Case series / case reports 7. Expert opinion