The double-hump graph illustrates overlapping distributions of a healthy population and a diseased population along a continuous screening variable (e.g., blood pressure, glucose level).
Key Concepts
| Cutoff Point | Effect | Clinical Significance |
|---|---|---|
| A (Low) | Includes almost all individuals | High sensitivity, low specificity |
| B (Healthy) | Correctly identifies all sick patients | Highest sensitivity → identifies all true positives; optimal negative predictive value |
| C (Intersection of curves) | The point where healthy and diseased overlap equally | Most accurate cutoff; balances sensitivity and specificity |
| D (Diseased) | Only identifies patients far from healthy | Highest specificity → identifies true negatives; optimal positive predictive value |
| E (High) | Excludes many diseases | Low sensitivity, very high specificity |
- Sensitivity ↑ → cutoff shifts toward healthy (more true positives, more false positives)
- Specificity ↑ → cutoff shifts toward diseased (fewer false positives, more false negatives)
Relationships Between Cutoff and Predictive Values
| Parameter | Optimal at |
|---|---|
| Sensitivity / Negative Predictive Value | Cutoff B (toward healthy) |
| Specificity / Positive Predictive Value | Cutoff D (toward diseased) |
| Overall Accuracy | Cutoff C (intersection of curves) |
Key Points:
- Moving the cutoff left increases sensitivity (catch all diseased) but decreases specificity.
- Moving the cutoff right increases specificity (avoid labeling healthy) but decreases sensitivity.
- The intersection point balances both and maximizes overall test accuracy.
Visual Summary
Learning Objective
After reviewing this graph, medical students should be able to identify optimal cutoffs for sensitivity, specificity, and overall accuracy, and understand how shifting the threshold affects positive and negative predictive values in clinical screening tests.
You must be logged in to post a comment.