Hemoglobin A1c (≥6.5%) — Type 2 Diabetes
Test Characteristics
| Metric | Value |
|---|---|
| False-negative rate | 49% (sensitivity 51%) |
| False-positive rate | 4% (specificity 96%) |
| Bayes factor (positive test) | 10× |
| Bayes factor (negative test) | 1/2× |
| Base rate | 5% undiagnosed diabetes among US adults |
Interpreting Results
| Scenario | Prior | + Result | − Result |
|---|---|---|---|
| Unselected adult | 5% | 10 × 5% = 50%5%× 1050% | 1/2 × 5% = 2.5%5%÷ 22.5% |
| Obese, family history, acanthosis nigricans | 15% | 10 × 15% ≥ 100%15%× 10100% | 1/2 × 15% = 7.5%15%÷ 27.5% |
10 × 15% ≥ 100%15%× 10100%: exact posterior is 64%. A positive A1c is strong evidence (10×). But the negative Bayes factor (1/2×) is remarkably weak — A1c misses half of diabetics. For the high-risk patient, a normal A1c still leaves 7.5% probability. Consider fasting glucose or OGTT if suspicion remains.
- + result: risk goes from vegetarian (5%) to coin flip (50%)
- − result: only drops to near red hair (2.5%) — it misses half of diabetics
Sources:
- Frontiers systematic review, 2023. Frontiers in Medicine.
- ADA. Diabetes diagnosis & tests.