Table 4.
Measures of Diagnostic Accuracy for 6 Dermascopic Algorithms
| Measure | 7-Point Checklist (Cut Point ≥3) | CASH (Cut Point ≥6) | Menzies Method | ABCD Rule (TDS Score >4.75) | 3-Point Checklist | Chaos and Clues |
|---|---|---|---|---|---|---|
| Sensitivity, % (95% CI) | 70.6 (61.5–78.6) | 77.9 (69.7–85.1) | 95.1 (89.0–98.4)a | 74.8 (66.0–82.3) | 68.9 (59.8–77.1) | 82.4 (66.1–96.5) |
| Specificity, % (95% CI) | 57.5 (52.2–62.7) | 50.9 (45.4–56.4) | 24.8 (20.1–30.1)b | 59.4 (54.0–64.6) | 58.7 (53.4–63.8) | 40.2 (35.1–45.5)c |
| ROC area (95% CI) | 0.65 (0.59–0.69) | 0.65 (0.59–0.69) | 0.60 (0.57–0.63) | 0.66 (0.62–0.72) | 0.64 (0.59–0.69) | 0.66 (0.63–0.70) |
Abbreviations: CASH, color, architecture, symmetry, and homogeneity; ROC, receiver operating characteristic; TDS, total dermatoscopy score.
a
Sensitivity of the Menzies method was significantly higher than any other algorithm.
b
Specificity of the Menzies method was significantly lower than any other algorithm.
c
Specificity of chaos and clues was significantly lower than the 7-point checklist, the 3-point checklist, and the ABCD rule.