Overall and Individual Performance for Diagnosis of HL Versus DLBCL Lymphomas
Parameter | Overall | Observer A | Observer B | Observer C | Observer D |
---|---|---|---|---|---|
Proposed diagnosis: HL, DLBCL | 110–180 | 33–45 | 22–58 | 27–45 | 28–32 |
Correct HL classification | 80/140: 57.1% (49.0%–65.3%) | 23/32: 71.9% (56.3%–87.5%) | 14/36: 38.9% (23.0%–54.8%) | 20/35: 57.1% (40.8%–73.5%) | 27/37: 73.0% (58.7%–87.3%) |
Correct DLBCL classification | 91/111: 82% (74.8%–89.1%) | 35/41: 85.4% (74.5%–96.2%) | 24/29: 82.8% (69.0%–96.5%) | 24/31: 77.4% (62.7%–92.1%) | 8/10: 80% (55.2%–100%) |
When observer said cancer and gold standard was cancer: | |||||
Correct HL classification | 79/137: 57.7% (49.4%–65.9%) | 23/30: 76.7% (61.5%–91.8%) | 14/36: 38.9% (23.0%–54.8%) | 20/34: 58.8% (42.3%–75.4%) | 22/37: 59.5% (43.6%–75.3%) |
Correct DLBCL classification | 91/111: 82% (74.8%–89.1%) | 35/41: 85.4% (74.6%–96.2%) | 24/29: 82.8% (69.0%–96.5%) | 24/31: 77.4% (62.7%–91.1%) | 8/10: 80% (55.2%–100%) |
Sensitivity | 0.58 (0.49–0.66) | 0.77 (0.58–0.90) | 0.39 (0.23–0.57) | 0.59 (0.41–0.75) | 0.59 (0.42–0.75) |
Specificity | 0.82 (0.74–0.89) | 0.85 (0.71–0.94) | 0.83 (0.64–0.94) | 0.77 (0.59–0.90) | 0.80 (0.44–0.97) |
PPV | 0.80 (0.71–0.87) | 0.79 (0.60–0.92) | 0.74 (0.49–0.91) | 0.74 (0.54–0.89) | 0.92 (0.73–0.99) |
NPV | 0.61 (0.53–0.69) | 0.83 (0.69–0.93) | 0.52 (0.37–0.67) | 0.63 (0.46–0.78) | 0.35 (0.16–0.57) |
Accuracy | 170/248: 68.5% (62.7%–74.3%) | 58/71: 81.7% (72.7%–90.7%) | 38/65: 58.5% (46.5%–70.5%) | 44/65: 67.7% (56.3%–79.1%) | 30/47: 63.8% (50.1%–77.5%) |
AUC | 0.70 (0.64–0.75) | 0.81 (0.72–0.91) | 0.60 (0.50–0.72) | 0.68 (0.57–0.79) | 0.70 (0.54–0.85) |
Data in parentheses are 95% CIs.