Discrimination of cases of BL from controls based on different log10 EBV copies per milliliter thresholds
| Threshold selection method∗ . | Log10 EBV copies per milliliter† . | Correctly classified . | Misclassified . | Sensitivity . | Specificity . | Accuracy‡ . | Youden’s Index§ . | ||
|---|---|---|---|---|---|---|---|---|---|
| Cases of BL . | Controls . | Cases of BL . | Controls . | ||||||
| All samples, N = 800 | |||||||||
| Maximum sensitivity | 1.35 | 309 | 338 | 91 | 62 | 77.3% | 84.5% | 80.9% | 61.8% |
| 1.50 | 297 | 360 | 103 | 40 | 74.3% | 90.0% | 82.1% | 64.3% | |
| 2.00 | 281 | 377 | 119 | 23 | 70.3% | 94.3% | 82.3% | 64.6% | |
| 2.50 | 271 | 391 | 129 | 9 | 67.8% | 97.8% | 82.8% | 65.6% | |
| Based on Xian et al23 | 2.57 | 270 | 393 | 130 | 7 | 67.5% | 98.3% | 82.9% | 65.8% |
| 3.00 | 259 | 396 | 141 | 4 | 64.8% | 99.0% | 81.9% | 63.8% | |
| 3.50 | 249 | 398 | 151 | 2 | 62.3% | 99.5% | 80.9% | 61.8% | |
| 4.00 | 238 | 399 | 162 | 1 | 59.5% | 99.8% | 79.6% | 59.3% | |
| Maximum specificity | 4.19 | 233 | 400 | 167 | 0 | 58.3% | 100.0% | 79.1% | 58.3% |
| Visual inspection | 2.21 | 278 | 385 | 122 | 15 | 69.5% | 96.3% | 82.9% | 65.8% |
| Bootstrap method | 2.27 | 275 | 387 | 125 | 13 | 68.8% | 96.8% | 82.8% | 65.6% |
| Excluding indeterminate samples,‖N = 689 | |||||||||
| Maximum sensitivity | 1.35 | 309 | 277 | 41 | 62 | 88.3% | 81.7% | 85.1% | 70.0% |
| 1.50 | 297 | 299 | 53 | 40 | 84.9% | 88.2% | 86.5% | 73.1% | |
| 2.00 | 281 | 316 | 69 | 23 | 80.3% | 93.2% | 86.6% | 73.5% | |
| 2.50 | 271 | 330 | 79 | 9 | 77.4% | 97.3% | 87.2% | 74.7% | |
| Based on Xian et al23 | 2.57 | 270 | 332 | 80 | 7 | 77.1% | 97.9% | 87.4% | 75.0% |
| 3.00 | 259 | 335 | 91 | 4 | 74.0% | 98.8% | 86.2% | 72.8% | |
| 3.50 | 249 | 337 | 101 | 2 | 71.1% | 99.4% | 85.1% | 70.5% | |
| 4.00 | 238 | 338 | 112 | 1 | 68.0% | 99.7% | 83.6% | 67.7% | |
| Maximum specificity | 4.19 | 233 | 339 | 117 | 0 | 66.6% | 100.0% | 83.0% | 66.6% |
| Visual inspection | 2.21 | 278 | 324 | 72 | 15 | 79.1% | 95.6% | 87.4% | 75.0% |
| Bootstrap method | 2.27 | 275 | 326 | 75 | 13 | 78.6% | 96.2% | 87.2% | 74.8% |
| Threshold selection method∗ . | Log10 EBV copies per milliliter† . | Correctly classified . | Misclassified . | Sensitivity . | Specificity . | Accuracy‡ . | Youden’s Index§ . | ||
|---|---|---|---|---|---|---|---|---|---|
| Cases of BL . | Controls . | Cases of BL . | Controls . | ||||||
| All samples, N = 800 | |||||||||
| Maximum sensitivity | 1.35 | 309 | 338 | 91 | 62 | 77.3% | 84.5% | 80.9% | 61.8% |
| 1.50 | 297 | 360 | 103 | 40 | 74.3% | 90.0% | 82.1% | 64.3% | |
| 2.00 | 281 | 377 | 119 | 23 | 70.3% | 94.3% | 82.3% | 64.6% | |
| 2.50 | 271 | 391 | 129 | 9 | 67.8% | 97.8% | 82.8% | 65.6% | |
| Based on Xian et al23 | 2.57 | 270 | 393 | 130 | 7 | 67.5% | 98.3% | 82.9% | 65.8% |
| 3.00 | 259 | 396 | 141 | 4 | 64.8% | 99.0% | 81.9% | 63.8% | |
| 3.50 | 249 | 398 | 151 | 2 | 62.3% | 99.5% | 80.9% | 61.8% | |
| 4.00 | 238 | 399 | 162 | 1 | 59.5% | 99.8% | 79.6% | 59.3% | |
| Maximum specificity | 4.19 | 233 | 400 | 167 | 0 | 58.3% | 100.0% | 79.1% | 58.3% |
| Visual inspection | 2.21 | 278 | 385 | 122 | 15 | 69.5% | 96.3% | 82.9% | 65.8% |
| Bootstrap method | 2.27 | 275 | 387 | 125 | 13 | 68.8% | 96.8% | 82.8% | 65.6% |
| Excluding indeterminate samples,‖N = 689 | |||||||||
| Maximum sensitivity | 1.35 | 309 | 277 | 41 | 62 | 88.3% | 81.7% | 85.1% | 70.0% |
| 1.50 | 297 | 299 | 53 | 40 | 84.9% | 88.2% | 86.5% | 73.1% | |
| 2.00 | 281 | 316 | 69 | 23 | 80.3% | 93.2% | 86.6% | 73.5% | |
| 2.50 | 271 | 330 | 79 | 9 | 77.4% | 97.3% | 87.2% | 74.7% | |
| Based on Xian et al23 | 2.57 | 270 | 332 | 80 | 7 | 77.1% | 97.9% | 87.4% | 75.0% |
| 3.00 | 259 | 335 | 91 | 4 | 74.0% | 98.8% | 86.2% | 72.8% | |
| 3.50 | 249 | 337 | 101 | 2 | 71.1% | 99.4% | 85.1% | 70.5% | |
| 4.00 | 238 | 338 | 112 | 1 | 68.0% | 99.7% | 83.6% | 67.7% | |
| Maximum specificity | 4.19 | 233 | 339 | 117 | 0 | 66.6% | 100.0% | 83.0% | 66.6% |
| Visual inspection | 2.21 | 278 | 324 | 72 | 15 | 79.1% | 95.6% | 87.4% | 75.0% |
| Bootstrap method | 2.27 | 275 | 326 | 75 | 13 | 78.6% | 96.2% | 87.2% | 74.8% |
The threshold selected in the prior study by Xian et al,23 visual inspection, or bootstrap method to optimize Youden’s Index.
The thresholds displayed start with the log10 copies per milliliter value (1.35) that would classify all those with detectible EBV as meeting the threshold, then possible thresholds are presented in increments of 0.50 log10 copies per mL to the threshold (4.19) that would produce a specificity of 100.0%.
Proportion of correct classifications calculated as the number of participants with BL and controls correctly classified divided by the total.
Composite measure of sensitivity and specificity, calculated by adding the sensitivity and specificity, and subtracting 1.
Indeterminate samples were those testing both RPP30 and EBV negative (RPP30 negative, but EBV positive samples were included).