Table 2.

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 classifiedMisclassifiedSensitivitySpecificityAccuracy Youden’s Index§ 
Cases of BLControlsCases of BLControls
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 67.8% 97.8% 82.8% 65.6% 
Based on Xian et al23  2.57 270 393 130 67.5% 98.3% 82.9% 65.8% 
 3.00 259 396 141 64.8% 99.0% 81.9% 63.8% 
 3.50 249 398 151 62.3% 99.5% 80.9% 61.8% 
 4.00 238 399 162 59.5% 99.8% 79.6% 59.3% 
Maximum specificity 4.19 233 400 167 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 77.4% 97.3% 87.2% 74.7% 
Based on Xian et al23  2.57 270 332 80 77.1% 97.9% 87.4% 75.0% 
 3.00 259 335 91 74.0% 98.8% 86.2% 72.8% 
 3.50 249 337 101 71.1% 99.4% 85.1% 70.5% 
 4.00 238 338 112 68.0% 99.7% 83.6% 67.7% 
Maximum specificity 4.19 233 339 117 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 classifiedMisclassifiedSensitivitySpecificityAccuracy Youden’s Index§ 
Cases of BLControlsCases of BLControls
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 67.8% 97.8% 82.8% 65.6% 
Based on Xian et al23  2.57 270 393 130 67.5% 98.3% 82.9% 65.8% 
 3.00 259 396 141 64.8% 99.0% 81.9% 63.8% 
 3.50 249 398 151 62.3% 99.5% 80.9% 61.8% 
 4.00 238 399 162 59.5% 99.8% 79.6% 59.3% 
Maximum specificity 4.19 233 400 167 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 77.4% 97.3% 87.2% 74.7% 
Based on Xian et al23  2.57 270 332 80 77.1% 97.9% 87.4% 75.0% 
 3.00 259 335 91 74.0% 98.8% 86.2% 72.8% 
 3.50 249 337 101 71.1% 99.4% 85.1% 70.5% 
 4.00 238 338 112 68.0% 99.7% 83.6% 67.7% 
Maximum specificity 4.19 233 339 117 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).

or Create an Account

Close Modal
Close Modal