Table 3. Test performance indices for various levels of toluidine blue vital staining and a comparison of this Kenyan study with studies from South Africa12 and Brazil11.
| Staining result | Sensitivity (95%CI) | Specificity (95%CI) | Positive predictive value (95%CI) | Negative predictive value (95%CI) | Area under the ROC curve (95%CI) |
|---|---|---|---|---|---|
| Any blue (dark, pale or mixed) | 92% (87% - 96%) | 31% (25% - 36%) | 41% (35% - 46%) | 88% (80% - 94%) | .61 (.58 - .65) |
| Dark royal blue or pale blue | 80% (73% - 87%) | 51% (45% - 57%) | 46% (40% - 52%) | 83% (77% - 89%) | .66 (.61 - .70) |
| Dark royal blue | 64% (56% - 72%) | 67% (61% - 72%) | 50% (43% - 58%) | 78% (72% - 83%) | .66 (.61 - .70) |
| Pale blue | 16% (11% - 23%) | 84% (79% - 88%) | 34% (23% - 47%) | 66% (61% - 71%) | .50 (.46 - .54) |
| Mixed patterna | 12% (7% - 19%) | 80% (74% - 84%) | 23% (14% - 35%) | 64% (58% - 69%) | .46 (.42 - .49) |
| Comparison with other studies | |||||
| Parameter | Kenya | S. Africa | Brazilb | ||
| Vital stain dye | toluidine blue 0.05% | methylene blue 1% | toluidine blue 1% | ||
| Number of participants | 419 | 75 | 47 | ||
| Gender, Female No. (%) | 277 (66) | 45 (60) | 18 (38) | ||
| Age, median (IQR), y | 37 (32-45) | 35c | 58c | ||
| OSSN prevalence by histopathology No. (%) | 142 (34) | 33 (44)d | 27 (57) | ||
| Sensitivity (95%CI) | 92% (87% - 96%) | 97% (85% - 100%) | 100% (87% - 100%) | ||
| Specificity (95%CI) | 31% (25% - 36%) | 50% (36% - 65%) | 50% (27% - 73%) | ||
| Positive predictive value (95%CI) | 41% (35% - 46%) | 60% (47% - 72%) | 73% (56% - 86%) | ||
| Negative predictive value (95%CI) | 88% (80% - 94%) | 95% (78% - 99%) | 100% (69% - 100%) | ||
Abbreviations: ROC, receiver operator characteristic; IQR, interquartile range
a
Some areas of the lesion were deep royal blue and others pale
b
intervals were calculated from the data presented in the paper
c
The interquartile range was not reported in S. Africa and Brazil.
d
In the South African study, 49 (65%) were on the OSSN spectrum, however, in the analysis that was presented CIN was combined in with benign lesions for the calculation of the test parameters.