Skip to main content
. 2020 Apr;8(4):407–419. doi: 10.1016/S2213-2600(19)30469-2

Table 4.

Sensitivity and specificity of a diagnostic algorithm combining the sputum Ultra test with blood transcriptional signature analysis

Ultra test alone Addition of signature to:
All Ultra-positive individuals Ultra trace-positive individuals Ultra-positive individuals with previous tuberculosis
Sensitivity
BATF2 82% (69–90) 76% (63–86) 82% (69–90) 80% (67–89)
Kaforou25 .. 66% (52–78); p=0·013 82% (69–90) 74% (60–84)
Roe3 .. 78% (65–87) 82% (69–90) 80% (67–89)
Sweeney3 .. 76% (63–86) 80% (67–89) 78% (65–87)
Specificity
BATF2 90% (83–95) 99% (95–100); p=0·0077 95% (89–98) 96% (91–98); p=0·041
Kaforou25 .. 100% (96–100); p=0·0044 96% (91–98); p=0·041 97% (92–99); p=0·023
Roe3 .. 99% (95–100); p=0·0077 95% (89–98) 96% (91–98); p=0·041
Sweeney3 .. 100% (96–100); p=0·0044 96% (91–98); p=0·041 97% (92–99); p=0·023

Data are % (95% CI). Only significant p values are shown. Sensitivity and specificity were calculated for 154 patients with Ultra results, with or without reclassification of selected Ultra-positive tests by transcriptional signatures. p value of comparison with Ultra alone using McNemar's test.