Sensitivity and specificity using alternate cutoff values.

Identification of prevalent TST+, TST converters, and TST- using i) TB Ag − Nil cutoffs, ii) combinations of TB Ag-Nil and CXCL10 Stimulated levels or CXCL10 Stimulated/Unstimulated ratio and iii) Menzies definition (excludes those with difference in TST<6 mm), and TST- using TB Ag − Nil cutoff OF 0.25 IU/ml. Participants come from 74 (IGRA testing)/46 (IGRA and CXCL10 testing) households. Estimation uses a GEE approach with an independent working correlation matrix to adjust for clustering in households. Models based on the sample that included CXCL10 cutoffs were weighted to reflect the larger study cohort.
Pr(T+|Prevalent TST+) is the percentage test-positive among those who are were TST+ at week 1–2.
Pr(T+|TST Converter) is the percentage test-positive among those who are TST-converters (TST- at week 1–2 and TST+ at week 8–12).
Pr(T+|TST Menzies Converter) is the percentage test-positive among those who are TST-converters (TST- at week 1–2 and TST+ at week 8–12 and change of at least 6 mm).
Pr(T-| TST-) is the percentage test-negative among those who are TST-negative at weeks 1–2 and 8–12.