Quantitative responses of IGRA, expressed as international units of IFN-γ or as spot-forming units (SFUs), showed good correlation with the size of the TST skin induration, expressed in millimeters (Fig 1). Overall, however, both TS.TB (18.4%) and QFT-IT (15.1%) identified more positive results than did TST (10.9%; p < 0.001 vs TS.TB; p = 0.033 vs QFT-IT). Significantly fewer persons were identified as latently infected by any test in the HIV group, compared with the LTC and HM groups (TST, p = 0.042; TS.TB, p < 0.001; QFT-IT, p < 0.001). Considering those patients with at least one positive test result to be infected, LTBI prevalence was significantly lower in the HIV group (9.5%) than in the LTC group (35.8%; p < 0.001) and HM group (29.5%; p < 0.001). Concordance among tests was substantial, ranging from 80.6% (TST vs TS.TB in LTC) to 95.4% (TST vs QFT-IT in HIV) [Table 3]. However, agreement between tests across different groups was moderate, with the κ statistic ranging from 0.40 to 0.65. In the HIV group, agreement between TS.TB and either TST (κ = 0.16) or QFT-IT (κ = 0.19) was slight (Table 3). In the 12 BCG-vaccinated individuals, agreement among all three tests was complete (κ = 1.00). “Highly discordant” results, that is, those clearly negative with one IGRA and clearly positive with another, were found in all groups and represented 12.1% of the whole population (Table 4). Patients with a positive TST and a negative IGRA (without a history of BCG vaccination) were also represented in all groups (LTC patients: TS.TB, 5 patients; QFT-IT, 4 patients; HIV patients: TS.TB, 5 patients; QFT-IT, 3 patients; HM: TS.TB, 1 patient; QFT-IT, 0 patients). Overall, the fraction of tests with borderline results (ie, close to the cutoff for a positive test result) was higher for TS.TB (18 tests with 4 to 5 SFUs) than for QFT-IT (8 tests with 0.20 to 0.34 international units of IFN-γ) [Table 4].