Table 1.
Characteristics of 774 participants.
| Characteristics | Overall | Bacteriological Positive * | Bacteriological Negative * | p value |
|---|---|---|---|---|
| N | 774 | 258 | 516 | |
| Gender = M/F (%) | 396/378 (51/49) | 140/118 (54/46) | 256/260 (50/50) | 0.25 |
| Age (median [IQR]) | 48.3 (18) | 44.7 (17) | 50.1 (19) | < 0.01 |
| Age Group (%) | < 0.01 | |||
| 15 to < 35 years | 219 (28) | 84 (33) | 135 (26) | |
| 35 to < 55 years | 249 (32) | 103 (40) | 146 (28) | |
| 55 years + | 306 (40) | 71 (28) | 235 (46) | |
| Prior TB History = Yes (%) | 197 (26) | 91 (35) | 106 (21) | < 0.01 |
| Currently receiving treatment for TB (%) | 28 (4) | 23 (9) | 5 (1) | < 0.01 |
| Previously treatment for TB (%) | 184 (24) | 79 (31) | 105 (20) | < 0.01 |
| HIV Status | ||||
| HIV+ (%) | 142 (18) | 65 (25) | 77 (15) | < 0.01 |
| HIV- (%) | 488 (63) | 150 (58) | 338 (66) | |
| Unknown (%) | 144 (19) | 43 (17) | 101 (20) | |
| Any Symptoms = Yes (%) | 405 (52) | 110 (43) | 295 (57) | < 0.01 |
| Cough > 2wks (%) | 148 (19) | 49 (19) | 99 (19) | 1 |
| Cough (%) | 258 (33) | 77 (30) | 181 (35) | 0.16 |
| Fever (%) | 113 (15) | 31 (12) | 82 (16) | 0.17 |
| Night sweats (%) | 178 (23) | 52 (20) | 126 (25) | 0.20 |
| Weight loss (%) | 141 (18) | 46 (18) | 95 (18) | 0.91 |
| Smoking = Yes (%) | 429 (56) | 123 (48) | 306 (60) | < 0.01 |
| Diabetes (%) | 0.08 | |||
| Yes | 58 (8) | 12 (5) | 46 (9) | |
| No | 703 (91) | 243 (94) | 460 (89) | |
| Unknown | 12 (2) | 3 (1) | 9 (2) | |
| Culture Result (%) | < 0.01 | |||
| Positive for MTB | 189 (24) | 189 (73) | 0 (0) | |
| Negative for MTB | 516 (67) | 57 (22) | 459 (89) | |
| Contaminated | 29 (4) | 8 (3) | 21 (4) | |
| Nontuberculous mycobacteria | 13 (2) | 0 (0) | 13 (3) | |
| Not done | 27 (4) | 4 (2) | 23 (5) | |
| Xpert Result (%) | < 0.01 | |||
| Positive | 197 (26) | 197 (76) | 0 (0) | |
| Trace positive | 26 (3) | 21 (8) | 5 (1) | |
| Negative | 539 (70) | 38 (15) | 501 (97) | |
| Invalid | 3 (0.4) | 0 (0) | 3(0.6) | |
| Not done | 9 (1) | 2 (1) | 7 (1) | |
| AI Reading (median [IQR]) | ||||
| JF CXR-2 | 0.38 [0.03, 0.93] | 0.95 [0.82, 0.98] | 0.08 [0.01, 0.62] | < 0.01 |
| Lunit | 0.09 [0.01, 0.82] | 0.90 [0.70, 0.96] | 0.02 [0.01, 0.14] | < 0.01 |
| Nexus | 0.53 [0.15, 0.84] | 0.89 [0.78, 0.93] | 0.25 [0.10, 0.61] | < 0.01 |
| CAD4TB | 8 [1, 59] | 70 [27, 91] | 2 [1, 18] | < 0.01 |
| qXR | 0.30 [0.05, 0.88] | 0.92 [0.66, 0.97] | 0.09 [0.03, 0.43] | < 0.01 |
| InferRead | 0.35 [0.17, 0.71] | 0.74 [0.57, 0.84] | 0.24 [0.14, 0.41] | < 0.01 |
| Genki | 0.07 [0.00, 0.57] | 0.62 [0.35, 0.84] | 0.01 [0.00, 0.17] | < 0.01 |
| XrayAME | 0.08 [0.02, 0.44] | 0.45 [0.08, 0.90] | 0.04 [0.01, 0.17] | < 0.01 |
| ChestEye | 0.10 [0.04, 0.42] | 0.46 [0.19, 0.62] | 0.06 [0.03, 0.14] | < 0.01 |
| Xvision | 0.14 [0.08, 0.46] | 0.51 [0.27, 0.62] | 0.10 [0.07, 0.16] | < 0.01 |
| TiSepX-TB | 0.26 [0.14, 0.51] | 0.60 [0.32, 0.88] | 0.19 [0.13, 0.31] | < 0.01 |
| RADIFY | 0.47 [0.00, 0.60] | 0.54 [0.23, 0.84] | 0.43 [0.00, 0.59] | < 0.01 |
| Radiologists TB findings (%) | ||||
| UK Radiologists | < 0.001 | |||
| Abnormality consistent with TB | 323 (41.73) | 203 (78.68) | 120 (23.26) | |
| Abnormality not consistent with TB | 91 (11.76) | 22 (8.53) | 69 (13.37) | |
| Normal | 360 (46.51) | 33 (12.79) | 327 (63.37) | |
| Nigerian Radiologists | < 0.001 | |||
| Abnormality consistent with TB | 324 (41.86) | 199 (77.13) | 125 (24.22) | |
| Abnormality not consistent with TB | 101 (13.05) | 25 (9.69) | 76 (14.73%) | |
| Normal | 349 (45.09) | 34 (13.18) | 315 (61.05) | |
| American Radiologists | < 0.001 | |||
| Abnormality consistent with TB | 258 (33.33) | 177 (68.60) | 81 (15.70) | |
| Abnormality not consistent with TB | 252 (32.56) | 61 (23.64) | 191 (37.02) | |
| Normal | 264 (34.11) | 20 (7.75) | 244 (47.29) | |
| Indian Radiologists | < 0.001 | |||
| Abnormality consistent with TB | 259 (33.46) | 173 (67.05) | 86 (16.67) | |
| Abnormality not consistent with TB | 90 (11.63) | 40 (15.50) | 50 (9.69) | |
| Normal | 425 (54.91) | 45 (17.44) | 380 (73.64) | |
* “bacteriological positive” and “bacteriological negative” denote the bacteriological status determined by the composite microbiological reference standard in this study, where anyone with Mycobacterium tuberculosis detected by either Xpert Ultra and/or culture is referred to as bacteriological positive.