Table 2.
Regression analyses examining baseline factors associated with progression to incident TB disease within 15 months in people living with HIV.
| No incident TB, n (%) or median (IQR) N = 808 | Incident TB, n (%) or median (IQR) N = 21 | Univariable Cox proportional hazards |
Multivariable Cox LASSO regression with selective inference |
|||
|---|---|---|---|---|---|---|
| HR (95%CI) | p value | aHR (95%CI) | p value | |||
| Sex | ||||||
| Female | 590 (73·0) | 15 (71·4) | Reference | – | – | |
| Male | 218 (27·0) | 6 (28·6) | 1·11 (0·43, 2·85) | 0·83 | ||
| Age | 35 (29–42) | 33 (30–40) | 0·98 (0·93, 1·03) | 0·41 | 0·67 (0·44, 8·16) | 0·51 |
| Ethnicity | ||||||
| Black African | 689 (85·3) | 11 (52·4) | Reference | Reference | ||
| Mixed Ancestry | 119 (14·7) | 10 (47·6) | 5·50 (2·34, 13·00) | <0·001 | 1·49 (1·30, >1000) | 0·005 |
| Site | 0·037‡ | |||||
| Durban, KwaZulu-Natal | 281 (34·8) | 6 (28·6) | Reference | – | – | |
| Klerksdorp, North West | 158 (19·6) | 1 (4·8) | 0·33 (0·04, 2·75) | 0·31 | 0·52 (0·00, 2·20) † | 0·061 |
| Rustenburg, North West | 154 (19·1) | 2 (9·5) | 0·68 (0·14, 3·39) | 0·64 | – | – |
| Ravensmead, Western Cape | 84 (10·4) | 5 (23·8) | 3·36 (1·03, 11·00) | 0·045 | – | – |
| Worcester, Western Cape | 131 (16·2) | 7 (33·3) | 2·42 (0·81, 7·20) | 0·11 | 0·99 (0·00, 1·10) † | 0·97 |
| Highest level of schooling | ||||||
| Primary school or lower | 95 (11·8) | 5 (23·8) | Reference | Reference | ||
| Secondary school or higher | 713 (88·2) | 16 (76·2) | 0·44 (0·16, 1·21) | 0·11 | 0·87 (0·49, 62·58) | 0·78 |
| Employment | ||||||
| Employed | 130 (16·1) | 1 (4·8) | Reference | Reference | ||
| Unemployed | 678 (83·9) | 20 (95·2) | 3·59 (0·48, 26·80) | 0·21 | 1·80 (0·53, 3·14) | 0·18 |
| Occupants per household | 4 (3–6) | 4 (3–8) | 1·09 (0·97, 1·23) | 0·15 | 1·15 (0·00, 1·67) | 0·85 |
| Smoking history | ||||||
| No | 501 (62·0) | 9 (42·9) | Reference | – | – | |
| Yes | 307 (38·0) | 12 (57·1) | 2·23 (0·94, 5·30) | 0·069 | ||
| Prior TB | ||||||
| No | 614 (76·0) | 12 (57·1) | Reference | Reference | ||
| Yes | 194 (24·0) | 9 (42·9) | 2·24 (0·94, 5·31) | 0·068 | 1·61 (0·73, 5·15) | 0·093 |
| TB household contacts | ||||||
| No | 654 (80·9) | 16 (76·2) | Reference | – | – | |
| Yes | 154 (19·1) | 5 (23·8) | 1·30 (0·48, 3·56) | 0·60 | ||
| Isoniazid preventive therapy during study conduct | ||||||
| Did not receive therapy | 409 (50·6) | 13 (61·9) | Reference | Reference | ||
| Received therapy | 399 (49·4) | 8 (38·1) | 0·56 (0·23, 1·36) | 0·20 | 0·67 (0·00, 24·08) | 0·13 |
| Antiretroviral therapy (ART) during study conduct | 0·065‡ | |||||
| ART experienced | 635 (78·6) | 12 (57·1) | Reference | – | – | |
| ART naïve | 42 (5·2) | 1 (4·8) | 1·84 (0·24, 14·20) | 0·56 | – | – |
| ART started after enrolment | 131 (16·2) | 8 (38·1) | 3·08 (1·26, 7·52) | 0·014 | 1·48 (1·01, 929·93) † | 0·023 |
| Body-mass index (kg/m2) | 24·5 (20·7–31·5) | 19·9 (18·5–24·9) | 0·89 (0·81, 0·97) | 0·007 | 0·56 (0·39, 142·36) | 0·68 |
| CD4-positive cell count (cells/mm3) | 533·5 (360·8–734·2) | 291·0 (145·0–573·0) | 0·86 (0·78, 0·95) * | 0·003 | 0·45 (0·29, 8·85) * | 0·37 |
| HIV plasma viral load | ||||||
| <100 copies/mL | 175 (46·1) | 3 (20·0) | Reference | # | # | |
| ≥100 copies/mL | 205 (53·9) | 12 (80·0) | 3·69 (1·04, 13·10) | 0·043 | ||
| Not tested (missing) | 428 | 6 | – | – | ||
| Season at enrolment | 0·10‡ | |||||
| Spring | 190 (23·5) | 6 (28·6) | Reference | – | – | |
| Summer | 210 (26·0) | 5 (23·8) | 0·70 (0·21, 2·30) | 0·56 | – | – |
| Autumn | 187 (23·1) | 1 (4·8) | 0·16 (0·02, 1·34) | 0·091 | 0·45 (0·01, 1·24) † | 0·048 |
| Winter | 221 (27·4) | 9 (42·9) | 1·18 (0·42, 3·30) | 0·76 | – | – |
| TB symptoms at enrolment | ||||||
| No | 764 (94·6) | 20 (95·2) | Reference | – | – | |
| Yes | 44 (5·4) | 1 (4·8) | 1·01 (0·14, 7·50) | >0·99 | ||
| IGRA status | ||||||
| Negative | 446 (55·2) | 7 (33·3) | Reference | Reference | ||
| Positive | 362 (44·8) | 14 (66·7) | 2·43 (0·98, 6·01) | 0·056 | 1·48 (0·60, 10·69) | 0·13 |
IQR, interquartile range. CI, confidence interval. HR, hazard ratio. aHR, adjusted HR. IGRA, interferon-γ release assay. TB, tuberculosis. *Per 50 cells/mm3. #Excluded from LASSO analysis. †Reference group includes all other levels of variable. ‡Likelihood-ratio test for significance of categorical variable.