Table 1.
Regression analyses examining factors associated with prevalent TB disease in people living with HIV.
| No prevalent TB, n (%) or median (IQR) N = 834 | Prevalent TB, n (%) or median (IQR) N = 17 | Univariable logistic regression |
Multivariable binomial LASSO regression with selective inference |
|||
|---|---|---|---|---|---|---|
| OR (95%CI) | p value | aOR (95%CI) | p value | |||
| Sex | ||||||
| Female | 608 (72·9) | 9 (52·9) | Reference | Reference | ||
| Male | 226 (27·1) | 8 (47·1) | 2·39 (0·89, 6·33) | 0·076 | 1·34 (0·08, 2·88) | 0·58 |
| Age | 35 (29–42) | 36 (33–42) | 1·02 (0·97, 1·08) | 0·39 | 1·29 (0·00, 2·17) | 0·77 |
| Ethnicity | ||||||
| Black African | 704 (84·4) | 12 (70·6) | Reference | – | – | |
| Mixed Ancestry | 130 (15·6) | 5 (29·4) | 2·26 (0·71, 6·19) | 0·13 | ||
| Site | <0·001‡ | |||||
| Durban, KwaZulu-Natal | 288 (34·5) | 7 (41·2) | Reference | 4·59 (1·49, 10·74) † | 0·006 | |
| Klerksdorp, North West | 160 (19·2) | 0 (0·0) | NA | NA | – | – |
| Rustenburg, North West | 158 (18·9) | 0 (0·0) | NA | NA | – | – |
| Ravensmead, Western Cape | 89 (10·7) | 0 (0·0) | NA | NA | – | – |
| Worcester, Western Cape | 139 (16·7) | 10 (58·8) | 2·96 (1·11, 8·31) | 0·031 | 5·08 (0·67, 10·63) † | 0·051 |
| Highest level of schooling | ||||||
| Primary school or lower | 100 (12·0) | 0 (0·0) | NA | Reference | ||
| Secondary school or higher | 734 (88·0) | 17 (100·0) | NA | 2·29 (0·54, 7·66) | 0·11 | |
| Employment | ||||||
| Employed | 131 (15·7) | 3 (17·6) | Reference | – | – | |
| Unemployed | 703 (84·3) | 14 (82·4) | 0·87 (0·28, 3·81) | 0·83 | ||
| Occupants per household | 4 (3–6) | 4 (1–6) | 0·87 (0·69, 1·04) | 0·17 | 0·83 (0·27, 26·81) | 0·65 |
| Smoking history | ||||||
| No | 513 (61·5) | 9 (52·9) | Reference | Reference | ||
| Yes | 321 (38·5) | 8 (47·1) | 1·42 (0·53, 3·75) | 0·47 | 0·60 (0·31, 17·78) | 0·52 |
| Prior TB | ||||||
| No | 630 (75·5) | 13 (76·5) | Reference | Reference | ||
| Yes | 204 (24·5) | 4 (23·5) | 0·95 (0·27, 2·72) | 0·93 | 0·75 (0·50, 78·17) | 0·73 |
| TB household contacts | ||||||
| No | 674 (80·8) | 17 (100·0) | NA | Reference | ||
| Yes | 160 (19·2) | 0 (0·0) | NA | 0·35 (0·13, 1·61) | 0·076 | |
| Isoniazid preventive therapy at enrolment | ||||||
| Not on therapy | 788 (94·5) | 17 (100·0) | NA | – | – | |
| On therapy | 46 (5·5) | 0 (0·0) | NA | |||
| Antiretroviral therapy at enrolment | 0·21‡ | |||||
| Naïve | 183 (21·9) | 7 (41·2) | Reference | 1·13 (0·03, 1·73) † | 0·77 | |
| <6 months | 112 (13·4) | 2 (11·8) | 0·47 (0·07, 1·97) | 0·35 | – | – |
| ≥6 months | 539 (64·6) | 8 (47·1) | 0·39 (0·14, 1·12) | 0·071 | – | – |
| Body-mass index (kg/m2) | 24·3 (20·6–31·3) | 21·3 (18·7–24·0) | 0·90 (0·81, 0·98) | 0·028 | 0·33 (0·10, 1·67) | 0·078 |
| CD4-positive cell count (cells/mm3) | 530·5 (354·2–731·0) | 393·0 (248·0–591·0) | 0·89 (0·80, 0·98) * | 0·028 | 0·62 (0·20, 1·71) * | 0·15 |
| HIV plasma viral load | ||||||
| <100 copies/mL | 179 (45·1) | 1 (9·1) | Reference | # | # | |
| ≥100 copies/mL | 218 (54·9) | 10 (90·9) | 8·21 (1·55, 151·41) | 0·046 | ||
| Not tested (missing) | 437 | 6 | NA | NA | ||
| Season at enrolment | 0·17‡ | |||||
| Spring | 196 (23·5) | 8 (47·1) | Reference | 1·41 (0·66, 196·86) † | 0·081 | |
| Summer | 218 (26·1) | 2 (11·8) | 0·22 (0·03, 0·91) | 0·061 | – | – |
| Autumn | 188 (22·5) | 3 (17·6) | 0·39 (0·08, 1·37) | 0·17 | – | – |
| Winter | 232 (27·8) | 4 (23·5) | 0·42 (0·11, 1·36) | 0·16 | 0·85 (0·58, >1000) † | 0·88 |
| TB symptoms at enrolment | ||||||
| No | 789 (94·6) | 15 (88·2) | Reference | Reference | ||
| Yes | 45 (5·4) | 2 (11·8) | 2·34 (0·36, 8·63) | 0·27 | 1·18 (0·04, 2·11) | 0·72 |
| IGRA status | ||||||
| Negative | 455 (54·6) | 5 (29·4) | Reference | Reference | ||
| Positive | 379 (45·4) | 12 (70·6) | 2·88 (1·06, 9·12) | 0·049 | 1·63 (0·62, 3·20) | 0·14 |
IQR, interquartile range. CI, confidence interval. OR, odds ratio. aOR, adjusted OR. IGRA, interferon-γ release assay. TB, tuberculosis. NA, not applicable. *Per 50 cells/mm3. #Excluded from LASSO analysis. †Reference group includes all other levels of variable. ‡Likelihood-ratio test for significance of categorical variable.