Table 3.
Risk factors for death during treatment versus post-treatment from multivariable analysis of the entire population (n=688 bedaquiline group; n=699 injectable group)
| Survived (n=1026) | Died during treatment (n=276) | Died after treatment (n=85) |
Adjusted risk difference, % (95% CI) |
||||
|---|---|---|---|---|---|---|---|
| Death at any timepoint versus survival | Death during treatment versus survival | Death post-treatment versus survival | |||||
| Regimen | |||||||
| Injectable group (n=699) | 500 (72%) | 159 (23%) | 40 (6%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Bedaquiline group (n=688) | 526 (76%) | 117 (17%) | 45 (7%) | −8% (−11 to −5) | −8% (−11 to −4) | 0% (−1 to 2) | |
| Age* | 36 (29 to 46) | 40 (33 to 51) | 35 (29 to 45) | 5% (3 to 6) | 4% (3 to 5) | 1% (0 to 2) | |
| Sex | |||||||
| Female (n=575) | 431 (75%) | 108 (19%) | 36 (6%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Male (n=812) | 595 (73%) | 168 (21%) | 49 (6%) | 0 (−3 to 4) | 0 (−3 to 3) | −1% (−2 to 0) | |
| AFB smear at baseline | |||||||
| Negative (n=629) | 474 (75%) | 118 (19%) | 37 (6%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Positive (n=545) | 410 (75%) | 104 (19%) | 31 (6%) | 2% (−2 to 5) | 2% (−2 to 5) | −2% (−3 to 0) | |
| Unknown (n=213) | 142 (67%) | 54 (25%) | 17 (8%) | 6% (1 to 11) | 4% (−1 to 8) | 2% (−2 to 6) | |
| Culture at baseline | |||||||
| Negative (n=254) | 204 (80%) | 42 (17%) | 8 (3%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Positive (n=676) | 507 (75%) | 119 (18%) | 50 (7%) | 7% (3 to 10) | 2% (−1 to 6) | 4% (3 to 6) | |
| Unknown (n=457) | 315 (69%) | 115 (25%) | 27 (6%) | 9% (3 to 15) | 4% (−1 to 9) | 2% (0 to 4) | |
| HIV status | |||||||
| Negative (n=420) | 326 (78%) | 74 (18%) | 20 (5%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Positive on antiretroviral therapy (n=918) | 674 (73%) | 182 (20%) | 62 (7%) | 6% (3 to 10) | 2% (−1 to 5) | 1% (0 to 2) | |
| Positive not on antiretroviral therapy (n=49) | 26 (53%) | 20 (41%) | 3 (6%) | 22% (6 to 38) | 18% (2 to 34) | 5% (−4 to 15) | |
| Previous treatment | |||||||
| Never treated (n=827) | 640 (77%) | 146 (18%) | 41 (5%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Treated with first-line drugs (n=560) | 386 (69%) | 130 (23%) | 44 (8%) | 8% (4 to 12) | 5% (2 to 8) | 2% (0 to 5) | |
| Isoniazid resistance | |||||||
| Susceptible (n=518) | 377 (73%) | 108 (21%) | 33 (6%) | 0 (ref) | 0 (ref) | 0 (ref) | |
| Resistant (n=553) | 430 (78%) | 86 (16%) | 37 (7%) | −4% (−8 to 1) | −3% (−6 to 1) | 0% (−2 to 1) | |
| Unknown (n=316) | 219 (69%) | 82 (26%) | 15 (5%) | 2% (−4 to 9) | 3% (−3 to 8) | −1% (−2 to 1) | |
Data are n (%), unless indicated otherwise. Comparisons were done only on patients who were alive during the analytical period (eg, analyses of post-treatment death versus survival were only done on patients who were alive at their treatment end date). Percentages are calculated by row, not by column; percentages might not sum up to 100% due to rounding. Model is adjusted for all factors listed in this table, including the province in South Africa where the patient initiated treatment. AFB=acid fast bacilli.
In this row, data are median (IQR) or risk difference (95% CI) per 10-year increase in age.