# Table 1 Predictors of an unfavourable outcome for all data (adjusted for treatment) and within each treatment arm. Prediction models fitted all factors significant in the “all data” model (p <0.05, likelihood ratio test) with the addition of TTP on MGIT which was significant in the “ethambutol arm” model. Factors not listed in this table were not significant in any model

Multivariable odds ratio (95 % CI) All data Control arm patients only Isoniazid arm patients only Ethambutol arm patients only
Control Reference
Isoniazid 1.81 (1.18, 2.78)
Ethambutol 2.88 (1.92, 4.33)
p <0.001
BMI per 1 kg/m2 0.92 (0.87, 0.98) 0.93 (0.82, 1.05) 0.95 (0.85, 1.05) 0.90 (0.83, 0.99)
p = 0.004 p = 0.193 p = 0.286 p = 0.027
History of smoking 1.63 (1.15, 2.31) 2.12 (0.99, 4.53) 2.34 (1.23, 4.46) 1.15 (0.69, 1.92)
p = 0.005 p = 0.046 p = 0.007 p = 0.592
HIV positive 2.93 (1.69, 5.08) 2.64 (0.90, 7.78) 1.95 (0.72, 5.28) 4.46 (1.79, 11.09)
p <0.001 p = 0.100 p = 0.210 p = 0.002
TTP on MGIT (per 0.1 log10(day)) 0.97 (0.90, 1.05) 1.17 (1.03, 1.32)a 1.02 (0.89, 1.17) 0.84 (0.74, 0.95)
p = 0.439 p = 0.024 p = 0.789 p = 0.003
Male gender 1.69 (1.13, 2.53) 1.10 (0.48, 2.55) 1.94 (0.94, 3.98) 2.09 (1.13, 3.85)
p = 0.009 p = 0.817 p = 0.060 p = 0.014
Cavities on X-ray 1.93 (1.22, 3.05) 1.62 (0.62, 4.19) 2.82 (1.15, 6.94) 1.75 (0.89, 3.41)
p = 0.003 p = 0.302 p = 0.013 p = 0.091
1. aThis association (p = 0.024) indicates a higher probability of an unfavourable outcome with higher TTP on MGIT, indicating a lower bacillary load which is biologically counter-intuitive. This is a modest odds ratio with a fairly wide confidence interval—similar results are seen in the univariable model. Due to the multiplicity in the number of tests done to evaluate baseline predictors this is therefore likely a chance finding