TABLE 2.
Proportional hazards models for association of hair drug concentrations and time to failure
Predictor | Patients n | Events n | HR (95% CI) | p-value |
Drug concentrations | ||||
Mean Z-score# | 47 | 11 | 0.60 (0.27–1.34) | 0.21 |
Isoniazid | 22 | 3 | 1.40 (0.39–5.0) | 0.61 |
Pyrazinamide | 44 | 11 | 0.58 (0.31–1.09) | 0.091 |
Ethambutol | 34 | 6 | 0.79 (0.34–1.8) | 0.58 |
Levofloxacin | 28 | 5 | 1.21 (0.52–2.8) | 0.66 |
Moxifloxacin | 19 | 6 | 0.95 (0.42–2.2) | 0.91 |
Bedaquiline | 25 | 4 | 0.42 (0.10–1.7) | 0.23 |
Clofazimine | 31 | 8 | 0.54 (0.27–1.08) | 0.081 |
Linezolid | 22 | 3 | 0.55 (0.10–3.1) | 0.50 |
Clinical and demographic risk factors | ||||
Age | 47 | 11 | 0.61 (0.28–1.31) | 0.20 |
HIV-positive | 47 | 11 | 3.1 (0.67–14.4) | 0.15 |
Mean anti-tuberculosis drug Z-score was calculated for each participant by averaging Z-scores (difference from the mean, divided by the standard deviation) for all hair drug concentrations for a given participant; this constituted the primary predictor. #: this was the primary predictor.