Table 2.
Association between common Mycobacterium tuberculosis genotypes and drug-resistant phenotype adjusted for relevant variables
| Independent variable | MDR | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Simple logistic regression* | Multiple logistic regression† | MLR with interaction‡ | |||||||
| OR | CI | P value | OR | CI | P value | OR | CI | P value | |
| SIT 1 (Beijing) | 0.69 | 0.24–2.0 | 0.5 | ||||||
| SIT 33 (LAM3) | 0.11 | 0.01–0.80 | 0.03 | ||||||
| SIT 42 (LAM9) | 14 | 7.30–25 | < 0.001 | 12 | 6.3–22 | < 0.001 | 2.6 | 0.83–8.2 | 0.1 |
| SIT 50 (Harlem3) | 0.59 | 0.29–1.2 | 0.144 | ||||||
| SIT 53 (T1) | 3.4 | 2.0–5.7 | < 0.001 | 3.8 | 2.2–6.3 | < 0.001 | |||
| HIV status | 5.0 | 3.4–7.4 | < 0.001 | 3.6 | 2.3–5.5 | < 0.001 | 3.1 | 2.0–4.8 | < 0.001 |
| Age | 0.99 | 0.97–1.0 | 0.2 | ||||||
| Sex | 1.5 | 1.0–2.3 | 0.045 | ||||||
| HIV/SIT 42 interaction | 8.5 | 1.9–38 | 0.006 | ||||||
| Pseudo R2 | 0.20 | 0.18 | |||||||
| PZA resistance | |||||||||
| Simple logistic regression* | MLR† | ||||||||
| OR | CI | P value | OR | CI | P value | ||||
| SIT 1 (Beijing) | 0.36 | 0.05–2.8 | 0.331 | ||||||
| SIT 33 (LAM3) | NA | ||||||||
| SIT 42 (LAM9) | 15 | 7.9–30 | < 0.001 | 12 | 5.9–23 | < 0.001 | |||
| SIT 50 (Harlem3) | 0.89 | 0.38–2.1 | 0.8 | ||||||
| SIT 53 (T1) | 4.9 | 2.7–9.1 | < 0.001 | 5.0 | 2.7–9.3 | < 0.001 | |||
| HIV status | 6.7 | 4.1–11 | < 0.001 | 4.4 | 2.6–7.4 | < 0.001 | |||
| Age | 0.99 | 0.97–1.0 | 0.3 | ||||||
| Sex | 0.94 | 0.60–1.5 | 0.8 | ||||||
CI = confidence interval; HIV = human immunodeficiency virus; LAM = Latino American-Mediterranean TB clade; MDR = multidrug resistant; MLR = multiple logistic regression; OR = odds ratio; PZA = pyrazinamide; NA = not applicable due to model omission, predicts failure perfectly; SIT = spoligotype international type.
Simple logistic regression using MDR or PZA resistance as dependent variable. An indicator variable for SIT genotype compares each of these five most common SITs against all SITs other than these five.
Multiple logistic regression using MDR or PZA resistance as dependent variable and SIT 42 and SIT 53 as independent variables of interest, adjusted for HIV status. Model chosen by backwards selection until all variables were statistically significant at level of (α) 5%.
Multiple logistic regression using MDR as dependent variable and SIT 42, HIV, and an interaction term for HIV multiplied by SIT 42.