Volume 19, Number 3—March 2013
Dispatch
Multidrug-Resistant Tuberculosis, Somalia, 2010–2011
Table
Drug-resistance pattern | Patients, % (95% CI) |
||
---|---|---|---|
New, n = 754 | Previously treated, n = 96 | All | |
Susceptible | 85.6 (81.5–89.7) | 46.2 (29.1–63.2) | 81.4 (77.7–85.2) |
Any resistance to | |||
Isoniazid | 10.9 (7.8–13.9) | 51.4 (35.8–67.0) | 15.2 (12.2–18.2) |
Rifampin | 8.7 (5.2–12.1) | 43.2 (25.9–60.6) | 12.3 (9.1–15.6) |
Total |
14.4 (10.3–18.5) |
53.8 (36.8–70.9) |
18.6 (14.8–22.3) |
Monoresistance to | |||
Isoniazid | 5.7 (4.1–7.4) | 10.6 (0.0–21.5) | 6.2 (4.5–7.9) |
Rifampin | 3.5 (1.1–6.0) | 2.4 (0.0–5.4) | 3.4 (1.0–5.8) |
Total |
9.2 (6.1–12.4) |
13.0 (1.4–24.5) |
9.6 (6.6–12.6) |
Multidrug resistance | 5.2 (2.8–7.5) | 40.8 (24.7–57.0) | 8.9 (6.5–11.4) |
*Prevalence estimates were obtained by using logistic regression (Stata’s svy: logit command, Stata Corp., College Station, TX, USA) on the binary treatment history variable; each new/retreatment case with a drug-susceptibility test result was weighted by the number of new/retreatment cases notified in its cluster in 2010 (the year in which the survey started), divided by the total number of new/retreated cases with a drug-susceptibility test result in its cluster. The estimation of odds ratios reported elsewhere in the text also included the expansion of categorical sex, age group, and zone variables (xi: logit), with clustering and CIs of variance. The findings were robust to multiple imputation of missing data (adding another 78 new and 18 retreatment cases to the sample), use of sampling weights based on 2007 notifications (the year in which cluster samples were calculated), and no use of sampling weights at all; prevalence estimates were equivalent to or slightly higher than those reported here.