Background: Despite the increasing prevalence of drug-resistant tuberculosis, most low- and middle-income countries use standardized regimens, without assessment of drug susceptibility.
Purpose: To perform a systematic review and meta-analysis of the effect of initial drug resistance and treatment regimen on tuberculosis treatment outcomes.
Data sources: PubMed, the Cochrane Central Database of Clinical Trials, and EMBASE were searched for studies published in English from 1965 to June 2007. Additional studies were identified from cited references.
Study selection: Randomized, controlled trials and cohort studies of standardized treatment of previously untreated patients with culture-confirmed pulmonary tuberculosis. Drug-susceptibility testing was done on pretreatment isolates from all patients and from patients with treatment failure or relapse.
Data extraction: Two authors reviewed the studies for methods, initial drug resistance, treatment regimens, and outcomes.
Data synthesis: Pooled cumulative incidences were computed with random-effects meta-analyses. Association between risk factors and outcomes were determined by using stratified analyses. The cumulative incidence of acquired drug resistance with initially pan-sensitive strains was 0.8% (95% CI, 0.5% to 1.0%) compared with 6% (CI, 4% to 8%) with initially single drug-resistant strains and 14% (CI, 9% to 20%) with initially polydrug-resistant strains. Failure and relapse were most strongly associated with initial drug resistance. Failure was also associated with shorter duration of rifampin therapy and nonuse of streptomycin, whereas the rate of relapse was higher with shorter duration of rifampin therapy and nonuse of pyrazinamide.
Limitations: Few studies included HIV-infected persons, and treatment outcomes were pooled despite considerable heterogeneity.
Conclusion: Treatment outcomes were substantially worse in the presence of initial drug resistance, which has important implications in resource-limited settings in which drug resistance is prevalent.