Sputum bacteriology conversion and treatment outcome of patients with multidrug-resistant tuberculosis

Multidrug-resistant tuberculosis (MDR-TB) requires long-term treatment, has a high fatality rate, and constitutes a global threat. Earlier detection of treatment failure is required to predict therapeutic efficacy.

We enrolled MDR-TB patients consecutively from January 2011 through December 2012 in Lianyungang, China. Sputum smear microscopy tests and sputum cultures were performed once a month for the first 6 months following initiation of antituberculosis treatment and once every 2 months thereafter until the end of therapy. The sensitivity, specificity and area under the receiver operating characteristic curve (AUC) were used with a 95% CI to estimate the role of sputum bacteriology conversion in predicting treatment outcomes.

Among the 92 MDR-TB patients enrolled in this study, 40.2% had poor treatment outcomes. The median initial sputum bacteriology conversion time was 1 month. Patients having 2-month sputum smear conversions (adjusted odds ratio [OR]: 7.19, 95% CI: 2.60-19.84) or culture conversions (adjusted OR: 2.88, 95% CI: 1.11-7.45) were more likely to experience good outcomes. The sensitivity and specificity obtained when using two-month sputum smear conversions to predict treatment outcomes were 67.6% (95% CI: 50.2-82.0) and 76.4% (95% CI: 63.0-86.8), respectively. The sensitivity and specificity obtained when using 2-month culture conversions to predict treatment outcomes were 48.6% (95% CI: 32.0-65.6) and 74.5% (95% CI: 61.0-85.3), respectively. The AUC for two-month smear conversions was 0.72 (95% CI: 0.62-0.81), significantly higher than that obtained for 2-month culture conversions (0.62, 95% CI: 0.52-0.72) (χ² = 4.18, P = 0.041).

The prognoses of MDR-TB patients displaying persistent sputum positivity were inferior to those for whom sputum bacteriology conversion was observed. Thus, sputum smear conversion results obtained 2 months after treatment initiation may provide a potential means for predicting MDR-TB treatment outcomes.