Are we justified in treating for multidrug-resistant tuberculosis based on positive follow-up smear results?

SETTING: National Institute for Research in Tuberculosis, India. OBJECTIVE: To assess, among new culture-confirmed smear-positive pulmonary tuberculosis (TB) patients, the proportion of follow-up smear-positives that were culture-negative (S+C-) by month of follow-up examination, human immunodeficiency virus (HIV) status, pre-treatment drug susceptibility status and smear grading. DESIGN: We extracted follow-up smear (fluorescence microscopy) and culture (Lowenstein-Jensen) results of patients enrolled in clinical trials from January 2000 to August 2012 and treated with the WHO Category I regimen (2EHRZ(3)/4HR(3)). RESULTS: Of 520 patients, including 176 who were HIV-infected, respectively 199, 81, 47 and 43 were smear-positive at months 2, 4, 5 and 6; of these, respectively 138 (69%), 62 (75%), 32 (68%) and 27 (63%) were culture-negative. The S+C- phenomenon was more pronounced among '1+ positive' patients than in 2+ or 3+ positive patients and in 'pan-susceptible' patients than in those with any resistance, and did not vary by HIV status. CONCLUSION: Nearly two thirds of patients with follow-up smears positive at months 5 and 6 were culture-negative. Starting multidrug-resistant TB (MDR-TB) treatment empirically based on smear results, even in resource-limited settings, is incorrect and can have hazardous consequences. There is an urgent need to revisit the WHO recommendation concerning empirical MDR-TB treatment.