408 www.thelancet.com Vol 379 February 4, 2012 a full 14 days. Sequential therapy was less eff ective than has been reported in other trials, but a careful appraisal of the data shows that TT-14 was not remarkably better than ST: the number need to treat in favour of TT14 was 18, with a wide 95% CI (9–175) suggesting uncertainty in the true estimate of eff ect. In conclusion, caution should be exercised in the interpretation of these interesting but heterogeneous data. Carefully designed trials with document ation of resistance in individual patients and with a suffi ciently large sample size from each country are necessary before any regimen can be determined as optimal in an individual Latin American country.