Diagnosis and treatment of pulmonary tuberculosis in patients with HIV

Abstract

Timothy Holtz and colleagues1 conclude that mortality in seriously ill patients with HIV and suspected smear-negative tuberculosis could be significantly reduced by early empiric antitubercular treatment according to the 2007 WHO algorithm for seriously ill patients. Although we agree that the performance of the algorithm is an important research topic, we contest that this study used the algorithm. We are also unconvinced by the mortality benefits recorded in view of the sequential cohort study design, which is prone to bias.