Setting: Reliable clinical algorithms that screen for active tuberculosis (TB) in human immunodeficiency virus (HIV) infected people initiating or receiving antiretroviral treatment (ART) in sub-Saharan Africa could reduce the need for diagnostic procedures.
Methods: We estimated the utility of six TB-related signs and symptoms, alone or in combination, compared with the Uganda Ministry of Health diagnostic guidelines for participants with prevalent (baseline), early ART (< or = 3 months on ART) and incident TB (>3 months on ART).
Results: Of 1995 participants screened for ART eligibility, 71 (3.6%) had prevalent TB. The presence of any one of the following: cough > or = 3 weeks, fever > or = 4 weeks, lymphadenopathy or baseline body mass index < or = 18 kg/m(2) had a sensitivity of 99% (95%CI 96-100), a specificity of 66% (95%CI 64-68) and a negative predictive value (NPV) of 100% (95%CI 99-100) for predicting active TB. During ART follow-up, TB incidence was 2.4 (95%CI 1.6-3.4)/100 person-years. The presence of cough > or = 3 weeks or general weakness was 100% sensitive (95%CI 99-100), 66% specific (95%CI 59-74) and had an NPV of 100% (95%CI 99-100).
Conclusion: Use of a simple TB screening algorithm can accurately identify, in a resource-poor African setting, HIV-infected individuals who require further procedures to diagnose active TB.