Objective: To evaluate the utility of the tuberculin skin test (TST) in detecting latent and active tuberculosis (TB) among human immunodeficiency virus (HIV) infected patients in South India.
Design: TSTs and CD4 counts were collected from 631 HIV-infected individuals without active TB and 209 antiretroviral and anti-tuberculosis treatment-naïve HIV-infected patients with TB. We calculated the proportion of TST-positive individuals, as well as the sensitivity, specificity, positive predictive value (PPV) and negative predictive value of TST in the diagnosis of TB.
Results: Among subjects without active TB, 28% with a CD4 count <100 cells/microl vs. 43% of the total cohort had a TST >5 mm (P = 0.14), while the proportions with induration >10 mm were 14% vs. 36%, respectively (P < 0.01). Among those with active TB, using a 5 mm cut-off, the sensitivity was 42% for those with CD4 counts <200 cells/mul compared to 70% for those with CD4 counts >or=200 cells/microl (P < 0.001). The PPV for detecting active TB was 29%.
Conclusions: TST is a poor predictor of both latent and active TB in HIV-infected individuals in TB endemic countries. Programmes offering treatment for latent TB should consider including all HIV-positive patients regardless of TST status, or use other indicators, such as CD4 count.