Invalid screening algorithm [Correspondence]

The study by Were et al. from Uganda contains two critical flaws in design that invalidate its main conclusions. First, screening for HIV-associated tuberculosis is not valid unless chest X-ray and sputum culture are included for all patients. Patients with HIV-associated TB may deny symptoms but have an abnormal chest X-ray as the only evidence of active disease. Further, it has been shown that patients with HIV and active TB may have what we refer to as sub-clinical tuberculosis, i.e., sputum culture-positive active TB with no symptoms and a negative chest X-ray. Among our cohort of HIV-positive subjects in Tanzania with CD4 counts .200, we found that 10% of subjects had subclinical TB. Had the authors employed comprehensive screening of all patients with sputum culture and chest X-ray, it is likely they would have detected rates of prevalent TB in the 9-12% range rather than the 3.6% they reported.