Background:  The WHO algorithm for the diagnostic evaluation of TB in HIV-infected ambulatory patients is based on cough of 2 to 3 weeks duration, with acid-fast bacillus (AFB) smear as the only initial diagnostic test. Our objective was to assess the yield and cost of an intensive TB screening program compared to the WHO guidelines in HIV-infected patients starting ART in Durban, South Africa.

Methods:  As part of an intensive TB screening study, we prospectively enrolled adults

(≥18 years old) in pre-ART training at an HIV clinic in Durban from May 2007 until May 2008. Patients were enrolled regardless of TB signs or symptoms. Following a baseline symptom screen by a trained nurse, patients expectorated sputum spontaneously or with ultrasonic nebulization for AFB smear (ZN/Auramine), and culture (MGIT liquid and 7H11 solid medium). Nebulizer tubing was single-use to prevent contamination. Sensitivity and specificity of cough as a screening tool for TB diagnosis were calculated using 6-week TB culture results (liquid and/or solid) as the gold standard. Program costs (2007 US$) included clinic personnel, materials and cultures.

Results:  During the 1-year study, 1035 patients were enrolled and have complete culture results. Median CD4 count was 100/μL (range 48 to 154/μL). Because they were already on treatment for active TB, 211 (20%) were excluded from further analyses. Of the remaining 824 patients, 159 (19%) had a positive TB sputum culture, only 14 (9%) of whom had a positive AFB smear. Only half (73 of 159, 49%) of culture-positive patients reported cough at the time of sputum collection; nearly a quarter (35 of 159, 22%) did not report any typical TB symptoms. The sensitivity and specificity of cough as an indicator of active TB were 52% (95%CI 44 to 60%) and 63% (95%CI 59 to 66%). Cough-based screening would have cost ~$20,000 or ~$240/case for the 83 cases identified. Total costs for the intensive screening program were $47,200, with an incremental cost/case of $360 to identify all 159 TB cases in the cohort.

Conclusions:  Nearly 20% of patients starting ART in one site in Durban had undiagnosed, culture-positive active pulmonary TB. Neither cough nor AFB smear were sensitive screening tools for TB. Compared to screening strategies based on cough, intensive screening doubles the cases identified with only a modest increase in the cost per case identified. TB sputum cultures should be performed prior to ART initiation, regardless of symptoms, in areas of high HIV/TB prevalence.