RESUMO
BACKGROUND: Clinical scores are promising case-finding tools for tuberculosis (TB) among HIV-infected patients. The Bandim TBscore has been shown to increase the diagnostic yield among patients with presumed TB in general, but has not previously been tested among newly diagnosed HIV patients at high risk of TB. METHODS: HIV-infected patients were included in this cross-sectional study. A pre-post-intervention study design was used to assess the outcome of a change in practice, i.e. the application of a clinical score (TBscore) consisting of 13 signs and symptoms to assess the need for further TB diagnostics. Patients with a TBscore ≥2 were evaluated using smear microscopy and Xpert MTB/RIF. A TB diagnosis was made based on microbiology or clinical evaluation. The sensitivity and specificity of the TBscore were compared with those of World Health Organization symptoms. RESULTS: The TB prevalence among newly enrolled HIV-infected patients during the study period was 13.4% (22/164). Using the TBscore and a diagnostic algorithm, it was possible to increase the proportion of patients started on TB treatment from 2.7% (10/367) the year before the study to 10.4% (17/164) during the study period. Five patients diagnosed with TB were not started on TB treatment as they were lost to follow-up or died. With a cut-off value of 2, the TBscore had a sensitivity, specificity, positive predictive value, and negative predictive value of 95.5% (21/22), 36.9% (41/111), 23.1% (22/118), and 97.6% (41/42), respectively. CONCLUSION: The TBscore is useful for standardized TB screening among HIV-infected individuals and may be a valuable tool to prioritize patients at high risk of TB.