Yield and Efficiency of Intensified Tuberculosis Case-Finding Algorithms in 2 High-Risk HIV Subgroups in Uganda

Background: Tuberculosis (TB) risk varies among different HIV subgroups, potentially impacting intensified case finding (ICF) performance. We evaluated the performance of the current ICF algorithm [symptom screening, followed by Xpert MTB/RIF (Xpert) testing] in 2 HIV subgroups and evaluated whether ICF performance could be improved if TB screening was based on C-reactive protein (CRP) concentrations. Methods: We enrolled consecutive adults with CD4 counts <= 350 cells/mu L initiating antiretroviral therapy and performed symptom screening, CRP testing using a low-cost point-of-care (POC) assay, and collected sputum for Xpert testing. We compared the yield and efficiency of the current ICF algorithm to POC CRP-based ICF among patients new to HIV care and patients engaged in care. Results: Of 1794 patients, 126/1315 (10%) new patients and 21/479 (4%) engaged patients had Xpert-positive TB. The current ICF algorithm detected >= 98% of all TB cases in both subgroups but required >= 85% of all patients to undergo Xpert testing. POC CRP-based ICF halved the proportion of patients in both subgroups requiring Xpert testing relative to the current ICF algorithm and had lower yield among patients engaged in care [81% vs. 100%, difference -19% (95% confidence interval: -41 to 3)]. Among patients new to care, POC CRP-based ICF had similar yield as the current ICF algorithm [93% vs. 98%, difference -6% (95% confidence interval: -11 to 0)]. Conclusions: Among patients new to care, POC CRP-based screening can improve ICF efficiency without compromising ICF yield, whereas symptom-based screening may be necessary to maximize ICF yield among patients engaged in care.