Setting: Tuberculosis (TB) diagnosis is challenging in patients who are unable to produce sputum. The string test, a method for retrieving enteropathogens, is a potential alternative diagnostic tool.
Objectives: To compare the TB detection yield and tolerability of the string test and that of sputum induction in adults with presumed TB in Uganda.
Design: Cross-sectional study. String test and sputum induction were performed consecutively in patients unable to produce sputum. The string was removed after a 2-h intra-gastric downtime. Sputum induction used nebulised 5% saline for 20 min. Light-emitting diode fluorescence microscopy, Löwenstein-Jensen and MGIT culture were performed on all specimens, and the Xpert(®) MTB/RIF assay on a subset. Tolerability questionnaires were administered.
Results: Of 210 patients included in the study, 59% were human immunodeficiency virus (HIV) positive and 50 (23.8%) were diagnosed with TB. Of these, 48 (96.0%) were detected with the string test and 46 (92.0%) with sputum induction. In patients with specimens collected using both methods for paired analysis, the yield of microscopy detection with the string test was 13.8% (26/188) vs. 13.3% (25/188) with sputum induction (P = 1.0). The yield increased to 22.9% (42/183) using culture for string test vs. 24.6% (45/183) for sputum induction (P = 0.37). Xpert detected TB in 15/96 (15.6%) patients with the string test vs. 17/96 (17.7%) with sputum induction (P = 0.62). Tolerability was comparable.
Conclusion: The string test was well tolerated and provided similar yields to sputum induction, offering a viable alternative in resource-limited settings with minimal risk of transmission.