Screening for active tuberculosis (TB) among individuals with latent tuberculosis infection (LTBI) is important for the initiation and evaluation of TB preventive treatment. The performances of different tools and their combinations had rarely been studied in community-level screening among individuals with LTBI in China. This study aimed to explore appropriate algorithms for screening for active TB among individuals with LTBI in rural China. Three sputum samples were collected from each participant for smear microscopy, culture, and an Xpert MTB/RIF assay. Chest digital radiography and TB symptoms were investigated as well. The performances of different testing algorithms were compared with that of sputum culture as the gold standard. Overall, 1,564 study participants with LTBI were investigated, with a final diagnosis of 20 TB cases by sputum culture. Compared with other tests, the Xpert MTB/RIF assay detected 80.00% (95% confidence interval [CI], 58.40% to 91.93%) of culture-positive cases, with the highest sensitivity. When tests were combined using "or," "and," or "step" algorithms, the highest sensitivity reached 90.00% (95% CI, 69.90% to 97.21%) for the combination of the Xpert MTB/RIF assay and chest radiography, but the positive predictive value (PPV) decreased to 22.22% (95% CI, 14.54% to 32.41%). The Xpert MTB/RIF assay alone showed the best agreement with sputum culture, with a kappa value of 0.840. Pathogen molecular detection alone showed good performance compared to the other algorithms, for ruling out active TB in general LTBI, but the high cost might be a challenge for scaling it up. Identifying those with a high risk for progression to TB more precisely and establishing a cost-effective screening algorithm deserve further exploration. IMPORTANCE Enhancing community-wide active case screening in target LTBI populations is important for achieving the early treatment of active TB, and ruling active TB out is a prerequisite for initiating preventive treatment. The current study evaluated the performances of multiple tests and their combinations in screening for active TB among individuals with LTBI at the community level. Compared with the classical "TB symptoms and chest radiography" algorithm, the application of Xpert MTB/RIF improved the sensitivity from 45% to 80%. When the Xpert MTB/RIF assay was combined with chest radiography, the sensitivity was further improved to 90.00%, which achieved the World Health Organization (WHO) target product profiles. However, the algorithm requires caution as the PPV decreased from 88.89% for Xpert MTB/RIF alone to 22.22% for the combination. Xpert MTB/RIF alone offered remarkable sensitivity without compromising the PPV but would have major resource implications. Thus, identifying target populations for LTBI treatment more precisely and developing cost-effective and high-throughput screening tools and algorithms deserve further efforts.
Keywords: Xpert MTB/RIF assay; active case screening; algorithms; latent tuberculosis infection; tuberculosis.