Comparative performance of stool preprocessing methods for Xpert MTB/RIF Ultra: optimizing diagnostic yield in elderly populations

ABSTRACT Diagnostic delays in elderly patients with pulmonary tuberculosis (PTB) frequently arise from difficulty obtaining adequate sputum specimens, yet evidence supporting stool-based Xpert MTB/RIF Ultra (Xpert Ultra) as a noninvasive alternative in adults remains limited. This prospective multicenter study conducted in Taiwan enrolled 483 participants, including 398 aged ≥65 years, of whom 167 had PTB, and 316 did not. Three stool preprocessing approaches were compared: homogenization and centrifugation (Stool-I), bead beating and chemical lysis (Stool-II), and a centrifuge-free protocol (Stool-III), with diagnostic performance assessed against a composite reference standard (CRS) and respiratory mycobacterial culture. Sputum Xpert Ultra achieved an overall sensitivity of 85.6% and specificity of 98.4%, while respiratory smear and culture were positive in 66.5% and 86.8% of PTB cases, respectively. Among stool preprocessing methods, Stool-I demonstrated the highest sensitivity, significantly outperforming Stool-III (73.1% vs 62.3%; P < 0.0001), while all three methods maintained high specificity of ≥98.4%. In subgroup analyses, Stool-I reached 90.7% sensitivity in patients with radiographic cavitation, and 68.6% in those unable to produce sputum. Notably, stool Xpert Ultra identified 29.2% of PTB cases that had initially tested negative on respiratory Xpert Ultra, highlighting its value as a complementary diagnostic tool. In conclusion, these findings indicate that stool-based Xpert Ultra is a highly specific, noninvasive adjunct for PTB diagnosis in elderly and sputum-scarce patients. With optimized preprocessing and context-specific implementation, this approach has the potential to improve diagnostic yield, reduce the need for invasive procedures, and shorten diagnostic delays across diverse healthcare settings. IMPORTANCE Tuberculosis (TB) remains a leading infectious cause of death worldwide, and diagnosis in elderly patients is often limited by inadequate sputum production and the need for invasive respiratory sampling. This study demonstrates that stool, a simple and noninvasive specimen, can be used with Xpert MTB/RIF Ultra to reliably detect Mycobacterium tuberculosis in this vulnerable population. Among the preprocessing methods evaluated, the centrifugation-based approach achieved the highest diagnostic sensitivity while maintaining excellent specificity. Notably, stool testing identified nearly 30% of tuberculosis cases that were initially negative on respiratory specimen testing, supporting its role as a complementary diagnostic strategy. These findings highlight a practical and patient-friendly approach that may reduce diagnostic delays, limit the need for invasive procedures, and improve TB care in an aging population that is particularly vulnerable to delayed diagnosis and adverse outcomes.