Diagnostic Performance and Test Savings of Pooled Testing for Molecular Detection of Tuberculosis: A Diagnostic Accuracy Study across Seven Countries

Read full text
Authors: Tushar Garg, Melissa Sander, Rachel L. Byrne, S. M. Mazidur Rahman, anon Investigators, See PDF
Cited by: 2

Abstract

Background: In 2024, only 54% of people diagnosed with tuberculosis received an initial WHO-recommended rapid molecular diagnostic test. Pooled testing could expand access to molecular detection using existing laboratory capacity. We evaluated the performance of pooled TB testing in a multicentre diagnostic accuracy study.<br><br>Methods: We recruited individuals with presumptive tuberculosis through facility-based case finding (district hospitals and primary health centres) and community-based case finding. Sputum was tested individually on Xpert MTB/RIF Ultra (Xpert Ultra) and in pools of two, three or four specimens on Xpert Ultra. Culture was used as the microbiological reference standard. We estimated the sensitivity and specificity of pooled testing versus culture, positive percent agreement (PPA) versus individual Xpert Ultra, and cartridge savings.<br><br>Findings: Results were available for 14,040 (95·2%) participants from 3,666 pools, including 3,195 (87·2%) pools of four. Using culture as the reference standard, sensitivity and specificity of pooled testing versus individual Xpert Ultra were 84·9% (95% CI 82·5–87·0) versus 88·0% (85·9–90·0) and 98·3% (98·0–98·5) versus 97·7% (97·4–97·9), respectively. PPA was 90·1% (95% CI 88·4–91·7), remaining high for high, medium, and low individual Xpert Ultra semiquantitative grades (98·1% [96·1–99·2], 98·3% [95·8–99·5], and 96·5% [93·8–98·2]), but lower for very low (86·6% [79·9–91·7]) and trace (55·4% [47·9–62·7]). Pooled testing used 7,028 cartridges versus 14,362 for individual testing, a 51·1% reduction.<br><br>Interpretation: Pooled testing with Xpert Ultra had slightly higher specificity and a 3·2% lower sensitivity compared to individual Xpert Ultra, while halving cartridge use. Scale-up of this testing approach may facilitate expanded access to molecular tuberculosis diagnosis without the need for additional infrastructure.

Fields of Study & Subjects

Tuberculosis Research and Epidemiology SARS-CoV-2 detection and testing Biosensors and Analytical Detection

Publication details

Journal / source
SSRN Electronic Journal
Publisher
Elsevier BV
Publication date
1 January 2026
Document type
Preprint