Diagnostic accuracy of GeneXpert MTB/RIF and TB-LAMP for pulmonary tuberculosis in Bayelsa State, Nigeria

Tuberculosis (TB) remains the leading cause of death from a single infectious agent, and accurate diagnosis is a bottleneck in resource-limited settings. Xpert MTB/RIF and loop-mediated isothermal amplification (TB-LAMP) are WHO-endorsed, but head-to-head comparisons against culture are scarce, none from Nigeria’s Niger Delta. To compare diagnostic accuracy and concordance of Xpert MTB/RIF and TB-LAMP against Löwenstein–Jensen (LJ) culture among presumptive pulmonary TB cases in Bayelsa State, Nigeria. In this cross-sectional study, 392 of 4,312 screened presumptive cases (aged 6–80 years; cough plus ≥ 1 constitutional symptom) were consecutively enrolled from 40 DOT sites (January–June 2025). One sputum specimen was tested in parallel by Xpert MTB/RIF, TB-LAMP, and LJ culture; isolates were speciated by SD Bioline MPT64, with culture readers blinded to index results. Sensitivity, specificity, predictive values, likelihood ratios, κ, and McNemar’s exact test were calculated (Wilson 95% CI). Reporting followed STARD 2015. Culture identified M. tuberculosis complex in 41/392 specimens (10.5%). Xpert MTB/RIF achieved 97.6% sensitivity (95% CI 87.4–99.6) and 98.9% specificity (97.1–99.6); TB-LAMP, 92.7% (80.6–97.5) and 94.3% (91.4–96.3). Cohen’s κ was 0.934 (Xpert) and 0.735 (TB-LAMP), with discordance skewed toward culture-unconfirmed positives (20 vs. 3; p = 0.00049). Positive predictive values were 90.9% and 65.5%; positive likelihood ratios 85.6 and 16.3. Xpert MTB/RIF closely matched culture. TB-LAMP performed comparably as a rule-out test, useful where Xpert is unavailable, but positivity raised post-test probability of culture-confirmed TB to only ~ 66%, warranting confirmation before treatment. Whether excess TB-LAMP positives are false or reflect disease missed by solid culture requires a composite reference standard.