Abstract Background In 2022, tuberculosis caused 11,200 deaths in Brazil and 1,300,000 globally. With 87,344 new cases, it remains a leading infectious cause of death, second only to COVID-19. Prompt detection is critical, particularly with the growing resistance to rifampicin and isoniazid, highlighting the need for molecular testing. External Quality Assessment Programs (EQAP) are essential to ensure the accuracy, reliability, and harmonization of laboratory results amid the complexities of M. tuberculosis infection. Here, we evaluate the overall performance of participating laboratories (labs) in three EQAP related to tuberculosis: M. tuberculosis molecular detection (MTB), Rifampicin resistance detection (RIF), and Isoniazid resistance detection (INH). All are organized by a Brazilian provider accredited according to the ABNT NBR ISO/IEC 17043:2011 standard. Methods The EQAP samples were suspensions containing human cells, with or without the addition of cultured, inactivated, and lyophilized M. tuberculosis. Labs received two samples quarterly for testing accuracy, sensitivity, and specificity. We computed these metrics along with participant and round numbers. The Mann-Kendall test assessed accuracy trends over time. Furthermore, kit type accuracies for in-vitro diagnostic (IVD) vs. laboratory-developed tests (LDT), World Health Organization recommended (WHO-recommended) vs. other kits (non-WHO), and methods (qPCR vs. PCR) were compared using chi-square tests and odds-ratio calculations. Results In the MTB EQAP (2018-2022), 63 labs participated over 20 rounds, with 22 (range: 6-48) labs per round on median, achieving 96.6% accuracy (853/883), 97.5% sensitivity (556/570), and 94.9% specificity (297/313). The Mann-Kendall test found no significant accuracy trends over time (τ = -0.91, p = 0.36). IVDs outperformed LDTs with 97.8% (751/768) vs. 87.9% (102/116) accuracy (OR=6.0, 95%CI 3.42-20.9, p<0.0001). WHO-recommended kits were more accurate than non-WHO kits, 98.2% (674/686) vs. 90.4% (179/198) (OR=5.96, 95%CI 1.05-12.7, p<0.0001). qPCR exceeded the accuracy of PCR, 97.2% (833/857) versus 74% (20/27) (OR=12.1, 95%CI 4.8-32.9, p<0.0001). For RIF EQAP (2020-2022), 49 laboratories in 9 rounds, with 25 (range: 18-42) laboratories per round on median, showed 97.4% accuracy (297/305), 97.2% sensitivity (176/181), and 97.6% specificity (121/124). In the INH EQAP (2021-2022), 7 laboratories over 7 rounds, with 4 (range: 2-5) laboratories per round on median, reached 95.2% accuracy (20/21), 100% sensitivity (4/4), and 94.1% specificity (16/17). Differences between types of kits and methods for RIF and INH, as well as trends over time, were not significant or could not be calculated due to the limited number of samples. Conclusions The MTB EQAP underscored the high participation and suitability of labs, along with excellent diagnostic performance; IVD kits were 6 times more likely to yield adequate results than LDTs. WHO-recommended kits were 5.9 times more likely to produce adequate results than non-WHO kits. qPCR was 12.6 times more likely to achieve adequate results than PCR. Despite fewer participants in RIF and INH, their performance rates were high; more data are needed for conclusive results on kit and method types. This study affirms the quality of molecular diagnostics for tuberculosis in Brazil, informs on kit and method selection, and underscores the critical role of EQAPs in combating a significant global health issue.
Read full abstract