Can the High Sensitivity of Xpert MTB/RIF Ultra Be Harnessed to Save Cartridge Costs? Results from a Pooled Sputum Evaluation in Cambodia

Abstract

Despite the World Health Organization recommending the use of rapid molecular tests for diagnosing tuberculosis (TB), uptake has been limited, partially due to high cartridge costs. Other infectious disease programs pool specimens to save on diagnostic test costs. We tested a sputum pooling strategy as part of a TB case finding program using Xpert MTB/RIF Ultra (Ultra). All persons were tested with Ultra individually, and their remaining specimens were also grouped with 3–4 samples for testing in a pooled sample. Individual and pooled testing results were compared to see if people with TB would have been missed when using pooling. We assessed the potential cost and time savings which different pooling strategies could achieve. We tested 584 individual samples and also grouped them in 153 pools for testing separately. Individual testing identified 91 (15.6%) people with positive Ultra results. One hundred percent of individual positive results were also found to be positive by the pooling strategy. Pooling would have saved 27% of cartridge and processing time. Our results are the first to use Ultra in a pooled approach for TB, and demonstrate feasibility in field conditions. Pooling did not miss any TB cases and can save time and money. The impact of pooling is only realized when yield is low.

Systematic Review of Pooling Sputum as an Efficient Method for Xpert MTB/RIF Tuberculosis Testing during the COVID-19 Pandemic

“…All 8 pools containing RIF-resistant samples tested positive for RIF-resistance ( 11 ). However, in Chry et al ( 15 ), of the 3 MTB-positive/RIF-resistant samples subjected to Ultra testing, the pools containing the samples yielded MTB-positive but RIF-sensitive results. Abdurrahman et al ( 12 ) included MTB-positive/RIF-resistant samples in all 4 pools, of which 3 were detected by Xpert MTB/RIF as MTB-positive/RIF-resistant and 1 as MTB-positive/RIF-sensitive.…”

“…One study collected smear and culture results from all participants in addition to the GeneXpert result ( 11 ). Four studies tested sputum using Xpert MTB/RIF ( 11 – 14 ) and 2 with Ultra ( 10 , 15 ) ( Table 1 ).…”

“…Most of the pools with high numbers of samples (≥6) per pool were in the bench-based study. Only 2 studies reported specificity, 1 in which pools were tested with Xpert MTB/RIF (99%, 95% CI 94%–100%) and 1 in which pools were tested with Ultra (100%, 95% CI 96%–100%; Table 2 ) ( 12 , 15 ).…”

“…The 2 studies ( 13 , 14 ) combining 2 sputum samples per pool reported 87.5% and 96.0% Xpert MTB/RIF sensitivity relative to individual testing ( Figure 2 , panel A). The 4 studies combining 4 samples per pool reported sensitivities of 88% ( 10 ) and 96% ( 12 ) for Xpert MTB/RIF and 95% ( 13 ) and 100% ( 15 ) for Ultra ( Figure 2 , panel B). In 2 studies ( 10 , 13 ), pools combining >4 sputum samples reported lower sensitivity ranges for Xpert MTB/RIF (63%–81%) and for Ultra (80%–100%) ( Table 2 ).…”

An Evaluation of Programmatic Community-Based Chest X-ray Screening for Tuberculosis in Ho Chi Minh City, Vietnam

“…The use of a CAR software to aide CXR interpretation would have been ideal in this high CXR throughput setting. Few CAR software platforms perform as well as experienced radiologists [ 40 , 41 ], and given the availability, quality and cost of radiologists in this setting, it is unlikely that a CAR software would be used to entirely replace human readers. However, a CAR software could be used to reduce workloads by triaging high-confidence normal CXR images, so that the radiologist can focus on providing clinical reports for abnormal images and interpreting the borderline abnormal images.…”