Pooled sputum testing for the detection of pulmonary tuberculosis by Xpert(R) MTB/RIF Ultra: a multi-site cross-sectional diagnostic evaluation study in Bangladesh

This multi-site cross-sectional study in Bangladesh demonstrates that pooled sputum testing using Xpert Ultra achieves diagnostic performance comparable to individual testing for pulmonary tuberculosis while reducing cartridge costs by 55.8%, offering a scalable and cost-effective strategy for resource-constrained, high-burden settings.

The Gist

The Big Problem: Too Many Patients, Not Enough Test Kits

Imagine you are a doctor in a busy city in Bangladesh. You have a very powerful, high-tech machine (called Xpert Ultra) that can sniff out Tuberculosis (TB) in a person's spit (sputum) in less than an hour. It's like a super-sniffer dog that never misses a scent.

However, there's a catch: The test kits (cartridges) are expensive. It's like trying to feed a whole village with a limited supply of gold coins. Because the kits cost so much, the health system can't test everyone who might have TB. Many people slip through the cracks, stay sick, and spread the disease to others.

The Clever Solution: The "Group Taste Test"

The researchers asked a simple question: Can we mix the spit from four different people into one cup and test that mixture instead of testing four separate cups?

Think of it like a group taste test at a bakery.

• The Old Way: You give a taste tester a cookie from Person A, then a cookie from Person B, then C, then D. If any cookie is stale, you know who it belongs to. But you have to taste four times.
• The New Way (Pooling): You mix a crumb from all four cookies into one bowl and give it to the taste tester.
  • If the mix tastes fresh: Great! None of the four people have the "stale" disease. You saved three taste tests.
  • If the mix tastes stale: You know someone in that group has the problem. You then go back and test Person A, B, C, and D individually to find the culprit.

What They Did

The researchers tested this "mixing" idea on 3,043 people in Bangladesh. They took samples from:

1. People visiting local clinics.
2. People in hospitals.
3. People living in crowded city slums (where TB spreads easily).

They tested everyone's spit twice:

1. Individually: The standard way (one person, one test).
2. Pooled: Mixing four people's spit together for one test.

They also compared both methods against the "Gold Standard" (growing the bacteria in a lab), which is the most accurate way to know if someone truly has TB.

The Results: A Win-Win Situation

The study found that mixing the samples worked almost as well as testing them separately, but it saved a massive amount of money.

1. The Accuracy (The "Sniffer Dog" didn't lose its nose)

• Individual Testing: Caught about 89% of the real TB cases.
• Pooled Testing: Caught about 86% of the real TB cases.
• The Verdict: The drop in accuracy was tiny. The "mixing" method missed very few cases, especially the serious ones. It was only slightly less sensitive when the bacteria count was extremely low (like finding a single grain of sand in a bucket of water).

2. The Cost (The "Gold Coins" saved)

• By using the pooling method, they reduced the number of test kits needed by 56%.
• The Analogy: Imagine you have a budget to buy 100 pizzas. With the old method, you feed 100 people. With the pooling method, you can feed 180 people with the same amount of money because you aren't wasting ingredients on people who don't need a full slice.
• In real numbers, this saved the program over $14,000 in just this one study, which could be used to test hundreds more people.

3. The "Trace" Issue
The only time the pooling method struggled was when a person had a very tiny amount of bacteria (called a "Trace" result). Mixing that tiny amount with three other people's spit diluted it so much the machine couldn't smell it.

• The Fix: The researchers suggest that for people who look very sick or have high-risk factors, doctors should still test them individually. For everyone else, the "mixing" method is perfect.

Why This Matters

This study is like finding a magic lever that allows a health system to do more with less.

In countries like Bangladesh, where TB is common and money is tight, this strategy means:

• More people get tested: Instead of testing 100 people, they can test 180 with the same budget.
• Faster detection: More people get diagnosed quickly, so they can start treatment and stop spreading the disease.
• Sustainability: The health system won't run out of money or test kits as quickly.

The Bottom Line

The researchers proved that you don't always need to test everyone individually to catch the disease. By smartly "pooling" samples, you can catch almost all the cases while cutting your costs in half. It's a simple, clever trick that could help save lives in many parts of the world.

Technical Summary

1. Problem Statement

Tuberculosis (TB) remains a critical global health threat, particularly in high-burden, low- and middle-income countries (LMICs) like Bangladesh. While the WHO-recommended Xpert® MTB/RIF Ultra (Xpert Ultra) assay offers rapid and sensitive molecular detection of Mycobacterium tuberculosis (MTB) and rifampicin resistance, its widespread implementation is constrained by the high cost of test cartridges. In resource-limited settings, this cost restricts testing capacity, leading to diagnostic gaps where many presumptive TB cases remain undiagnosed. The study addresses the need for a cost-effective strategy to expand molecular testing coverage without compromising diagnostic accuracy.

2. Methodology

Study Design and Setting:

• Type: Prospective, multi-site, cross-sectional diagnostic accuracy study.
• Location: Conducted in Bangladesh between July 2024 and February 2025 across three settings:
  • Primary Healthcare Facilities (PHCFs) in Rajshahi and Narsingdi.
  • Secondary Healthcare Facilities (SHCFs) in Rajshahi and Narsingdi.
  • Urban Slum Communities (USCs) in Dhaka (via community-based case finding).
• Participants: 3,050 adults with presumptive pulmonary TB (PTB) were enrolled. Inclusion criteria included a cough of ≥2 weeks or at least two of the WHO four symptoms (cough, fever, weight loss, night sweats). In slums, AI-enabled portable X-rays were also used for screening.

Procedures:

• Sample Collection: Each participant provided two sputum samples.
  • Sample 1: Processed for Xpert Ultra testing. One aliquot was tested individually; the remaining processed material was pooled with samples from 2–3 other participants (typically groups of 4) to create a pooled sample.
  • Sample 2: Sent for culture (the microbiological reference standard) using the BACTEC MGIT 960 system.
• Pooling Protocol: Pools were created from the remnant sputum-sample reagent (SR) mixture. A minimum of 1 mL from each individual was combined to reach a total volume of ≥2 mL. If a pool tested positive, all individual samples within that pool were retested to identify the positive case(s).
• Reference Standard: Culture results served as the gold standard for calculating sensitivity and specificity. Individual Xpert Ultra results served as the comparator for Positive Percent Agreement (PPA).

Data Analysis:

• Diagnostic performance metrics (sensitivity, specificity, PPV, NPV, accuracy, Kappa) were calculated against culture.
• PPA was calculated against individual Xpert Ultra results.
• Semi-quantitative bacterial burden (High, Medium, Low, Very Low, Trace) and Cycle Threshold (Ct) values were analyzed to assess the impact of dilution.
• Cost analysis compared cartridge usage and total costs between individual and pooled strategies (using a unit cost of $7.97/cartridge).

3. Key Contributions

• First Culture-Validated Study in Bangladesh: This is the first study in Bangladesh to evaluate pooled Xpert Ultra testing using culture as the reference standard, providing a rigorous assessment of diagnostic accuracy.
• Diverse Setting Evaluation: The study uniquely covers primary care, secondary care, and high-density urban slum communities, offering insights into the strategy's applicability across different healthcare delivery models (facility-based vs. community-based).
• Granular Burden Analysis: The study provides detailed data on how pooling affects the detection of different bacterial loads, specifically analyzing the "Trace" category which is often clinically ambiguous.
• Operational Feasibility: It demonstrates that pooling can be implemented with minimal workflow changes and no additional personnel, making it highly scalable for National Tuberculosis Programs (NTP).

4. Key Results

Diagnostic Performance:

• Sensitivity vs. Culture:
  • Individual Xpert Ultra: 89.2% (95% CI: 83.7–93.4).
  • Pooled Xpert Ultra: 85.8% (95% CI: 79.8–90.6).
  • Note: The modest 3.4% reduction in sensitivity is primarily attributed to the dilution effect on low-burden samples.
• Specificity: Both methods maintained high specificity (>98%), with pooled testing showing slightly higher specificity (98.9%) compared to individual testing (98.1%).
• Bacterial Burden Detection:
  • Pooled testing detected 100% of samples with High, Medium, and Low bacterial burdens.
  • Detection rates dropped for lower burdens: 81.8% for "Very Low" and 31.4% for "Trace."
  • When "Trace" results were excluded from the analysis, pooled sensitivity improved to 88.2%, and PPA against individual testing rose to 97.6%.
• Rifampicin Resistance: All 5 rifampicin-resistant cases detected by individual testing were also detected by pooled testing.

Cost and Efficiency:

• Cartridge Reduction: Pooled testing reduced total cartridge usage from 3,192 (individual) to 1,411 (pooled + confirmatory).
• Cost Savings: Total costs decreased from $25,440.20 to $11,245.70, representing a 55.8% cost reduction.
• Operational Efficiency: Pooled testing had a lower rate of invalid/error results (3%) compared to individual testing (4.9%).

Setting-Specific Savings:

• Highest savings were observed in Primary Healthcare Facilities (62.1%), followed by Secondary Facilities (56.2%) and Urban Slums (48.7%).

5. Significance and Conclusion

The study concludes that pooled sputum testing with Xpert Ultra is a viable, scalable strategy for TB diagnosis in high-burden, resource-constrained settings like Bangladesh.

• Trade-off: While there is a slight loss in sensitivity for paucibacillary (very low/trace) cases, the strategy maintains high accuracy for clinically significant bacterial loads (High/Medium/Low).
• Strategic Recommendation: The authors suggest that pooled testing is ideal for community screening and facility-based case finding where the prevalence of high-burden disease is the primary concern. However, for high-risk populations (e.g., HIV co-infection, critically ill, or those with abnormal chest X-rays) where paucibacillary TB is more likely, individual testing should remain the priority.
• Impact: Implementing this approach could more than double the testing capacity with the same financial resources, significantly narrowing the diagnostic gap and enabling earlier detection of TB cases in Bangladesh and similar settings.

This research provides robust evidence for policy-makers to adopt pooled testing as a standard operational procedure to maximize the utility of limited molecular diagnostic resources.