Implementation of pooled testing of sputum specimens from multiple individuals for rapid molecular detection of tuberculosis in Cameroon: retrospective evaluation of efficiency, cost, and instrument time to result

This retrospective evaluation of Cameroon's large-scale implementation of pooled sputum testing for tuberculosis demonstrates that combining 2 to 8 specimens significantly increased molecular testing capacity, reduced assay costs from $7.97 to as low as $1.19 per result, and shortened time-to-result compared to individual testing, thereby enabling more people to access diagnosis with limited resources.

The Gist

Imagine you are running a massive bakery in a city where everyone is worried about a specific type of mold (Tuberculosis, or TB) in their bread. You have a very expensive, high-tech machine that can detect this mold instantly, but the machine is slow, and the special "test cartridges" it needs are incredibly costly and in short supply.

If you test every single loaf of bread one by one, you will run out of money and time before you can check half the city. Many people will go hungry (or in this case, sick and undiagnosed) because you can't afford to test them all.

This paper is about a clever trick the bakers in Cameroon used to solve this problem: "The Group Taste-Test."

Here is the story of how they did it, broken down simply:

1. The Old Way: Testing One by One

Previously, if 100 people brought in bread samples, the lab had to run 100 separate tests.

• The Cost: If one test costs $8, testing 100 people costs $800.
• The Time: The machine takes about an hour to run one test. So, 100 tests take 100 hours of machine time.
• The Result: They could only test a few people before running out of money.

2. The New Way: The "Pool" Strategy

The researchers in Cameroon decided to try something different. Instead of testing every loaf individually, they took samples from 2 to 8 people and mixed them together into a single "soup" (a pool) to test at once.

Think of it like this: You have 8 cups of water. You suspect one might be dirty. Instead of testing all 8 cups separately, you pour a little bit from all 8 cups into a new bowl and test that one bowl.

• Scenario A (The Good News): If the "soup" tests negative (clean), you know all 8 people are safe! You didn't have to test them individually. You saved 7 tests and 7 hours of machine time.
• Scenario B (The Bad News): If the "soup" tests positive (dirty), you know someone in that group of 8 is sick. You then take the original 8 samples and test them one by one to find the specific culprit.

3. What Happened in Cameroon?

The team ran this experiment in 16 different labs across Cameroon over a year and a half. They tested over 71,000 people.

Here is what they found:

• Super Efficiency: By mixing samples, they managed to test 38,000 more people than they could have if they tested everyone individually. It's like getting a free ticket to the party for 38,000 extra people because they shared a ride.
• Huge Savings: The cost per person dropped from $8 down to just $2.81 on average. If they used big pools (8 people), it cost only $1.19 per person!
• Speed: Because most pools came back negative (meaning the "soup" was clean), the machine didn't have to run the full hour-long cycle for everyone. The average time to get a result dropped from 66 minutes down to 24 minutes.
• Accuracy: They were worried that mixing samples might dilute the mold so much the machine wouldn't see it. But the machine they used (called "Ultra") is very sensitive. It still found the TB in almost all the positive cases, even when mixed with other samples.

4. How Did They Decide Who to Mix?

The lab workers weren't just guessing. They acted like smart traffic controllers:

• If a patient already had a positive result from a simpler, cheaper microscope test, they tested them individually (no mixing needed).
• If the patient looked healthy or had a negative microscope test, they were put into a pool.
• If the lab was very busy and had many samples, they made bigger pools (up to 8). If the lab was quiet or had many sick people, they made smaller pools (2 or 3).

5. The "False Alarm" Problem

Sometimes, the "soup" would test positive, but when they tested the 8 individuals, none of them were actually sick. This happened about 8% of the time.

• Why? The machine is so sensitive it sometimes picks up tiny, harmless traces of the bacteria that aren't enough to make a person sick, or it's just a glitch in the machine.
• The Fix: The doctors just re-tested those specific people to be sure. It wasn't a disaster; it was just a little extra work to be safe.

The Big Takeaway

This study proves that when money is tight and machines are slow, you don't have to stop testing. By being smart about how you group samples (like carpooling), you can:

1. Test way more people.
2. Save a lot of money.
3. Get results faster.

It's a simple, low-tech solution to a high-tech problem that could help save thousands of lives in countries struggling with Tuberculosis, and it could be used for other diseases too. The "Group Taste-Test" is a winner.

Technical Summary

1. Problem Statement

Despite the availability of sensitive molecular diagnostics like the Xpert MTB/RIF Ultra (Ultra) assay, access remains limited in high-burden settings due to high costs (typically ~$8/test) and infrastructure constraints. In 2024, only 54% of people with tuberculosis (TB) in Cameroon received a molecular test as their initial diagnostic. Smear microscopy, while cheaper, lacks sensitivity. The World Health Organization (WHO) recently recommended pooled testing (combining multiple specimens into one assay) to maximize resource efficiency, but large-scale programmatic data on its implementation, cost-effectiveness, and operational feasibility in real-world settings were lacking.

2. Methodology

Study Design:
A retrospective evaluation of laboratory data collected from October 2023 to March 2025 across 16 GeneXpert laboratories in six regions of Cameroon (Far North, North, West, Littoral, Southwest, and Northwest).

Intervention:

• Pooled Testing Strategy: Laboratory personnel were trained to decide whether to test specimens individually or in pools (sizes 2 to 8) based on local factors: smear microscopy results (smear-positive tested individually), daily testing volume, historical positivity rates, cartridge availability, and instrument capacity.
• Procedure:
  • Specimens <1mL were tested individually.
  • For pooling, sputum-reagent mixtures (2:1 ratio with Sample Reagent) were combined in a new container to create a pool volume of at least 2mL.
  • Algorithm: If the pool tested negative (MTB NOT DETECTED), all individuals in the pool were reported negative. If the pool tested positive (MTB DETECTED) or indeterminate, the original individual mixtures were re-tested individually to identify the specific positive case(s).
• Data Analysis: Researchers extracted test results, cycle threshold (Ct) values, and instrument run times from GeneXpert instruments. They calculated efficiency (cartridges saved), cost per result, and time to result, comparing pooled testing against a theoretical individual testing baseline.

3. Key Contributions

• First Large-Scale Programmatic Report: This is the first report detailing the large-scale, operational implementation of pooled TB testing in a national program setting, moving beyond theoretical models or small research studies.
• Pragmatic Implementation Model: It demonstrates a decentralized decision-making model where local lab technicians, rather than a central algorithm, determine pool sizes based on real-time operational constraints (e.g., cartridge shortages, daily volume).
• Validation of Larger Pool Sizes: While WHO guidelines currently recommend pools of up to 4, this study successfully implemented and validated pools of up to 8 specimens in low-prevalence settings.

4. Key Results

Efficiency and Scale:

• Total Volume: 71,328 specimens were tested; 59,164 (83%) were tested in 14,111 pools.
• Cartridge Savings: Pooled testing used only 0.35 cartridges per result obtained. This saved 65% of cartridges compared to individual testing.
• Expanded Access: The strategy enabled 38,326 additional people to receive molecular test results who would not have been tested due to cartridge shortages.
• Positivity Rate: The overall positivity rate in pooled specimens was 3.4% (1,999 positive specimens). Most positive pools (73%) contained only a single positive specimen.

Time to Result:

• Instrument Time: Pooled testing significantly reduced instrument time.
  • Negative Pools: Average time per specimen dropped from ~66 minutes (individual) to 15.5 minutes (pooled).
  • Positive Pools: Even accounting for the follow-up individual testing, the average time per specimen was 88.3 minutes, compared to ~66 minutes for a single individual test.
  • Overall: A 64% reduction in total instrument time was achieved.
• Pool Size Impact: Time efficiency increased with pool size. For example, pools of 8 yielded an average of 10.0 minutes per specimen result, compared to 45.4 minutes for pools of 2.

Cost Analysis:

• Cost Per Result:
  • Individual Testing: $7.97 per specimen.
  • Pooled Testing (Overall): $2.81 per specimen.
  • By Pool Size: Costs ranged from $5.29 (pool of 2) down to $1.19 (pool of 8).
• Total Savings: The program saved approximately $305,458 in cartridge costs ($471,537 projected vs. $166,079 actual) for the 59,164 specimens tested.

Diagnostic Performance:

• Agreement: Bland-Altman analysis showed good agreement between pooled and individual Ct values, with a mean bias of +0.20 to +1.37 cycles depending on pool size.
• Discordance: Approximately 7.8% of positive pools resulted in all negative individual re-tests (likely due to assay variability or low bacterial load). This rate is comparable to discordance seen in other pooled testing applications (e.g., SARS-CoV-2).

5. Significance and Implications

• Resource Optimization: The study proves that pooled testing is a highly effective strategy to overcome cartridge shortages and budget constraints without requiring new infrastructure. It allows existing GeneXpert networks to serve significantly more patients.
• Policy Impact: The findings support the WHO recommendation for pooled testing and suggest that larger pool sizes (up to 8) are feasible and highly efficient in settings with low positivity rates (<5%), challenging the current conservative limit of 4.
• Operational Feasibility: The success of a decentralized, technician-led decision model suggests that national TB programs can implement this strategy rapidly using existing human resources and training protocols.
• Future Directions: The authors suggest that further optimization could be achieved through "informative pooled testing" (using apps to guide pool size based on patient data) and that this approach should be integrated with the rollout of new, lower-cost near-point-of-care diagnostics.

Conclusion:
The implementation of pooled testing in Cameroon successfully expanded molecular TB diagnostics to tens of thousands of additional patients, reduced assay costs by 65%, and saved significant instrument time. It serves as a scalable, low-cost model for other high-burden, resource-constrained settings to mitigate diagnostic gaps.