In-home molecular testing of tongue swabs and sputum to inform household-level screening with diagnostic escalation strategies for tuberculosis contact investigation: a prospective cohort study in South Africa

This prospective cohort study in South Africa demonstrates that in-home molecular testing using tongue swabs, including pooled specimens, enables near-universal screening of asymptomatic tuberculosis household contacts and facilitates rapid linkage to care, offering a scalable alternative to traditional clinic-based sputum testing despite a predictable sensitivity trade-off associated with lower bacterial loads.

The Gist

The Big Problem: The "Clinic Hurdle"

Imagine Tuberculosis (TB) is a hidden fire spreading through a neighborhood. The most effective way to stop it is to check everyone living in the same house as a person who already has the fire (the "Index Patient").

However, the current way of doing this is like telling everyone in the house: "You need to walk 5 miles to a specific building, stand in line, and cough up a sample into a cup. If you can't cough, or if you feel fine, you probably don't need to go."

The result? Many people don't go. Many can't cough up a sample. Many feel fine and think they are safe. The fire spreads because we miss the people who are hiding it.

The New Idea: The "Home Team" Approach

This study tested a new strategy in South Africa. Instead of making people travel to a clinic, the researchers brought the lab to the living room.

They used two main tools:

1. Tongue Swabs: Instead of asking people to cough up mucus (which is hard and embarrassing), they simply swabbed the back of their tongues. It's as easy as brushing your teeth.
2. Pooling: Instead of testing every single person individually right away, they mixed swabs from up to three people into one "soup" and tested that.

The "Fruit Salad" Analogy: How Pooling Works

Think of the household like a fruit bowl.

• The Old Way: You taste every single apple, orange, and banana individually to see if one is rotten. It takes a long time and uses a lot of taste buds (test kits).
• The New Way (Pooling): You take a spoonful of fruit from three people and mix it together.
  • If the spoonful tastes rotten, you know at least one person in that group has the "rotten fruit" (TB). You then go back and test those three people individually to find the specific culprit.
  • If the spoonful tastes fresh, you know none of those three people have the disease. You save time and money, and you can move on to the next group.

What They Found

The researchers tested this on nearly 1,000 people in 439 households. Here is what happened:

1. Everyone could participate (The "Universal" Win)

• Old Way: Only about 1 in 3 people could produce a cough sample. The other 2/3 were left out.
• New Way: Almost everyone (99%) could give a tongue swab. It was easy, non-invasive, and didn't require them to be sick.

2. The "Rotten Fruit" Detection (Accuracy)

• When a person had a lot of TB bacteria (a "big fire"), the tongue swab caught it 100% of the time.
• When the bacteria count was very low (a "tiny spark"), the tongue swab sometimes missed it, especially when mixed with others in a pool.
• The Takeaway: The test isn't perfect for tiny, hidden sparks, but it is excellent for finding the bigger fires that are easier to catch.

3. Speeding Up the Rescue (Linkage to Care)

• Because the test happened at home, people who tested positive didn't have to wait days to get results. They were referred to the clinic immediately.
• 86% of the people who tested positive went to the clinic the very next day to start treatment. This is a huge success because, in the old system, many people would get lost on the way to the clinic.

The Verdict: A Better Strategy

This study suggests we should change how we fight TB in households.

• Don't wait for symptoms: Don't just ask "Do you have a cough?" Many people with TB feel fine.
• Don't rely on coughing: Use tongue swabs because everyone can do them.
• Use the "Group Test" first: Mix swabs to screen the whole house quickly. If the mix is positive, zoom in on the specific people.

The Bottom Line:
By bringing the test to the home and using a simple tongue swab, we can catch TB earlier, test more people, and get them treatment faster. It's like switching from waiting for people to call the fire department, to sending a fire truck to every house to check the smoke detectors.

Technical Summary

1. Problem Statement

Household Contact Investigation (HCI) for Tuberculosis (TB) is a critical strategy for early case detection and breaking transmission chains. However, current implementation in high-burden settings like South Africa faces significant barriers:

• Referral Attrition: Traditional models rely on referring contacts to clinics for screening and sputum collection, leading to high loss-to-follow-up rates.
• Sputum Scarcity: A large proportion of household contacts (HHCs), particularly those with asymptomatic or paucibacillary (low bacterial load) TB, cannot produce sputum, rendering sputum-dependent diagnostics ineffective for this high-risk group.
• Symptom Misalignment: Symptom-based screening is poorly suited for HHCs, as TB in this population is frequently asymptomatic.
• Lack of In-Home Data: There is limited evidence on the diagnostic performance of non-sputum specimens (specifically tongue swabs) collected in-home, and no published data on using pooled tongue swabs as a primary household-level screening tool.

2. Methodology

Study Design & Setting:

• Type: Prospective cohort study.
• Location: Buffalo City Metro Health District, Eastern Cape, South Africa (a high-burden area with ~900–999 TB cases per 100,000).
• Population: 909 adult household contacts (HHCs) of 725 index patients receiving treatment for drug-sensitive TB.
• Timeframe: June 2021 to October 2024.

Intervention & Procedures:

• In-Home Visits: Trained staff visited households within 3–4 days of index patient recruitment.
• Specimen Collection:
  • Tongue Swabs (TS): Up to three TSs were collected per household using HydraFlock swabs and PrimeStore Molecular Transport Medium (PS-MTM).
  • Pooling Strategy: Swabs from up to three different household members were pooled into a single tube (2 mL PS-MTM) for testing.
  • Sputum: Collected individually (not pooled) where possible.
• Testing Platform: Xpert MTB/RIF Ultra performed on portable GeneXpert Omni and Edge devices located in the home.
• Diagnostic Algorithm:
  • Primary Screening: Pooled TS testing.
  • Escalation: Positive pools triggered individual testing; negative pools ruled out the household.
  • Linkage to Care: Positive results triggered immediate referral to clinic-based treatment; negative results triggered referral for TB Preventive Treatment (TPT).

Statistical Analysis:

• Reference Standard: Sputum Xpert Ultra results.
• Primary Outcome: Diagnostic accuracy (sensitivity/specificity) of TS vs. sputum.
• Secondary Outcomes: Linkage-to-care rates, time-to-clinic presentation, and performance stratified by sputum semi-quantitative categories (High, Medium, Low, Very Low, Trace).

3. Key Contributions

• First Evaluation of Pooled In-Home TS: This is the first study to evaluate the diagnostic performance of pooled tongue swab specimens as a primary household-level screening strategy for TB.
• Near-Universal Coverage: Demonstrated that TS collection achieves ~99% coverage in HHCs compared to only ~32% for sputum, effectively addressing the "sputum-scarce" population.
• Biological Gradient Validation: Provided evidence that reduced sensitivity in pooled testing is not a technical failure but a biological gradient correlated with low mycobacterial burden (sputum semi-quantitative categories).
• Operational Model: Proposed a "Household-Level Screening with Diagnostic Escalation" model that shifts from individual referral to pooled screening, optimizing resource use (cartridges and staff time).

4. Key Results

Participant Characteristics:

• Enrollment: 909 HHCs from 439 households.
• Symptom Status: 58.5% were asymptomatic.
• Specimen Yield: 99.1% provided a tongue swab; only 32.1% provided sputum.

Diagnostic Performance:

• Overall Sensitivity/Specificity:
  • Overall TS sensitivity (all pools): 61.9% (95% CI: 38.4–81.9%).
  • Overall TS specificity: 100%.
  • Individual (Unpooled) TS Sensitivity: 100% (5/5 positive cases detected).
• Pooled Testing Performance:
  • In pools where at least one member was sputum-positive:
    • 2-swab pools: 55.6% detected.
    • 3-swab pools: 42.9% detected.
• Impact of Bacterial Burden: Sensitivity declined as sputum bacterial load decreased:
  • High/Medium burden: 100% sensitivity.
  • ≥ Low burden: 70.6% sensitivity.
  • ≥ Very Low/Trace burden: ~54–62% sensitivity.
• Yield Comparison:
  • Symptom-restricted sputum strategy yield: 1.8%.
  • Symptom-agnostic (TS + Sputum) strategy yield: 3.2%.
  • TS-only detection: Identified 3 households with TB that sputum testing missed.

Linkage to Care:

• Of 29 individuals with positive sputum results, 86.2% (25/29) presented to a clinic within a median of 1 day.
• 24/25 (96%) re-tested positive at the clinic and initiated treatment.
• The 4 individuals who did not present were all asymptomatic, highlighting a challenge in engaging asymptomatic patients even after in-home diagnosis.

5. Significance and Implications

• Paradigm Shift: The study supports moving TB HCI from an "individual referral" model to a "household-level screening" model. Pooled TS testing acts as an efficient initial rule-out/rule-in step.
• Efficiency vs. Sensitivity Trade-off: While pooling reduces sensitivity compared to individual testing, the study argues this is acceptable if viewed as a screening tool within a sequential algorithm. Negative pools rule out the household, saving resources; positive pools trigger targeted individual testing.
• Addressing Asymptomatic TB: The ability to test nearly all contacts (including those who cannot produce sputum) significantly expands the diagnostic net for subclinical TB, which is often missed by current referral-based systems.
• Programmatic Recommendations:
  • TB programs should consider pooled non-sputum testing to improve coverage.
  • Diagnostic algorithms must account for the sensitivity drop in low-burden cases (e.g., by prioritizing "sputum-swab" hybrids or ensuring negative pools still receive TPT).
  • Linkage-to-care strategies must be strengthened for asymptomatic individuals who may not perceive a need for treatment despite a positive molecular result.

Conclusion:
In-home molecular testing using tongue swabs, particularly when pooled, offers a scalable, feasible, and high-coverage strategy for TB contact investigation. It effectively overcomes the limitations of sputum dependence and referral attrition, provided that programs implement a sequential diagnostic algorithm that manages the sensitivity trade-offs inherent in pooling.