Estimating the impact of nutritional transition and ending hunger on TB in 12 high-burden countries [TBN 068]

Who

The study modeled the adult population aged >15 years in 12 high TB burden countries: Bangladesh, Brazil, Cambodia, China, India, Indonesia, Myanmar, Pakistan, Philippines, Russia, Thailand, and Viet Nam. These countries represented 64.2% of global incident TB cases among WHO’s 2015 high-burden list. Countries with HIV prevalence >1% and North Korea were excluded because of high uncertainty in TB and BMI estimates.

What

The study examined how population nutrition changes, especially BMI distribution and elimination of undernutrition (“zero hunger”), could affect TB incidence and mortality and progress toward the WHO End TB Strategy targets by 2030.

Key findings:

• Under the current-level scenario, TB burden declined only slowly:
  • TB incidence fell from 196.9 to 172.1 per 100,000 between 2015 and 2030, a 12.9% reduction (CrI: 2.76%–23.6%).
  • TB mortality fell from 46.5 to 40.3 per 100,000, a 14.0% reduction (CrI: 3.7%–23.2%).
• Under the continuing trend scenario (more overweight/obesity, less underweight), compared with current level:
  • Incidence declined by an additional 14.7% (CrI: 12.7%–16.7%).
  • Mortality declined by an additional 15.6% (CrI: 12.5%–19.2%).
• Under the current level + zero hunger scenario, compared with current level:
  • Incidence declined by an additional 32.0% (CrI: 20.0%–43.8%).
  • Mortality declined by an additional 37.3% (CrI: 26.1%–49.6%).
• Under the optimal scenario (continuing trend + zero hunger):
  • Incidence declined by an additional 38.2% (CrI: 27.0%–49.1%).
  • Mortality declined by an additional 42.4% (CrI: 32.1%–53.5%).
  • This corresponded to 20.6 million TB cases prevented and 5.4 million TB deaths averted across the 12 countries from 2015 to 2030.

Countries with the largest projected gains in the optimal scenario included Cambodia, Viet Nam, and Bangladesh. Smaller gains were seen in Brazil and Russia.

Authors’ interpretation: Reducing undernutrition and increasing BMI at population level could substantially reduce TB burden and meaningfully support progress toward the End TB Strategy, though not fully achieve the 80% incidence reduction target.

Policy implication: TB control should be linked with nutrition policy, hunger reduction, and coordination with non-communicable disease sectors.

When

The model projected outcomes from 2015 to 2030, with 2015 as the baseline year, aligned with the start of the WHO End TB Strategy. Historical calibration used WHO TB incidence estimates from 2006 to 2015. BMI trend estimates covered 1980–2015.

Where

The analysis focused on 12 high TB burden countries across Asia, Latin America, and Eastern Europe. TB incidence data came from WHO country-specific estimates, and BMI estimates came from a systematic review and pooled analysis of 1,698 population-based measurement studies.

Why

The study aimed to address a gap in understanding how nutrition transition and nutritional interventions may influence TB epidemiology and whether these shifts could help countries move closer to the WHO End TB Strategy target of an 80% reduction in TB incidence by 2030. The rationale was that low BMI increases TB risk, while higher BMI may be protective despite diabetes-related risks.

How

This was a model-based epidemiological analysis using dynamic compartmental TB transmission models stratified by BMI category:

• underweight: BMI <18.5
• normal weight: 18.5 to <25
• overweight: 25 to <30
• obese: ≥30

For each BMI category, the model included six TB states:
susceptible, fast infection, slow infection, pulmonary active disease, recovered, and fast infection after reinfection.

Model features:

• Included TB natural history, case detection rate, and treatment success rate at country level.
• Incorporated BMI effects on:
  • progression from infection to active TB,
  • TB case fatality,
  • general mortality.
• Included both direct BMI effects and indirect effects mediated through diabetes.
• BMI estimates were stratified by age and sex.
• Models were calibrated to WHO data using Approximate Bayesian Computation rejection algorithm and a Euclidean distance fit metric.

Sensitivity analysis showed results were especially sensitive to:

• the relative risk of TB incidence in underweight vs normal weight, and
• the transmission parameter.

Limitations

From the provided text:

• The analysis ignored the impact of COVID-19 on TB and BMI.
• Evidence on BMI and risk of TB infection was described as limited.
• Some countries were excluded due to high HIV prevalence or uncertain estimates, which may affect generalizability.

Strength of evidence

This is a modeling study, not a randomized or observational trial. It provides scenario-based projections rather than direct causal empirical evidence, so conclusions depend on the validity of model structure and input assumptions.

Source: Wu CY, Ku CC, McQuaid CF, Lönnroth K, Cegielski JP, Bentham J, Ezzati M, Lin HH. Estimating the impact of nutritional transition and ending hunger on tuberculosis in 12 high-burden countries: a model-based scenario analysis. BMJ Global Health. 2025 Dec 25;10(12):e018839. https://benangmerah.net/record/58/estimating-the-impact

Sequencing of Pleural Fluid and Plasma for Tuberculous Pleuritis [TBN 067]

Who

• Study population: Adults aged 18 years or older with new-onset unilateral pleural effusion requiring thoracentesis.

• Setting: Consecutive patients screened at the Department of Medicine & Therapeutics, Prince of Wales Hospital, Hong Kong SAR.

• Eligible and analyzed cohort: 329 patients with established causes of pleural effusion.

• TB pleuritis group (TBP): 34 patients (10.3% of the cohort).

  • 16/34 had positive pleural fluid M. tuberculosis culture.

  • 13/34 did not undergo pleural biopsy because of physician preference or patient refusal.

• Non-TBP group: 295 patients, including:

  • 134 malignant pleural effusion

  • 93 fluid overload

  • 55 parapneumonic effusion

  • 13 other causes

• Excluded patients:

  • 39/403 eligible patients had uncertain diagnosis

  • 19 lacked pleural fluid adenosine deaminase results

  • 16 lacked M. tuberculosis culture results

• Exclusion criteria: Prior TBP or bacterial pleural infection, prior instrumentation in the ipsilateral pleural space, or >14 consecutive days of anti-TB treatment in the prior 3 months.

What

• Study focus: To assess the diagnostic utility of targeted Mycobacterium tuberculosis sequencing in pleural fluid and plasma for diagnosing tuberculous pleuritis (TBP), compared with conventional diagnostic methods.

• Key innovation: A bioinformatics masking strategy was developed to exclude regions of high genomic similarity between M. tuberculosis and nontuberculous mycobacteria (NTM), aiming to improve taxonomic specificity.

• Primary findings in pleural fluid:

  • M. tuberculosis DNA fragments were detected in all TBP cases, with median 267.6 RP10M (IQR 30.8–2644.3).

  • In non-TBP, fragments were absent in 288/295 (97.6%); 7/295 (2.4%) had low-level signal (0.31–4.34 RP10M).

  • Diagnostic discrimination was excellent:

    • AUC 0.9996 (95% CI 0.9988–1.0000)

  • At a cutoff of 2 RP10M:

    • Sensitivity: 97.1% (33/34; 95% CI 84.7–99.9%)

    • Specificity: 99.7% (95% CI 98.1–100.0%)

• Comparison with conventional pleural fluid tests:

  • M. tuberculosis culture sensitivity: 47.1% (95% CI 29.8–64.9%)

  • Xpert MTB/RIF Ultra sensitivity: 26.5% (95% CI 12.9–44.4%)

  • Sequencing sensitivity was significantly higher than culture (P<0.001).

• Performance by culture subgroup:

  • In culture-positive TBP, sequencing sensitivity was 93.8% (15/16; 95% CI 69.8–99.8%)

  • In culture-negative TBP, sequencing sensitivity was 100.0% (18/18)

  • Xpert Ultra sensitivities:

    • 37.5% for culture-positive TBP

    • 16.7% for culture-negative TBP

• Plasma findings:

  • Paired plasma available for 31 TBP and 257 non-TBP patients.

  • M. tuberculosis DNA detected in 28/31 TBP plasma samples, median 6.2 RP10M (IQR 1.5–27.9)

  • Low-level signal found in 9/257 non-TBP plasma samples

  • AUC 0.9475 (95% CI 0.8929–1.000)

  • Plasma Xpert Ultra sensitivity was 0%

• Drug resistance analysis:

  • 16 TBP samples had reads covering at least 1 of 15 WHO-relevant resistance-associated regions.

  • No resistance-associated mutations were detected.

  • Sequencing findings agreed with phenotypic susceptibility in culture-positive cases, all susceptible to first-line agents.

• Additional biological findings:

  • M. tuberculosis cell-free DNA fragments in pleural fluid and plasma were mostly short fragments.

  • Pleural fluid human cell-free DNA end motif profiles clustered by disease subgroup, suggesting association with pleural pathology.

• Authors’ conclusion from provided text: Targeted M. tuberculosis sequencing of pleural fluid showed very high sensitivity and specificity for TBP and outperformed pleural fluid culture and Xpert Ultra, including in culture-negative TBP cases.

• Practical implication: This approach may be a promising diagnostic method for paucibacillary TB pleuritis, where conventional microbiologic tests often have limited sensitivity.

When

• Recruitment period: September 1, 2022, to March 31, 2024

• Follow-up information for non-TBP diagnoses: reviewed through July 2025

• Study design timeframe: Prospective enrollment during the above study period.

Where

• Location: Prince of Wales Hospital, Hong Kong SAR

• Institutional context: Department of Medicine & Therapeutics

• Ethics and registration: Approved by the Chinese University of Hong Kong / New Territories East Cluster Ethics Committee; registered at ClinicalTrials.gov (NCT05397730) as the MYDNITE study.

Why

• Rationale: TB pleuritis is difficult to diagnose because conventional tests such as pleural fluid culture and PCR have limited sensitivity, especially in paucibacillary disease.

• Knowledge gap: Whether targeted sequencing of pleural fluid and plasma, coupled with a method to distinguish M. tuberculosis from NTM background DNA, can improve diagnostic yield.

• Objective: To evaluate the diagnostic performance of targeted M. tuberculosis sequencing for TBP against conventional diagnostic methodologies.

How

• Study design: Prospective cohort study

  • This is an observational diagnostic accuracy study, which provides moderate evidence for test performance in a real-world consecutive cohort.

• Sampling approach: Consecutive recruitment of eligible patients with unilateral pleural effusion.

• Reference standard for TBP diagnosis: Positive microbiology, histologic evidence, or a composite reference standard.

• Specimens collected:

  • Pleural fluid for targeted sequencing, Xpert Ultra, acid-fast bacilli stain, culture, cell counts, biochemistry, and cytology

  • Paired plasma when available

• Index test: Targeted M. tuberculosis sequencing using a custom-designed Roche Diagnostics capture-probe system

• Bioinformatics method:

  • Compared the M. tuberculosis genome with 2543 reference genomes from 1075 microbial species in the Actinobacteria phylum

  • Regions shared with NTM were masked

  • Only reads aligning to the masked M. tuberculosis genome were counted as M. tuberculosis DNA

  • DNA abundance reported as reads per 10 million (RP10M)

• Sequencing depth: Pleural fluid sequencing had median 41.9 million raw reads (IQR 31.7–52.5 million)

• Statistical/analytic methods mentioned:

  • McNemar’s test for sample size planning

  • AUC for diagnostic performance

  • Bonferroni correction for multiple comparisons in end motif analysis

  • Unsupervised hierarchical clustering for motif profile analysis

• Sample size calculation:

  • Based on assumed sensitivity difference between culture (27%) and cell-free DNA analysis (80%)

  • Required 28 TBP patients, inflated to 32 allowing for 10% attrition

• Blinding: Investigators performing molecular analyses were masked to the clinical diagnosis

• Limitations noted or evident from the text:

  • 13/34 TBP patients lacked pleural biopsy and additional histology/microbiology

  • Some diagnostic classification relied on a composite reference standard

  • There was at least one apparent false-positive at the selected cutoff

  • Plasma analysis was limited to samples “if available”

  • Single-center study, which may limit generalizability

• Funding/conflicts of interest: Not specified in the provided text.

Bottom line

In this prospective Hong Kong cohort of 329 patients with pleural effusion, targeted M. tuberculosis sequencing of pleural fluid using a genome-masking alignment strategy showed 97.1% sensitivity and 99.7% specificity for diagnosing TB pleuritis at a cutoff of 2 RP10M, clearly outperforming pleural fluid culture and Xpert Ultra. Performance remained strong even in culture-negative TBP, suggesting this method may be particularly valuable for diagnosing paucibacillary pleural TB.

Source: Lam WJ, Chan KK, Wang G, Lai CK, Kang G, Chan C, Leung AC, Wong NH, Tso CS, Chow KM, Ramakrishnan S. Sequencing of Pleural Fluid and Plasma for Tuberculous Pleuritis. NEJM evidence. 2026 Mar 24;5(4):EVIDoa2500237. https://benangmerah.net/record/46/sequencing-of-pleural

Targeted sputum sequencing for rapid and broad drug resistance of MTB [TBN 066]

Who
The study analyzed 55 clinical sputum samples from Kaohsiung Veterans General Hospital (KSVGH): 51 Mtb-positive samples and 4 Mtb-negative controls. Positive samples included 12 samples sequenced on a Flongle flow cell and 40 samples sequenced on a MinION flow cell, with overlap in the described cohorts as labeled F01–F12, M01–M36, and T01–T03. Samples were categorized by acid-fast staining (AFS) grades, including trace, 1+, 2+, 3+, and 4+. The study also used spike-in sputum experiments with Mtb H37Ra and analyzed 442 publicly available nanopore whole-genome datasets (291 from PRJNA650381 and 151 from PRJEB49093).

What
The study developed a streamlined thermo-protection DNA preparation method, a single-tube 17-plex PCR assay targeting 16 resistance-associated genes, and a Python-based resistance prediction pipeline called ARapidTb for targeted nanopore sequencing directly from clinical sputum. The assay was designed around resistance genes and mutations identified from the WHO mutation catalogue and tbdb/TBProfiler database.

Key findings reported in the text:

• The thermo-protection approach increased mapped Mtb read counts from 6,183 to 32,301 and increased the overall mapped-read percentage from 5.73% to 61.08% in the pilot comparison.
• Average read length increased from 844.2 bp in control samples to 942.6 bp in thermo-protected samples.
• In spike-in experiments, the thermo-protection buffer improved enrichment, and the apparent limit of detection was around 873 CFU/ml for sequencing 12 spike-in samples on a MinION flow cell.
• On the Flongle, 6 clinical samples were correctly predicted as susceptible; the workflow was presented as enabling next-morning AMR prediction, with a reported total run cost of about $200 USD, or less than $35 per sample.
• On the MinION, Mtb was detected in 28 of 36 Mtb-positive samples, and correct AMR predictions were made for 21 of those 28 detected samples using 15-hour reads.
• Among 17 clinical samples with AFS 2+ or above, predictions were described as accurate except for one INH-resistant sample (M03) that was falsely predicted susceptible.
• Using 4-hour MinION reads, correct AMR predictions were still obtained for 13 AFS 2+ or above samples plus one trace sample, supporting same-day reporting in some cases.
• In PRJNA650381, ARapidTb showed 95.8% agreement with phenotypic drug susceptibility testing across 1,417 AMR predictions, outperforming Mykrobe and TBProfiler according to the authors.
• In PRJEB49093, ARapidTb results were described as concordant with prior findings that nanopore-based genotyping aligns well with Illumina-based resistance calls. Across 442 isolates, ARapidTb was reported as comparable to Mykrobe and better than TBProfiler overall.

The authors conclude that this workflow can provide rapid, targeted, sputum-based AMR prediction for tuberculosis, especially in higher-smear samples, while expanding mutation coverage beyond some prior targeted assays.

When
The study duration, enrollment period, and dates of sample collection were not specified. Sequencing analyses were evaluated using 4-hour, 15-hour, 24-hour, and 72-hour run scenarios depending on platform and simulation of laboratory turnaround. Spike-in cultures for CFU validation were incubated for 3–4 weeks.

Where
Clinical samples came from Kaohsiung Veterans General Hospital (KSVGH). The study used NaOH-treated sputum samples from that clinical setting. Public external validation datasets were downloaded from PRJNA650381 and PRJEB49093. The reference genome used for primer design was NC_000962.3.

Why
The rationale was to address the need for faster and broader antimicrobial resistance detection for tuberculosis directly from sputum, without relying on slower culture-based workflows or more limited targeted assays. The authors sought to:

• reduce human DNA contamination during DNA preparation,
• expand coverage of resistance-associated mutations beyond existing assays,
• exploit long-read nanopore sequencing more effectively,
• and enable same-day or next-day AMR prediction from clinical samples.

The study also aimed to improve on earlier assays such as Deeplex-MycTB, “nanopore 2021,” and “nanopore 2023” by covering additional genes and larger targeted regions.

How
This was a diagnostic assay development and evaluation study combining laboratory protocol optimization, clinical sample testing, spike-in analytical validation, and comparative bioinformatics benchmarking.

Methods included:

• Primer design: A panel of 10 full-length genes and 6 partial genes was designed using oli2go with target amplicons of about 1000–5000 bp. The final assay became a 17-plex PCR covering 16 resistance-associated genes.
• Sample preparation: Sputum pellets were treated with a thermo-protection buffer (2 M KCl, 0.025 M HEPES, pH 7.5), heat-inactivated at 99 °C for 30 min, bead-cleaned, and eluted for PCR.
• Pilot comparison: One clinical sample was split into control and thermo-protection workflows to assess enrichment.
• Spike-in testing: Mtb H37Ra was serially diluted into Mtb-negative sputum to assess enrichment and detection limit.
• PCR: Multiplex PCR used Platinum SuperFi II Green PCR Master Mix with 35 cycles.
• Sequencing: Libraries were prepared with the Oxford Nanopore Rapid Barcoding Kit 96, then sequenced on Flongle and MinION R9.4.1 flow cells. Basecalling/demultiplexing used MinKNOW v4.3.4 and Guppy v5.0.11.
• Bioinformatics: Reads were mapped to expected amplicons with Minimap2 v2.26; consensus and variants were generated with Medaka v1.5.0; resistance-associated variants were compiled into Tbresdb from tbdb and the WHO catalogue; AMR prediction was performed with ARapidTb.
• Comparators: ARapidTb was compared against Mykrobe v0.12.1 and TBProfiler v2.3.0 using public nanopore datasets.

Limitations noted in the text

• Clinical drug-resistant samples were limited in number.
• Sensitivity was low for some drugs, particularly PZA and ETH.
• ethA deletions were incompletely represented in the resistance database, contributing to false negatives.
• Nanopore indel characterization issues contributed to false positives in other tools.
• The omission of the gid gene from primer design likely reduced streptomycin resistance detection.
• Performance was best in samples with higher smear grades, indicating limited sensitivity in lower-burden samples.

Level of evidence
This is a diagnostic development and validation study, not a randomized or interventional trial. Its evidence is strongest for technical feasibility and comparative diagnostic performance, but more limited for broad clinical effectiveness because the clinical evaluation sample was relatively small and resistant cases were uncommon.

Source: Dou, H. Y., Huang, T. S., Wu, H. C., Hsu, C. H., Chen, F. J., & Liao, Y. C. (2025). Targeted sputum sequencing for rapid and broad drug resistance of Mycobacterium tuberculosis. Infection, 53(4), 1413-1423. https://benangmerah.net/record/31/targeted-sputum-sequencing

Geospatial codistribution of TB and DM in Indonesia [TBN 065]

Who

• Population: 514 districts in Indonesia

• Data Source Population: ~345,000 households from 34,500 census blocks

• Subgroups Identified:

  • Age ≥40 years (higher TB and DM prevalence)

  • Gender differences:

    • TB higher in males (0.38%) vs females (0.22%)

    • DM higher in females (2.85%) vs males (1.90%)

  • Higher-risk groups:

    • Urban residents

    • Lower education levels

    • Informal or non-working populations

• Data Sources:

  • 2023 Indonesian Health Survey (SKI)

  • Badan Pusat Statistik (BPS)

  • Ministry of Health databases

  • Ministry of Home Affairs data

---

What

Study Focus

• Mapping spatial distribution of:

  • Tuberculosis (TB)

  • Diabetes Mellitus (DM)

• Identifying:

  • High-risk districts

  • Socio-demographic risk factors

  • Co-occurring TB-DM burden

Key Findings

• National prevalence

  • TB: 0.30%

  • DM: 2.37%

• Age ≥40 years

  • TB: 0.42%

  • DM: 4.37%

Socio-demographic Associations

TB

• Poverty positively associated

  • β = 0.015 (95% CrI: 0.005–0.024)

• Population density: positive but not significant

DM

• Population density positively associated

  • β = 0.059 (95% CrI: 0.039–0.080)

• Poverty negatively associated

  • β = −0.007 (95% CrI: −0.013 to −0.001)

Geographic Patterns

High TB prevalence

• Papua (highest)

• West Java

• Banten

High DM prevalence

• Central Java

• East Java

• Riau

• Sumatra regions

High TB-DM overlap (62 districts >50% probability)

• West Java

• Banten

• Aceh

• East Kalimantan

• Central Kalimantan

• North Sulawesi

Authors' Conclusions

• TB and DM co-burden shows distinct but overlapping spatial patterns

• High-risk areas often:

  • Urban

  • High population density

  • Low-income settings

Policy Implications

• Targeted geographic interventions recommended:

  • Integrated TB-DM screening

  • Strengthening primary care

  • Resource prioritization for high-burden districts

  • Improved referral pathways

  • Community health worker engagement
    

---

When

• Data year: 2023 Indonesian Health Survey (SKI)

• Study type: Cross-sectional spatial analysis

• Follow-up: Not applicable (ecological cross-sectional)
  

---

Where

• Country: Indonesia

• Geographic unit: 514 districts

• Settings:

  • National survey data

  • District-level socio-demographic indicators

  • Spatial modelling using district shapefiles
    

---

Why

• TB and diabetes increasingly co-occur

• Limited understanding of:

  • Spatial overlap

  • Shared risk factors

  • Geographic clustering

• Objective:

  • Identify districts with dual burden

  • Inform targeted policy and healthcare planning
    

---

How

Study Design

• Ecological spatial study

• Cross-sectional national survey data

• Level of Evidence:

  • Observational ecological modelling study

Methods

• Bayesian Model-Based Geostatistics (MBG)

• Binomial logistic regression

• Fixed + random effects (BYM2 model)

Model Evaluation

• Deviance Information Criterion (DIC)

• WAIC

• RMSE

• Probability Integral Transform (PIT)

• Fivefold cross-validation

Spatial Analysis

• Quintile classification

• Joint exceedance probability mapping

• Residual spatial random effect mapping

Covariates Included

• Population density

• Poverty proportion

• Hospital service ratio

• Primary healthcare availability

Limitations (Implied / Reported)

• Ecological design (district-level aggregation)

• Non-public dataset access

• Potential under-diagnosis in remote areas

• Cross-sectional data limits causal inference
  

---

Strength of Evidence

• Moderate (Ecological spatial modelling study)

• Strong national dataset

• No causal inference possible

---

Narrative Summary

This ecological spatial study analyzed data from 514 districts in Indonesia using the 2023 Indonesian Health Survey to examine the geographic distribution and co-occurrence of tuberculosis (0.30%) and diabetes mellitus (2.37%). Using Bayesian geostatistical modelling, the study identified significant associations between TB and poverty, and between DM and population density, with overlapping high-burden districts concentrated in West Java, Banten, Aceh, Kalimantan, and North Sulawesi. Approximately 62 districts showed high joint probability of TB-DM co-occurrence. Urban residence, lower education, and informal employment were associated with higher prevalence of both diseases. The findings highlight geographic clustering and emphasize the need for integrated TB-DM screening, targeted resource allocation, and strengthened primary healthcare in high-burden districts within Indonesia's decentralized health system.

---

Source: Dwinata I, Tsheten T, Ansariadi A, Wagnew F, Alene KA, Sutarsa IN, Moraga P, Putra IW, Kelly M. Geospatial codistribution of tuberculosis and diabetes mellitus in Indonesia. Infectious Diseases of Poverty. 2026 Mar 30;15(1):37.

Structural Interventions at the Community Level

(Yoseph Samodra)

Host Vulnerability and Clinical Risk Stratification Across the TB Spectrum

This theme integrates evidence on how biological vulnerability, comorbid disease, age, and prior TB exposure shape TB risk, recurrence, diagnosis, and outcomes. See also: Lin TB Lab

1. TB risk is amplified well before “end-stage” disease: Predialysis CKD increased active TB risk even at stage 1, suggesting immune dysfunction occurs earlier than traditionally assumed and warrants earlier TB surveillance.
2. Recurrence is strongly cumulative and non-linear: In treated TB patients, recurrence risk rose dramatically when low BMI, prior TB, and delayed culture conversion coexisted, reaching nearly 30%, far exceeding average recurrence estimates.
3. Extreme age alters TB biology and presentation: In adults ≥85 years, atypical pulmonary TB was more strongly predicted by age itself than by frailty markers, highlighting aging as an independent modifier of TB expression.
4. Malnutrition is both a cause and consequence of TB vulnerability: Low BMI predicted TB recurrence in Taiwan and undernutrition affected one-third of Bangladeshi TB patients, reinforcing malnutrition as a bidirectional driver of poor TB outcomes.
5. Chronic lung disease creates a convergence of risks: Older COPD patients had high LTBI prevalence, with smoking, steroid exposure, and disease duration synergistically increasing infection risk, rather than any single factor dominating.
6. Diabetes consistently worsens TB risk and detection: Diabetes emerged across studies as a predictor of recurrence, delayed diagnosis (aPTB), and undernutrition, reinforcing its role as a central TB amplifier.
7. Short-term TB mortality reflects baseline host reserve: CKD patients who developed TB had markedly higher 1-year mortality, indicating that TB outcomes depend heavily on pre-existing physiological resilience.
8. Traditional symptom-based TB screening fails in high-risk hosts: Nearly 70% of late-elderly TB patients lacked classic symptoms, yet were identifiable through simple clinical–radiologic risk scores, underscoring the need for host-adapted diagnostic strategies.

---

Environment, Living Conditions, and Structural Determinants of TB Transmission

This theme synthesizes evidence showing that where and how people live remains a decisive determinant of TB risk, especially for children and communities.

1. Indoor micro-environments outweigh macro-climate factors: Ventilation, lighting, and housing density predicted pediatric TB, while rainfall, humidity, and population density did not—challenging assumptions about climate-driven TB risk.
2. Household exposure is more decisive than ambient exposure: A family history of TB increased pediatric TB risk nearly tenfold, far exceeding the effect size of environmental or behavioral variables.
3. Housing quality operates as a population-level protective factor: Ecological data showed that coverage of healthy housing correlated inversely with TB incidence, supporting structural interventions beyond individual case management.
4. Orientation and airflow are under-recognized risk modifiers: Non–east-facing bedrooms and inadequate air circulation were significantly associated with childhood TB, highlighting design-level prevention opportunities.
5. Behavioral risk is context-dependent: Cigarette smoke exposure did not independently predict pediatric TB once household and environmental factors were considered, suggesting indirect rather than primary effects in children.
6. Urban TB risk is driven by deprivation, not density alone: In Bangladesh and Indonesia, poor housing and sanitation—not simple crowding—were linked to TB and undernutrition.
7. Environmental interventions are scalable and preventive: Improving ventilation, lighting, and housing standards offers population-wide risk reduction without relying on individual adherence.
8. Community-level indicators can guide TB prevention: Healthy-house coverage and healthy-lifestyle adoption function as actionable surveillance metrics for targeting TB control resources.

---

Health Systems, Treatment Strategies, and Policy-Level Leverage Points

This theme highlights how program design, treatment choices, and funding decisions profoundly influence TB outcomes across populations.

1. Early microbiological response is a critical prognostic marker: Failure of sputum culture conversion at 2 months strongly predicted TB recurrence, supporting intensified monitoring or prolonged therapy in slow responders.
2. Shorter preventive regimens improve adherence—but not without trade-offs: Rifapentine- and rifampin-based LTBI regimens achieved >90% completion in COPD patients but carried higher risks of systemic drug reactions.
3. TB control gains are fragile and funding-dependent: Modelling showed that abrupt withdrawal of US and Global Fund support could more than double paediatric TB deaths, reversing decades of progress.
4. Speed of funding restoration matters more than duration of cuts: Even a one-year interruption in global TB financing caused large mortality surges, but rapid restoration could avert over 90% of excess deaths.
5. Risk-stratified follow-up is more efficient than universal surveillance: Identifying patients with high recurrence risk (e.g., low BMI, prior TB) allows focused use of limited public health resources.
6. Diagnostic tools must be adapted to frontline realities: A simple clinical score enabled non-pulmonologists to detect atypical TB with >95% accuracy, addressing diagnostic delays where specialists are unavailable.
7. Nutrition support is a missing pillar of TB care: Despite strong links between nutrition and TB outcomes, nutritional counseling and food security interventions remain inconsistently integrated into TB programs.
8. TB elimination depends on systems, not just drugs: The studies collectively show that TB outcomes hinge on coordinated action across clinical care, housing policy, nutrition, and global financing, rather than biomedical advances alone.

References:

1. Hsu, C.M., Wu, C.J., Chang, C.J., Pan, S.W., Tseng, Y.H., Huang, J.R., Su, W.J., Feng, J.Y. and Chen, Y.M., 2025. Recurrence of tuberculosis and associated risk factors among Non-HIV patients in Taiwan: A retrospective cohort study. Journal of Infection and Public Health, p.102912.
2. Huang, H.L., Cheng, M.H., Lee, M.R., Chien, J.Y., Lu, P.L., Sheu, C.C., Wang, J.Y., Chong, I.W., Yang, J.M. and Huang, W.C., 2025. Prevalence and treatment outcomes of latent tuberculosis infection among older patients with chronic obstructive pulmonary disease in an area with intermediate tuberculosis burden. Emerging Microbes & Infections, 14(1), p.2497302.
3. Syukur, A., Yulia, Y. and Istikomah, N.R., 2024. Hubungan Kondisi Lingkungan Rumah Dengan Kejadian Tb. Paru Pada Anak Di Kabupaten Sambas. Journal of Innovation Research and Knowledge, 4(6), pp.3795-3806.
4. Menzies, N.A., Brown, T.S., Imai-Eaton, J.W., Dodd, P.J., Cohen, T. and Martinez, L., 2025. Potential paediatric tuberculosis incidence and deaths resulting from interruption in programmes supported by international health aid, 2025–34: a mathematical modelling study. The Lancet Child & Adolescent Health, 9(11), pp.787-795.
5. Park, S., Lee, S., Kim, Y., Lee, Y., Kang, M.W., Cho, S., Han, K., Han, S.S., Lee, H., Lee, J.P. and Joo, K.W., 2019. Association of CKD with incident tuberculosis. Clinical Journal of the American Society of Nephrology, 14(7), pp.1002-1010.
6. Nabi, S.G., Aziz, M.M., Uddin, M.R., Tuhin, R.A., Shuchi, R.R., Nusreen, N., Jahan, R., Afroz, F. and Islam, M.S., 2024. Nutritional status and other associated factors of patients with tuberculosis in selected urban areas of Bangladesh. Well Testing Journal, 33(S2), pp.571-590.
7. Alma, L.R., Olivionita, V. and Wardani, H.E., 2024. An Ecological Study of Determinants of The Incidence of Tuberculosis in Malang Raya Area. Preventia: The Indonesian Journal of Public Health, 9(1), 112–120.
8. Yeh, J.J., Chen, J.H., Kuo, Y.L., Tsai, C.H. and Ko, Y.E., 2025. A Clinical Prediction Model for Atypical Tuberculosis Manifestations Among Older Adults. Medicina, 61(10), p.1888.

TBN 003