Tuberculosis (TB) remains the leading cause of death from a single infectious agent globally, with millions of people still undiagnosed or diagnosed late. Conventional case-finding strategies rely heavily on symptom screening using the WHO Four-Symptom Screen ((W4SS; comprising any one of current cough, fever, night sweats, or weight loss) and sputum testing, but these approaches miss a substantial proportion of individuals with active TB disease, particularly those who are asymptomatic or unable to produce sputum. Missed and delayed diagnoses drive ongoing transmission and undermine global TB elimination goals. Recent evidence has shown that diagnostic tools which are more accessible, even if somewhat less sensitive, can still substantially improve TB case detection by reducing diagnostic loss associated with access barriers. This suggests that near point-of-care (NPOC) tests might be highly cost-effective in many settings, because the gains from earlier diagnosis, reduced delays, and broader reach could outweigh losses from slightly lower accuracy. The purpose of this study is to evaluate new, symptom-agnostic screening and diagnostic approaches that can be implemented at lower-level health facilities in high TB-burden, low and middle-income (LMIC) countries for adults ≥15 years and 10-14 years old young adolescents
The study will generate evidence on the performance, cost-effectiveness, feasibility, acceptability, and scalability of symptom-agnostic algorithms initiated by of computer-aided detection chest radiography (CAD CXR-AI) and near point-of-care (NPOC) molecular assays applied to tongue and sputum swabs. These tools have the potential to identify TB earlier, including among asymptomatic individuals, and to reduce dependence on sputum-based diagnostics alone. The research questions being addressed are of direct global relevance. There is currently limited real-world evidence on: how CAD CXR-AI and NPOC tongue swab and sputum swab assays compare as initial screening tools; how they can be integrated with WHO-recommended low-complexity nucleic acid amplification tests (LC-NAATs), in efficient algorithms; and whether these approaches can be delivered effectively in primary care and outpatient settings in high TB burden LMIC. Data generated through this study will directly inform WHO guideline development and national TB programme decisions, especially concerning the detection of asymptomatic TB and the role of non-sputum samples.
Study Type
INTERVENTIONAL
Allocation
NA
Purpose
DIAGNOSTIC
Masking
NONE
Enrollment
37,000
* Near point of care instrument that can test tongue swabs and sputum swabs. * Rapid molecular detection system for detecting infectious diseases included TB, able to provide accurate test results that are comparable to top laboratory PCR tests, while it is easier to use and move around and only takes 15 to 35 minutes to conclude the result.
Semi-quantitative, nested real-time polymerase chain reaction (PCR) diagnostic test for the detection of Mycobacterium tuberculosis (MTB) complex DNA in unprocessed sputum samples\[18\]. It can also detect rifampicin-resistance associated mutations in MTB. Results are automatically displayed on the screen of the system in less than 80 minutes
Pooled testing involves combining equal volumes from multiple individuals' samples and testing them together using a single test\[. Pools will be created using remaining samples from 2-4 participants who have screened positive and were able to produce a sputum, guided by CAD CXR-AI thresholds\[20\]. To the possible extend, pools will be suggested by CAD band score: CAD \<0.3 pooled together and 0.3 ≤ CAD \< 0.8 pooled together.
Portable X-ray systems are designed to bring diagnostic imaging to environments where conventional radiography is impractical. They are lightweight, compact, and battery-powered, making them suitable for use in remote or resource-limited settings, or for reaching people with limited mobility. Depending on the model, they can produce between 100 and 400 images on a full charge, allowing extended use without access to electricity.
Computer-aided detection (CAD) software for chest X-rays is designed to support rapid, automated screening for tuberculosis and other thoracic abnormalities. Software for the study has not been selected yet. It will be a WHO-approved CAD software with final selection through tender processes and in compliance with national regulatory approvals. Operating on mobile or computer platforms, these tools can analyse chest X-rays in less than a minute, distinguishing normal from abnormal scans and highlighting findings in the lungs, pleura, mediastinum, bones, diaphragm, and heart. In addition to detecting disease, some systems can assist clinicians with tasks such as verifying device placement and measuring distances from anatomical landmarks.
The SCREEN TB\&HIV substudy is implemented only in Cameroon, Nigeria and Kenya. HIV testing will therefore not be conducted in Bangladesh or Viet Nam, as HIV testing is not part of routine care pathways at the participating facilities and the study does not introduce additional HIV testing. In addition, Bangladesh and Viet Nam have substantially lower HIV prevalence, making implementation of the HIV substudy operationally unnecessary and not aligned with clinical need
The CD4 cell count is performed in venous blood in HIV positive patients to assess progression of HIV disease, including risk for developing opportunistic infections. The normal range of CD4 count is from 500 to 1500 cells/mm3 of blood, and it progressively decreases over time in persons who are not receiving or not responding well to ART. Someone with a CD4 count below 200 is described as having advanced HIV disease.
LAM is a glycolipid of the cell wall of Mycobacterium tuberculosis. LAM is excreted in urine, where it can be detected using rapid lateral flow tests. In inpatient settings, WHO strongly recommends using LAM to assist in the diagnosis of active TB in HIV-positive adults, adolescents and children with signs and symptoms of TB (pulmonary and/or extrapulmonary), or with advanced HIV disease (1) or who are seriously ill (2) or else irrespective of signs and symptoms of TB and with a CD4 cell count of less than 200 cells/mm3\[23\]. In outpatient settings, WHO suggests using LF-LAM to assist in the diagnosis of active TB in HIV-positive adults, adolescents and children: with signs and symptoms of TB (pulmonary and/or extrapulmonary) or seriously ill; or else irrespective of signs and symptoms of TB and with a CD4 cell count of less than 100 cells/mm3\[23\]. LAM tests evaluated in SCREEN TB\&HIV are:
Primary Objective 1: To evaluate the diagnostic yield and comparative accuracy of diagnostic algorithms initiated by CAD CXR-AI and/or NPOC tongue swab and sputum swab screening as initial screening tools in a facility-based case finding strategy
Primary Endpoint 1.1: Diagnostic Yield of TB by algorithm, disaggregated by symptom status * WHO: All participants enrolled in the study (≥10 years) * WHAT: Number and proportion (out of attempted TB testing) of participants diagnosed with TB (microbiologically confirmed and clinically diagnosed) per diagnostic algorithm pathway * WHEN: Primary = at completion of diagnostic work-up (Day 1-3); Secondary = confirmed via NTP registry data at treatment initiation * WHERE: Study healthcare facilities and referral laboratories in Bangladesh, Cameroon, Kenya, Nigeria, Viet Nam * WHY: To determine incremental case detection across different screening approaches * HOW MEASURED: oNumerator = number of TB cases identified; oDenominator = total participants screened by each algorithm (including those with invalid and inconclusive results); oStratification = symptomatic vs asymptomatic (per WHO 4-symptom screen); oCase definitions = WHO TB definitions (microbiologically confirmed, clinicaly diagnose
Time frame: Completed within 6 month of data collection
Primary Objective 1: To evaluate the diagnostic yield and comparative accuracy of diagnostic algorithms initiated by CAD CXR-AI and/or NPOC tongue swab and sputum swab screening as initial screening tools in a facility-based case finding strategy
Primary Endpoint 1.2: Comparative diagnostic accuracy (sensitivity, specificity, PPV, NPV) of (i) CAD CXR-AI as an initial screening tool, (ii) NPOC tongue swab and sputum swab testing as initial screening tools. * WHO: All participants with interpretable test results * WHAT: Sensitivity, specificity, PPV, NPV of CAD CXR-AI and NPOC tongue swab and sputum swab as initial screening tools * WHEN: Calculated after completion of reference standard testing for all participants * WHERE: Central data analysis * WHY: To compare performance characteristics of screening tools * HOW MEASURED: * Reference standard = LC-NAAT; * Using standard diagnostic accuracy definitions and reporting all estimates together with 95% confidence intervals; * Receiver operating characteristic (ROC) curves for CAD CXR-AI thresholds * Pre-specified handling of indeterminate/uninterpretable results
Time frame: Completed within 6 month of data collection
Secondary Objective 1: To assess timeliness (time to treatment initiation) and proportion initiated on treatment, of CAD CXR-AI and/or NPOC tongue swab and sputum swab -based initiated algorithms
Secondary Endpoint 1.1: Time from entering healthcare facility to being initiated on TB treatment according to National TB Programme (NTP) register record * WHO: Participants diagnosed with TB * WHAT: Time from entering healthcare facility to initiation of TB treatment, using NTP register records * WHEN: From date of facility attendance (and, where available, reported symptom onset) to date of treatment start * WHERE: Facility records and NTP registers * WHY: To evaluate timeliness of linkage to care under different algorithms * HOW MEASURED: Extract dates of facility attendance (or specimen collection), diagnosis confirmation, and treatment start from registers; calculate time to treatment initiation in days; analyse using descriptive and time-to-event methods; assessment limited to treatment initiation (no post-treatment follow-up).
Time frame: Completed within 6 month of data collection
Secondary Objective 1: To assess timeliness (time to treatment initiation) and proportion initiated on treatment, of CAD CXR-AI and/or NPOC tongue swab and sputum swab -based initiated algorithms
Secondary Endpoint 1.2: Proportion of individuals screening positive on the diagnostic algorithm who then initiate TB treatment. * WHO: All participants screening positive on diagnostic algorithms * WHAT: Proportion who initiate TB treatment * WHEN: At time of registry-confirmed treatment initiation * WHERE: NTP registers * WHY: To determine effectiveness of linkage to care * HOW MEASURED: Numerator = number starting treatment; Denominator = number screening positive
Time frame: Completed within 6 month of data collection
Secondary Objective 2: To evaluate the cost-effectiveness of CAD CXR-AI and NPOC tongue swab and sputum swab as initial screening tools in facility-based case finding
Secondary Endpoint 2.1: Modelled incremental cost per person diagnosed with TB from a societal perspective (disaggregated by provider and beneficiary), comparing multiple screening and diagnostic algorithms where: CAD CXR-AI and NPOC are incorporated as initial screening tests Varying CAD thresholds for determining whether pooled or individual LC-NAAT testing is subsequently used are treated as independent and separate initial CAD CXR-AI screening tests * WHO: All participants enrolled * WHAT: Direct evaluation of costs and cost-effectiveness of CAD vs NPOC tongue swab and sputum swab as: 1. stand-alone initial screening tools 2. part of diagnostic algorithms * WHEN: During study and at study end * WHERE: Facility costing and central analysis * WHY: To inform programmatic adoption * HOW MEASURED: Measured cost per TB case diagnosed, modelled incremental cost-effectiveness ratios
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.
Time frame: Completed within 6 month of data collection
Secondary Objective 2: To evaluate the cost-effectiveness of CAD CXR-AI and NPOC tongue swab and sputum swab as initial screening tools in facility-based case finding
Secondary Endpoint 2.2: Modelled incremental cost per person diagnosed with TB who initiates treatment from a societal perspective (disaggregated by provider and beneficiary), comparing multiple screening and diagnostic algorithms where: CAD CXR-AI and NPOC are incorporated as initial screening tests Varying CAD thresholds for determining whether pooled or individual LC-NAAT testing is subsequently used are treated as independent and separate initial CAD CXR-AI screening tests * WHO: All participants enrolled * WHAT: Direct evaluation of costs and cost-effectiveness of CAD vs NPOC tongue swab and sputum swab as: 1. stand-alone initial screening tools 2. part of diagnostic algorithms * WHEN: During study and at study end * WHERE: Facility costing and central analysis * WHY: To inform programmatic adoption * HOW MEASURED: Modelled cost per person who initiates treatment, modelled incremental cost-effectiveness ratios
Time frame: Completed within 6 month of data collection
Secondary Objective 3: To evaluate feasibility, acceptability, and scalability of CAD CXR-AI and NPOC tongue swab and sputum swab in routine facility workflows
Secondary Endpoint 3.1: Feasibility and acceptability of CAD CXR-AI and NPOC tongue swab and sputum swab from diverse perspectives including people seeking care, and health system, and policy makers * WHO: People seeking care, healthcare providers, policymakers * WHAT: Feasibility, acceptability, and scalability of CAD CXR-AI and NPOC tongue swab and sputum swab * WHEN: During and after implementation * WHERE: Study facilities and through qualitative sub-studies * WHY: To evaluate real-world integration and sustainability * HOW MEASURED: Participant and provider interviews, FGDs; structured observation; time-motion analysis; metrics such as proportion able to provide samples and proportion of valid results
Time frame: Completed within 6 month of data collection
Secondary Objective 4: To evaluate the diagnostic performance, efficiency, and feasibility of CAD-guided pooling compared with individual testing
Secondary Endpoint 4.1: Sensitivity and specificity of CAD-guided pooling relative to individual testing * WHO: All enrolled participants providing tongue or sputum swabs eligible for both individual and pooled testing. * WHAT: Diagnostic accuracy (sensitivity, specificity, PPV, NPV) of CAD-guided pooling compared with individual LC-NAAT results as reference. * WHEN: completion of diagnostic work-up; secondary confirmation against NTP registry data at treatment initiation. * WHERE: Study facilities and laboratories in Bangladesh, Cameroon, Kenya, Nigeria, Viet Nam. * WHY: To determine whether CAD-guided pooling maintains diagnostic accuracy while reducing testing volumes. * HOW MEASURED: * Numerator (sensitivity) = number of TB-positive pools correctly identified. * Denominator (specificity) = total number of TB-negative individuals per reference testing. * Stratification = by CAD score band and specimen type (tongue swab, sputum swab).
Time frame: Completed within 6 month of data collection
Secondary Objective 4: To evaluate the diagnostic performance, efficiency, and feasibility of CAD-guided pooling compared with individual testing
Secondary Endpoint 4.2: Proportion of tests saved and associated change in turnaround time * WHO: All pooled samples from enrolled participants eligible for CAD-guided pooling. * WHAT: Number and proportion of tests saved by CAD-guided pooling compared with individual and turnaround time for reporting results. * WHEN: Measured in real-time during study implementation; time recorded from specimen collection to availability of test result. * WHERE: Study facilities and laboratories in Bangladesh, Cameroon, Kenya, Nigeria, Viet Nam. * WHY: To assess efficiency gains from pooling strategies and their impact on timeliness of results. * HOW MEASURED: * Numerator (tests saved) = number of individual cartridges not required due to pooling. * Denominator (tests saved) = total number of cartridges that would have been required for individual testing alone. * Turnaround time = median hours from specimen collection to result availability; stratified by pooled vs individual pathways.
Time frame: Completed within 6 month of data collection
Secondary Objective 4: To evaluate the diagnostic performance, efficiency, and feasibility of CAD-guided pooling compared with individual testing
Secondary Endpoint 4.3: Feasibility and acceptability of CAD-guided pooling from perspectives of laboratory staff, providers, and policymakers * WHO: Laboratory staff, healthcare and policymakers engaged in or overseeing CAD-guided pooling activities. * WHAT: Perceptions of feasibility (integration into, training , workload, error rates) and acceptability (appropriateness, confidence, value) * WHEN: During implementation and at study end via interviews, discussions, and surveys. * WHERE: Healthcare facilities, laboratories, and NTB programme offices in-country. * WHY: To identify barriers and facilitators to implementation of CAD-guided pooling informing potential scale-up. * HOW MEASURED: * Qualitative data: thematically analysed from key informant interviews and focus groups. * Quantitative data: survey scores on feasibility and acceptability domains (e.g. workload, clarity of purpose, perceived reliability). * Stratification = by stakeholder group (lab staff, providers, policymakers).
Time frame: Completed within 6 months of data collection