World Health Organization · 1st edition, 2026 · 121 pages
WHO guidance for researchers, developers and funders on planning, conducting and reporting studies on TB diagnostic tests to inform global policy.
- Issuer
- World Health Organization
- Edition
- 1st edition, 2026
- Audience
- Test developers, researchers conducting diagnostic accuracy studies, systematic reviewers, research funders, national TB programmes, ministries of health, guideline development group members
- Evidence
- Expert consensus from Scientific GEG Development Group comprising leading trialists, scientists, public health officials, regulators, economists, social scientists and end users; based on WHO handbook for guideline development and GRADE methodology
- Reviewed
- 22 Oct 2026 · GMJ
What this guideline says
GMJ summary
This WHO guidance document provides comprehensive direction for stakeholders involved in generating evidence on new tuberculosis diagnostics, including test developers, researchers, funders, and national TB programmes. It describes how to design, conduct and report studies that can optimally inform WHO policy recommendations, with specific focus on diagnostic test accuracy studies but also covering evidence needs for cost-effectiveness, equity, acceptability and feasibility.
The guidance introduces an analytical framework that maps the intended role of each diagnostic within the clinical pathway—from patient access to testing, test performance, linkage with treatment and health outcomes. It outlines 28 key messages covering study design considerations such as participant selection, reference standard selection, sample size requirements, and reporting standards following STARD and QUADAS-3 frameworks.
The document explains the two main approaches to diagnostic evaluation: the diagnostic-accuracy-based approach (using test accuracy data combined with other evidence to estimate effects on health outcomes) and the people-important-outcomes-based approach (directly measuring the effects of diagnostic interventions on patient outcomes through randomized controlled trials or other comparative studies). It serves as a critical complement to WHO target product profiles for TB diagnostics.
Key points
For clinicians — doses and thresholds as stated in the guideline
- Studies should enrol a consecutive series or random sample of people requiring TB evaluation using cross-sectional or cohort designs; case-control designs should be avoided
- For TB detection, the reference standard should be multiple liquid cultures (ideally from multiple specimens on different days); a single liquid culture may be considered but solid culture alone should not be used
- Sample sizes should include ≥300 participants with confirmed TB (including ~30% paucibacillary specimens) to yield reasonably precise sensitivity estimates
- Studies should include populations with lower bacillary load, people unable to expectorate sputum, children, and people living with HIV early in test evaluations
- All studies should include at least one relevant comparator test representing standard of care (e.g. WHO-recommended molecular diagnostics or smear microscopy for TB detection)
- Specimen processing and testing should follow manufacturer's instructions or standardized final protocols without changes during the study conduct
- Confidence intervals for sensitivity and specificity should be computed using the Wilson score method
- De-identified individual participant data should be made widely available through established repositories using standardized data dictionaries
Conditions for strong versus conditional WHO recommendations for TB diagnostics
Table extracted from the guideline
| Factor | Strong recommendation conditions | Conditional recommendation conditions |
|---|---|---|
| Certainty of evidence | High or moderate certainty of evidence | Low or very low certainty of evidence |
| Balance of benefits and harms | Benefits clearly outweigh harms or vice versa | Balance between benefits and harms is close |
| Stakeholder values and preferences | All or almost all informed stakeholders value outcomes similarly | Variability or uncertainty in how stakeholders value outcomes |
| Resource considerations | Benefit clearly justified in all or almost all circumstances | Benefit may not be justified in some circumstances |
Source: World Health Organization, Guidance on evidence generation on new tuberculosis diagnostics. Reproduced in summary form.
From the guideline
Verbatim
“For adults and adolescents with signs or symptoms of TB or who screened positive for pulmonary TB, low-complexity automated nucleic acid amplification tests (NAATs) should be used on respiratory specimens as initial diagnostic tests for TB, rather than smear microscopy or culture.”
Editor’s perspective
Why this matters for Georgia and the region
Prof. Giorgi Pkhakadze, MD, MPH, PhD
Editor-in-Chief, Georgian Medical Journal · Chair, Public Health Institute of Georgia · WHO consultant, health and migration
As someone who has worked extensively with WHO on health systems strengthening, I recognise this guidance as an essential tool for ensuring that evidence generated on new TB diagnostics can actually inform global and national policy. For the South Caucasus region, where we have active TB programmes and increasing capacity for diagnostic research, the key recommendation I would stress is the imperative to include challenging populations—children, people living with HIV, those unable to produce sputum—early in evaluations rather than as afterthoughts. The analytical framework approach is particularly valuable for our region's researchers and programme managers in articulating how a new test would fit into existing clinical pathways. The realistic implementation caveat is that achieving the recommended sample size of 300 confirmed TB cases with 30% paucibacillary specimens will require multi-centre collaboration across the region, as individual countries may struggle to enrol sufficient numbers within reasonable timeframes, particularly for subgroup analyses.
ORCID 0000-0001-7609-4515 · Perspective written for the GMJ Guidelines Library, 22 Oct 2026
Applicability
Georgia and the South Caucasus
For Georgia and the South Caucasus region, this guidance is directly relevant as countries strengthen TB diagnostic capacity and consider adoption of new diagnostic technologies. National TB programmes in Georgia, Armenia and Azerbaijan should use this framework when planning local evaluations of new diagnostics, ensuring studies capture relevant patient populations (including people living with HIV, children, and those with extrapulmonary TB). The emphasis on generating evidence across diverse geographical regions to capture strain diversity and mutation patterns is particularly important for the Caucasus, where patterns of drug resistance may differ from other regions. Programmes should consider how to balance the desire for locally-generated evidence with participation in multi-country studies that can provide sufficient sample sizes for robust policy recommendations. The guidance on extrapolation of evidence may help programmes determine when international evidence can be applied locally versus when additional national studies are needed.
Questions this guideline answers
Quick answers
What is the difference between the diagnostic-accuracy-based approach and the people-important-outcomes-based approach?
The diagnostic-accuracy-based approach evaluates how well a test correctly identifies disease, then uses that information combined with other evidence to indirectly estimate effects on health outcomes. The people-important-outcomes-based approach directly measures how using a test affects patient health through randomized controlled trials or other comparative studies. The accuracy-based approach is typically used for initial policy recommendations when a test is new, while the outcomes-based approach requires the test to already be in clinical use.
Why can't banked specimens be used for all diagnostic accuracy studies?
Banked specimens may be used if processed and stored appropriately and if equivalency studies demonstrate storage does not affect test performance. However, they are more acceptable for resistance testing than TB detection. Use of banked specimens for paediatric populations may be considered given challenges in obtaining fresh specimens. Derived specimens (spiked with TB culture or DNA) are not acceptable for diagnostic accuracy evaluations used for WHO guidelines as they do not generate clinically relevant sensitivity and specificity values.
How should studies handle inconclusive or invalid test results?
Studies should separately report results that are borderline, unsuccessful (errors or invalid results), indeterminate, and missing, as well as instrument failures for index, reference and comparator tests. Results from repeat testing following inconclusive results should be reported separately. Main analyses should be based on initial test results, with results from repeat testing reported separately. This approach helps assess test robustness and real-world performance.
What sample size is needed for a diagnostic accuracy study of a new TB test?
A body of evidence comprising 300 or more participants with confirmed TB (as defined by the reference standard), including about 30% with paucibacillary specimens, yields reasonably precise estimates of test sensitivity, with diminishing returns beyond this number. Given usual TB prevalence levels, specificity estimates will be highly precise at this sample size. Key subgroups should be adequately represented. Evidence may come from a single study or systematic review of multiple studies, so smaller studies also have value. For drug resistance testing, sample size should consider desired precision separately for each drug.
Related on GMJ
Interconnected resources
Dictionary of Public Health
Conditions A–Z
GMJ News coverage
Quality & accreditation
Other guidelines on this topic
- Special situations in multidrug-resistant and rifampicin-resistant tuberculosis treatment — MSF Tuberculosis
- Treatment regimens in operational research conditions — MSF Tuberculosis
- Treatment Regimens in Programmatic Conditions for Multidrug-Resistant and Rifampicin-Resistant Tuberculosis — MSF Tuberculosis
- Treatment of Multidrug-Resistant and Rifampicin-Resistant Tuberculosis: Introduction — MSF Tuberculosis
- Patient monitoring — MSF Tuberculosis
- Issuer
- World Health Organization
- Type
- Guideline
- Edition
- 1st edition, 2026
- Status
- Current
- Published
- 2026-10-20
- Languages
- English
- Scope
- Global guidance for evidence generation on tuberculosis diagnostics for detection of TB disease and drug resistance
- Audience
- Test developers, researchers conducting diagnostic accuracy studies, systematic reviewers, research funders, national TB programmes, ministries of health, guideline development group members
- Pages
- 121
- Licence
- CC BY-NC-SA 3.0 IGO
- ISBN
- 978-92-4-012379-3
- Evidence
- Expert consensus from Scientific GEG Development Group comprising leading trialists, scientists, public health officials, regulators, economists, social scientists and end users; based on WHO handbook for guideline development and GRADE methodology
- Record history
- Created 22 Oct 2026 · updated 22 Oct 2026 · next check Apr 2027
World Health Organization. Guidance on evidence generation on new tuberculosis diagnostics. Geneva: World Health Organization; 2026. Licence: CC BY-NC-SA 3.0 IGO.GMJ Guidelines Library records are original summaries written by the GMJ editorial team from the issuer’s published document and reviewed by a physician. They do not replace the full guideline. The document is the copyright of its issuer and is linked, not hosted. Spotted an error or a newer edition? Tell the Guidelines Desk.
Was this article helpful?
