GXT - GeneXpert or Chest-X-ray or Tuberculin Skin Testing for Household Contact Assessment

Background:

At a recent United Nations high level conference on TB a target was set that more than 30 million individuals should be treated for LTBI over the next 5 years. This ambitious goal will require massive scaling up of LTBI diagnosis and treatment - which will be challenging, if not impossible, with current approaches to diagnosis and treatment of latent TB. Two aspects of management of LTBI - need for CXR and need for TST - are controversial, as both can create important barriers to LTBI initiation, thereby reducing the individual as well as public health benefits of treatment. For TST, difficulties in training and quality assurance of administration and reading, and a global shortage of tuberculin material have reduced uptake. CXR was recommended by WHO in its 2018 guidelines for LTBI management as part of investigations to rule our active TB, however CXR services are not accessible in many settings and even if accessible, the cost for a CXR often falls on patients and is often prohibitively expensive. An alternate strategy is to use GeneXpert equipment, with the Xpert MTB/Rif test (hereafter labelled GX) to replace CXR to exclude prevalent active TB.

Study Objectives:

To compare outcomes from three different strategies for the management of HIV uninfected persons aged 5-50 years who are HH contacts of newly diagnosed microbiologically confirmed active pulmonary TB. (0-5 years only in Brazil)

Design:

This will be a cluster randomized trial with three arms of equal size; clusters will be defined as all the household contacts of patients with newly diagnosed active pulmonary TB. The first eligible member of the HH who provides signed informed consent to participate will be randomized to one of the three strategies. All subsequently enrolled members of the same HH will be assigned to the same study arm.

Study participants will be randomized to one of three different study arms:

1. Standard care (control arm): Participants will receive symptom screening and tuberculin skin testing (TST). If symptom screen positive and/or TST positive, they undergo chest x-rays (CXR). If CXR abnormal, they undergo microbiological investigation. If CXR normal or if microbiological investigation negative, TST positive receive latent TB infection (LTBI) treatment. If microbiological investigation is positive, they will be offered treatment for active TB. For children under 5 years of age in Brazil, sputum induction will be performed for bacteriological investigation.
2. GeneXpert (GX): Participants follow an algorithm similar to the standard care, however participants with positive symptom screen and/or positive TST will receive GX (i.e., GX replaces CXR in standard care algorithm). GX positive are considered to have active TB. TST positive and GX negative receive LTBI treatment. If an individual is not able to provide sputum, they will undergo a CXR.
3. CXR for all/NoTST: Participants will receive symptom screening and CXR. No TST will be performed. If CXR abnormal or symptom positive, they undergo microbiological investigation. If the CXR is normal, and/or microbiological investigations negative - they receive LTBI treatment as per national guidelines. If microbiological investigation is positive they will be offered treatment for active TB.

Population and Setting:

HIV uninfected persons aged 5-50 years who are household contacts newly diagnosed microbiologically confirmed active pulmonary TB can participate in the study. The planned number of household contacts to recruit is about 1434 in total, or about 455 for each of the three arms. The study will take place in five cities in two countries: Benin (Cotonou and Porto Novo) and Brazil (Rio de Janeiro, Manaus and Porto Alegre). In each city clinics will be selected that are representative of the diagnostic facilities and capacities available in most clinics in the country. For Brazil only they will include under 5 HIV uninfected.

Study duration:

Study participants will be recruited over a period of 18 months. Participants will be followed until LTBI treatment is completed (4-6 months depending on what treatment is given)

Outcomes:

Primary Outcome:

Of those eligible (measured or estimated) for LTBI therapy, the proportion starting therapy within 3 months of the index TB patient starting active TB treatment.

Secondary outcomes:

1. Societal costs (health system and patient costs) of the full cascade of care - from initial identification to LTBI therapy completion.
2. Prevalence of microbiologically confirmed and clinically diagnosed active TB - detected as part of the initial contact investigation, who initiate LTBI treatment within 3 months of the index TB patient starting active TB treatment.
3. Prevalence of positive TST (>5 mm or >10 mm) - overall, and by age group.
4. Incidence of grade 1-4 adverse events related to LTBI therapy.
5. Proportion who complete LTBI therapy - defined as having taken at least 80% of doses in 120% of allowed time.
6. Sensitivity and specificity of CXR reading by usual providers in each study site (reference standard will be readings by external reviewers).
7. Prevalence of active TB diagnosed using CXR in participants who cannot produce a sputum sample.
8. Applicable for Brazil only: To evaluate the applicability and performance of material for bacteriological investigation obtained from induced sputum in children under 5 years of age.

Statistical analyses:

Primary analysis The primary outcome is the proportion starting LTBI therapy of those eligible (measured or estimated) for latent TB therapy. 'Eligible' will be defined as: aged 5-50 years, HIV uninfected and TST >5mm. For the no TST arm, the expected prevalence for HIV uninfected HH contacts aged 5 years and older will be estimated from age-specific prevalence of positive TST among HH contacts tested in the other two arms. This expected prevalence will be used to estimate the proportion 'eligible' for LTBI therapy in the no TST arm. Treatment initiation will be defined as being given a prescription for LTBI therapy, or dispensed the first month of pills needed for LTBI therapy. Since this is a dichotomous outcome, the primary analysis will be a logistic regression, using an identity link, and estimated via generalized estimating equations (GEE) to account for clustering by household. An exchangeable correlation structure and empirical standard errors will be used. The proportion starting LTBI therapy within 3 months of the index TB patient starting active TB treatment will be compared in each experimental arm against the standard arm.

Secondary analyses.

1. Societal costs (health system and patient costs) of the full cascade of care - from initial identification to LTBI therapy completion will be considered. Using all relevant cost components and an ingredients approach, differences in health system costs will be assessed for each of the strategies. Patient time will be valuated based on an assumption of income equivalent to the average per capita income in the country. Health care personnel time will be valuated based on average salaries from information provided by facility management in each setting.
2. Prevalence of microbiologically confirmed and clinically diagnosed active TB - detected as part of the initial contact investigation, who initiate LTBI treatment within 3 months of the index TB patient starting active TB treatment, will be compared between all three arms.
3. Prevalence of positive TST (5 mm or 10 mm cut-points) by age group - 0-4 (for Brazil only), 5-10, 11-17, 18-24, 25-34 years, and older in Standard and GX Strategies. This is simple descriptive analysis - and will be presented as overall prevalence in the specified age groups, plus stratified by country.
4. Incidence of serious adverse events related to LTBI therapy. Adverse events are relatively rare dichotomous outcomes. As such, Poisson regression will be used to compare the occurrence of the adverse events between each of the two experimental arms and the conventional arm. To account for clustering by household, GEE, with an exchangeable correlation structure and empirical standard errors will be used. Investigators will compare in the same way - the occurrence of grade 1-2 adverse events reported by study investigators.
5. Completion of LTBI therapy - defined as taking at least 80% of doses in 120% of allowed time. (For 6H this means taking at least 144 doses within 216 days; for 4R this means 96 doses within 144 days, and for 3HR - 72 doses within 108 days.) Since this is a dichotomous outcome, the primary analysis will use logistic regression, with an identity link, and estimated via GEE to account for clustering by household. An exchangeable correlation structure and empirical standard errors will be used. The proportion completing treatment in each experimental arm against the standard arm will be compared.
6. Sensitivity and specificity of CXR reading by usual providers in each study site. For this analysis, the reference standard will be the readings by the external CXR review.
7. Active TB - defined as treatment initiated for active TB - detected only as a result of the CXR done in persons who could not produce a sputum sample.