Recruitment in clinics.

Children aged 1–5 years will be identified for recruitment from those attending vaccination clinics, and from those accompanying other people attending routine primary health services, such as antenatal and postnatal care, family planning, and cervical cancer screening. On each day of recruitment, children’s guardians will be approached sequentially by a trained study nurse for consideration of enrolment and, if guardians consent, children will be recruited either while waiting (with care taken that they do not lose their place in the queue) or after finishing their routine visit, until a daily recruitment target is met.

Community sampling.

Children aged 1–5 years, and an additional cohort of adolescents and adults aged 10–40 years, will be eligible for recruitment from households in the study communities. Enumeration of possible household structures within each geographically-defined area has been performed using data from Open Buildings, a large-scale open-access dataset developed from a deep-learning model trained to identify building footprints from high-resolution satellite imagery [24]. Population weights will be defined using open-source gridded population data from WorldPop [25]. Population-weighted random samples will be drawn from the sampling frame of potential building structures within each study neighbourhood, and selected households will be visited by field teams (who will also capture the Global Positioning System (GPS) location of the dwelling, to validate the sampling approach.) Trained study team members will discuss the study with household members and recruit participants on the same visit if they consent. All eligible household residents (or their guardians) will be approached for potential recruitment, with a second visit if required to recruit additional household members not present on the first visit. Where it is not possible to recruit from a location (e.g. a building structure is not a household, or household members decline to participate) teams will visit the next sequential location from the sample until the sample size is reached.

Inclusion and exclusion criteria.

Inclusion criteria are deliberately broad, as the aim is to recruit a community-representative population within these age groups.

Inclusion criteria

1. Children aged 12–60 months (over 1 and under 5 years) of age (clinics and community), adolescents 10–17 years and adults 18–40 years old (community only).
2. Normally resident in one of the study neighbourhoods defined above
3. In clinics: attending for a routine appointment (e.g. vaccination clinic), or accompanying a family member or other person attending for care (e.g. mothers attending for antenatal care).
4. In community: normally resident in the household.

Exclusion criteria

1. Participant or guardian respectively decline to consent.
2. Presenting to PHCs because of their own symptomatic illness.
3. Participant already recruited from the other recruitment group (i.e. from community if recruited in PHCs, or PHCs if recruited in community).

Rationale for sample sizes.

There are limited data on the prevalence of TB immunoreactivity in Blantyre on which to base sample size calculations. The last National Tuberculin Survey in Malawi in 1994 (which did not include Blantyre) found a 12.6% prevalence of TST-positivity amongst children aged 6–11 years, corresponding to an ARTI of 0.6–1.4% [26], with considerable variation by region. In a 2012 tuberculin survey of children aged 2–4 years in Karonga, a rural district in northern Malawi, TST positivity prevalence was 1.1% and the estimated ARTI was 0.3% [15]. Urban areas such as Blantyre are known to have substantially higher incidence of TB disease than rural areas, and in a recent study of 165 adults attending HIV services in urban Blantyre, 44% had a positive IGRA [27]; however the incidence of TB has been falling over the past decade [20], which may result in a lower-than-expected prevalence in young children. In early pilot work, the prevalence of IGRA positivity in children under 5 years was 6/78 (7.7%, 95% CI 2.9–16.0%), with a mean age of 2.4 years, giving an estimated ARTI of 3.3%.

Sample sizes for children aged 1–5 years.

We have based sample size estimates on an ability to precisely estimate a prevalence of 5% with an absolute precision of ±1% (i.e. 4–6%), corresponding to an ARTI of 2.1% (1.7–2.5%) with 95% confidence. This requires a sample size of 1,825 children aged 1–5 years. Of these, we would anticipate around 91 children to have a positive QFT-Plus IGRA. We therefore plan to recruit a total of 1,825 children under 5 with a QFT-Plus result, or 100 children with a positive QFT-Plus IGRA–whichever milestone is reached first–from each of the clinic and community recruitment groups.

Sample sizes for adults and adolescents.

For this objective, we will test the hypothesis that TB immunoreactivity in adolescents and young adults in Blantyre City differs by sex. Based on pilot data, we have assumed an Mtb infection prevalence among females of 15% in adolescents aged 10–17, 20% and 25% in adults aged 18–25 and 26–40 years, respectively. While there are little data from Malawi to inform estimates of the expected sex differences in prevalence, results from adults attending HIV services in Blantyre found a male to female (M:F) ratio of Mtb infection prevalence of 1.6 [27], while data from other settings [28–30] show ratios of 1.2 to 1.8, which themselves vary by age. We have therefore based our sample size on an ability to detect a M:F ratio of 1.35 in each age category, with 80% power and 95% confidence. This requires a sample size of 1,652 (826 male, 826 female) in adolescents aged 10–17, and 1,150 (575 male, 575 female) and 850 (425 male, 425 female) in adults aged 18–25 and 26–40 years respectively. We have not adjusted sample sizes for clustering at household or area level. Firstly, there are no recent data from Malawi to inform clustering estimates of Mtb infection. Secondly, we found no strong basis to assume clustering of Mtb infection at household/area level, as Mtb infection in older age groups may reflect historical exposure trends, which will likely differ even among older people living in the same household or neighbourhood. In analysis (see below) however, we will explore clustering of Mtb infection at household and area level from our data. We will additionally explore implications of such clustering, if any, on potential loss of precision on estimated Mtb immunoreactivity prevalence, and male-to-female ratios.

eLocate.

Guardians of participants recruited in clinics will be asked to provide their (or the child’s) GPS home coordinates using the electronic participant location (ePal) application, which has previously been validated and is used to geolocate record the homes of TB patients in Blantyre [9,23]. A random 5% sample of participants will be selected for independent verification of home coordinates; study teams will visit their home (using the captured GPS coordinates and the verbal description of their address provided in the questionnaire), and compare the true GPS location with that recorded using ePal for the purposes of quality assurance.

QFT-Plus IGRA testing.

A 4mL venous blood sample will be taken for IGRA testing using QFT-Plus, following manufacturer’s instructions. Briefly, samples will be drawn into lithium heparin tubes and transported at room temperature to the laboratory on the same day for transfer into QFT-Plus blood tubes within 12 hours of sample collection, followed by overnight (16–24 hours) incubation at 37°C. Samples will then be centrifuged, and harvested plasma tested using the QFT-Plus assay according to manufacturer’s instructions. A second venous blood sample (maximum volume 1mL) will be taken for either QIAreach QFT IGRA, additional Mtb infection assays, or serum for storage.

Results and follow-up.

Communication of results (children 1–5). Negative IGRA results will be communicated by short message service (SMS), along with standard advice about seeking healthcare if they experience any symptoms of tuberculosis. To facilitate results communication to those who may lack access to their own phone, guardians will also be given the option to attend their local clinic after 2 weeks to collect the results, and will also be given the phone number of a study representative (which they can use if they have temporary access to a phone). Guardians of children aged 1–5 will be informed that all positive results will be delivered either by phone call or, if they do not have access to a phone, via a home visit. Guardians of any children testing positive will be asked to bring their child to the PHC for an in-person review, repeat symptom screen and a follow-up blood sample by the research team, and will be given transport money to do so.

Asymptomatic children aged 1–5 with a positive IGRA. Asymptomatic children with a positive IGRA (defined based on standard definition using manufacturer’s instructions) will be referred to the PHC TB clinic, with a recommendation to initiate TB preventive therapy (TPT) if there are no contraindications. The World Health Organization (WHO) recommends that TB contacts under 5 years old be considered for treatment of “latent” TB infection, given their high risk of progression to active and disseminated disease. TB household contacts under 5 years of age with a positive TST/IGRA have an estimated 19% chance of developing TB over the next 2 years, and the effectiveness of preventive therapy is estimated at 91% [11]. By contrast, the risk of severe (Grade 3+) adverse events in children receiving 6 months of isoniazid (6H) (the current preventive therapy regimen recommended by the Malawi National TB and Leprosy Programme for young children exposed to TB) is estimated at 0.2% [32,33]. (While alternative TPT regimens are also recommended by WHO, these are not widely available for children in Malawi, nor included in the current Malawi National Guidelines for treatment of under-5 TB contacts.) Asymptomatic children will be contacted at 1 and 6 months, and the records of the relevant TB clinics checked to see if they have attended and received/completed TB preventive therapy, and whether they have developed active TB disease or any adverse events. Those who have not attended will be actively traced.

Symptomatic children aged 1–5 with a positive IGRA. If children have TB symptoms on the review visit, they will be referred to the paediatric TB clinic at the Queen Elizabeth Central Hospital (a tertiary-level referral hospital in Blantyre) to be investigated for active TB disease, and will be provided with money for transport. (All diagnostic tests are available free of charge in the government health system). They will receive a diagnostic work-up according to the standard clinical procedures (typically includes a clinical examination, blood tests, chest radiograph and stool and/or gastric aspirate samples for Xpert MTB-Rif/Ultra). Clinic staff will be asked to complete a proforma detailing any investigations performed, the results, and whether the child initiated TB treatment. Children’s guardians will be contacted at 1 and 6 months to ascertain outcomes, which will be cross-checked against clinic records.

Adolescents and adults. WHO does not recommend routine TB preventive therapy in adults and adolescents without HIV or other risk factors [34], and current Malawi guidance does not include provision for preventive therapy for older household contacts of people with TB. In this group, the risk of developing active TB disease in the 2 years following after a positive IGRA/TST result is estimated to be below 5% [35,36], and the risk of side effects from preventive therapy is higher than in children (estimated at 1.5% in people aged 18–34) [33]. A positive test of TB exposure does not require any specific management in this age group in the absence of other risk factors, and a negative test does not exclude the possibility of TB in the future. All participants will therefore be given advice at the time of enrolment, to seek care at a clinic if they develop TB symptoms, and to seek HIV testing if their HIV status is unknown. Those with HIV will be asked if they have previously received TB preventive therapy and, if they have not, advised to ask their ART clinic about this. Adolescents and adults testing positive will be notified of their results via SMS, with a generic reminder message of information given during recruitment about implications of a positive test result (including the need to seek care if they develop symptoms; participants with HIV will be advised to inform their ART clinic of the result). They will also be given the phone number of a study representative, and if they do not have access to a telephone will be able to attend their local clinic after 2 weeks to collect the results.

QIAreach QFT diagnostic accuracy study.

In a nested diagnostic accuracy study, the first 1,000 sequentially-recruited participants will have a 1mL paired sample taken for QIAreach QFT, which will be taken directly into the QIAreach QFT sample tube, transported to the laboratory, incubated overnight and then analysed using the QIAreach QFT eHub and eStick (according to manufacturer’s instructions). Results will be compared against paired samples processed using the gold-standard QFT-Plus assay, and sensitivity, specificity, and Cohen’s kappa (with 95% confidence intervals) calculated for both the overall cohort, and subgroups including different age groups and people living with and without HIV. We will also compare quantitative (time to positivity) QIAreach QFT values with quantitative interferon-gamma levels from QFT-Plus.

Additional assays and diagnostics.

Surveillance based on blood tests may be more efficient if other diseases are also included [37]. A subset of consenting participants will also have an additional saved blood sample taken (specific consent will be sought for this and participants can decline additional blood tests while still consenting to participate in the wider study). This may be processed in future for evidence of exposure to other infections, such as malaria, arboviruses, CMV or respiratory viruses, in order to explore the spatial distribution of other conditions and to evaluate the possibility of combining TB surveillance with serosurveillance for other infectious diseases, and may also be used to evaluate novel tests of TB exposure or infection, including CRP, CRISPR-TB [38], and mycobacterial serology. Participants will not be routinely informed of the results of these tests, unless they reveal clinically-relevant information about active infection. Detailed protocols for additional nested studies will be submitted for ethical approval.

Household contacts of children aged 1–5 with a positive IGRA.

All adult household contacts of children aged 1–5 years with a positive IGRA will be advised to test for active TB disease. Guardians will be given one sputum pot per household member, with pre-filled laboratory request forms and specimen bags, and verbal and written (in Chichewa and with pictorial tools) instructions on how to safely produce a single sputum sample [39]. Guardians will be asked to return sputum pots and cards to the study nurse at the clinic from which their child was recruited, and reimbursed for their travel. All sputum samples will be tested using Xpert-MTB/Rif in government laboratories, and the results checked by the clinic nurses. The proportion of household members returning a sputum sample, and the prevalence of Xpert-positive pulmonary TB amongst household contacts, will be calculated.

Acceptability of Mtb infection surveillance.

The participant questionnaire will include items asking about acceptability of the test. A subset of adult participants ≥18 years and guardians from both households and the clinic will be invited to a more in-depth qualitative study on these themes, with around 10 semi-structured in-depth interviews, using an interview guide, performed with guardians from each setting. Adult participants ≥18 years or guardians who decline participation in the overall research study will also be invited to participate in the qualitative study, to ensure that the views of those who did not consent to Mtb infection testing are also included. These interviews will be conducted by trained qualitative researchers, in participants’ first language. These will be supplemented by key-informant interviews with research staff and healthcare workers in the clinic.

Prevalence of Mtb immunoreactivity and the annual risk of TB infection (ARTI) in Blantyre.

The primary study outcome is the prevalence of Mtb immunoreactivity (number of children with a positive IGRA/number tested, expressed as a percentage) in Blantyre. From this, and the mean age of children tested, the mean ARTI can be calculated using[40]: Similarly, age- and sex-specific prevalence of TB immunoreactivity and mean ARTI among adolescents and adults will be calculated.

Mtb infection tests are interpreted by converting a continuous immunological measure (IFN-γ release or skin induration) into a binary positive/negative result. There is incomplete consensus on the most appropriate thresholds [35] or whether the same thresholds are appropriate for capturing individual risk (influenced by age, HIV status and other factors) v.s. population exposure. Quantitative IGRA responses are associated with the intensity of recent TB exposure [41] and with risk of progression to active TB [42]. Sensitivity analyses will also be performed to explore the use of different quantitative cut-offs for QFT-Plus positivity, and to model immunoreactivity as a continuous variable.

Risk factors for Mtb immunoreactivity.

Multilevel models will be developed to examine the individual-, household- and neighbourhood-level risk factors for Mtb immunoreactivity (Table 1). We will specifically explore whether a household contact with TB, and reported time spent in congregate spaces, independently increase the risk of Mtb infection in young children.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Risk factors for Mtb infection, and sources of data.

https://doi.org/10.1371/journal.pone.0291215.t001

Comparison of sampling methodologies.

We will investigate whether children recruited by each sampling methodology (convenience sampling in PHCs vs community-based household sampling) differ with respect to their demographic and health characteristics. We will additionally investigate whether TB immunoreactivity prevalence differs by recruitment methodology, once other factors which may differ between the two sampling strategies are controlled for (by incorporating this into a variable in the models outlined above).

Geospatial distribution of Mtb immunoreactivity.

We will describe the continuous geospatial distribution of Mtb immunoreactivity prevalence in children under 5 across the included areas. The dependent variable is Mtb immunoreactivity (primarily modelled as a binary outcome, but sensitivity analyses may include alternative cut-offs and continuous distributions). We will capture the precise home GPS coordinates of participants and can therefore estimate a continuous distribution of Mtb immunoreactivity prevalence across the city using a Bayesian geostatistical model.

Using an unadjusted model, we will investigate whether there are spatial hotspots of Mtb immunoreactivity in Blantyre (i.e. areas in which posterior risk estimate exceedance is above a specified cut-off), and describe their spatial scale and the magnitude of the variance in prevalence.

We will explore whether these hotspots correspond to areas of high population-adjusted case notification rates, and to previously-identified areas of possible underdiagnosis. We will formally explore whether incorporating the spatial distribution of case notification rates explains some of the spatial variance of the model–i.e. do high case notifications predict high levels of TB transmission or, as we suspect, are case notifications a poor predictor of the burden of undiagnosed TB?

Developing spatial models of Mtb immunoreactivity to explore drivers of spatial hotspots.

We will develop a geostatistical model of Mtb immunoreactivity prevalence which incorporates the individual- and household-level variables explored in the epidemiological analysis, as well as risk factors operating at a neighbourhood level e.g. population density and poverty.

Once other spatial covariables have been adjusted for, we will explore whether living close to an individual with notified tuberculosis is a risk factor for TB immunoreactivity and, if so, over what spatial and temporal scale this occurs. We will further examine for evidence of residual auto-correlation between Mtb infection prevalence, which may correspond to areas of high transmission, and which can then be compared with the results of ongoing sequencing analyses.

Evaluation of the performance of QIAreach QFT relative to QFT-Plus in population-based samples of healthy children, adolescents and adults.

Samples for the novel QIAreach QFT IGRA will be taken for a subset of participants enrolled in the study, and run in parallel to QFT-Plus. The sensitivity, specificity, positive predictive value and negative predictive value of QIAreach QFT will be calculated against the reference standard QFT-Plus, in the whole population and in subgroups, including children under 5. Quantitative QFT-Plus and QIAReach results will also be compared.

Informed consent.

Adults and adolescents ≥ 18 years, and guardians (Legally Acceptable Representatives; generally, a close relative of the participant who may act on their behalf) of participating children who meet inclusion criteria, will be provided with oral and written information about the study by the research team. They will be invited to provide individual informed consent, either in writing or, if illiterate, by a thumbprint confirmation which will be observed and recorded by a witness independent of the study team. Additionally, adolescents aged 10–17 years will be provided with oral and written information about the study and asked for their assent to participate. Consent forms will be stored securely. All study staff are trained in informed consent and Good Clinical Practice.

Participants will be compensated a small monetary amount for their time and transport costs, according to published local guidance [44].

Possible positive and negative impacts of testing for Mtb infection.

Taking blood from participants, especially children, may cause distress. Trained experienced healthcare professionals will conduct all study procedures. The small volume of blood being taken is not expected to cause any health problems. General infection prevention and COVID-19 protocols will be followed to protect the safety of participants and staff.

Testing for evidence of TB immunoreactivity involves testing for a pre-disease state, which may never lead to active TB disease. A diagnosis of Mtb infection may cause distress and stigma. Great care will therefore be taken to explain the nature of Mtb infection, particularly that it is not infectious to others. The potential benefit of household members being screened for active TB disease if a participant tests positive for infection will also be highlighted. Specific questions and focus groups with participants will address the questions of understanding and acceptability of testing for Mtb infection.

WHO recommends that TB contacts under 5 years old be considered for treatment of “latent” TB infection, given their high risk of progression to active and disseminated disease [11], the high efficacy of preventive therapy [11], and the low risk of serious adverse events in this age-group [32,33].We therefore recommend that children under 5 years in this study with a positive IGRA, by definition representing recent TB exposure, are appropriately investigated for active TB and receive TB preventive therapy once active TB has been excluded. The approach to adults and adolescents with a positive test of TB infection is less clear; current Malawi guidelines do not routinely recommend TPT for adult and adolescent household contacts of people with TB; furthermore older participants in this study are not household contacts with a known, recent exposure but rather healthy community members who may have been exposed to TB years ago, who have a low risk of progression to TB disease [35,36] and an appreciable risk of side effects from preventive therapy [33]. Nevertheless as countries undergo epidemiological shifts towards TB elimination it becomes more important to understand the individual- and community-level risks and benefits of Mtb infection screening and treatment [6].

Approvals and regulatory issues.

Ethical approval has been received from both COMREC (Kamuzu University of Health Sciences Research Ethics Committee) (Ref P.04/22/3611) and LSHTM (Ref 2774). Approval has been received from the Blantyre District Health Office to conduct the study within PHCs.