Community Intervention for Child Tuberculosis Active Contact Investigation and Management: Study Protocol for A Cluster Randomized Trial


 Background: There are major gaps in the management of pediatric tuberculosis (TB) contact investigation for rapid identification of active tuberculosis and initiation of preventive therapy. This study aims to evaluate the impact of a community-based intervention as compared to facility-based model for the management of children in contact with bacteriologically confirmed pulmonary TB adults in low-resource high-burden settings.Methods/design: This multicenter cluster-randomized trial is composed of three phases: I, baseline phase in which retrospective data are collected, quality of data recording in facility registers is checked and expected acceptability and feasibility of the intervention is assessed; II, intervention phase with enrolment of index cases and contact cases in either facility- or community based models; and III, explanatory phase including endpoint data analysis, cost effectiveness analysis and post-intervention acceptability assessment by heath care providers and beneficiaries. The study uses both quantitative and qualitative analysis methods. The community-based intervention includes identification and screening of all household contacts, referral of contact with TB-suggestive symptoms to the facility for investigation, and household initiation of preventive therapy with follow-up of eligible child contacts by community healthcare workers, i.e. all young (<5 years) child contacts or older (5-14 years) child contacts living with HIV, and with no evidence of TB disease. Twenty clusters representing TB diagnostic and treatment facilities with their catchment areas are randomized in a 1:1 ratio to either the community-based intervention arm or the facility-based standard of care arm in Cameroon and Uganda. The primary endpoint is the proportion of eligible child contacts who initiate and complete the preventive therapy. The sample size is of 1500 child contacts to identify a 10% difference between the arms with the assumption that 60% of children will complete the preventive therapy in the standard of care arm.Discussion: This study will provide evidence of the impact of a community-based intervention on household child contact screening, and management of TB preventive therapy in order to improve care and prevention of childhood TB in high low resource high-burden settings.Trial registration: the study has been registered on the 6th of February 2019 on ClinicalTrials.gov with the number NCT03832023 (https://clinicaltrials.gov/ct2/show/NCT03832023?term=CONTACT&cond=Tuberculosis&cntry=UG&draw=2&rank=1 )

contribute to the lost proportion of child contacts initiated on TPT for a long time (13). However, there is evidence that the additional yield of TB disease detection from CXR in asymptomatic child contacts is extremely low (9,(21)(22)(23). In addition, WHO has recommended since 2006 that high-risk child TB contacts -young (< 5 years) or are living with HIV of any age -receive TPT after exclusion of TB disease without systematically con rming TB infection with TST (24,25). Therefore, a symptom-based approach that does not require further investigations for asymptomatic child contacts could facilitate a more decentralized, community-based implementation to initiate TPT in asymptomatic children (8). In addition, the recent WHO recommendations (4) that include shorter TPT combination regimens (isoniazid and rifampicin or rifapentine for 3 months) are associated with improved adherence compared to the standard TPT regimen of isoniazid monotherapy for at least 6 months and provide an important opportunity for increasing completion of TPT (9,26). (27). Further, follow-up of children receiving TPT at the household could further improve TPT completion rates, and could be easily integrated with activities to support treatment of TB disease of the index cases in the household.
There is no published study that has evaluated the impact on the cascade of care of pediatric TB case detection and preventive therapy management of a community-based approach compared to a facility-based standard of care. We therefore aim to evaluate the implementation of a community-based approach to child TB contact screening and management in two TB-endemic African countries.

Study objective and endpoints
The primary objective of this study is to compare the proportion of household child TB contacts eligible for TPT who initiate and complete TPT under a facility-based standard of care and under a decentralized community-based intervention model of care for contact screening and management.
The corresponding primary endpoint is the proportion of child TB contacts < 5 years of age and HIV-infected children of 5-14 years of age who initiate and complete the TPT declared by the index case.
The secondary objectives compare the aforementioned models in terms of: i) cascade of care of TPT initiation and completion in child contacts < 5 years or HIV positive children 5-14 years; ii) cascade of care for TB detection and treatment in all included contacts; iii) tolerability and adherence in children initiated on TPT; iv) acceptability and feasibility of the two models by the parents/guardians, health personnel and community; v) the effect of the community based intervention on the number of adult contacts diagnosed with TB; and the cost-effectiveness. The number of children and adults diagnosed with TB and the number of children initiated on TPT will be also compared before and after the intervention.
The secondary endpoints of the study are presented in Table 1 below:  Study design This is a two arm parallel cluster randomized study comparing two models of care for TB contact investigation and management. This study contains three phases: 1. Baseline phase (phase I) in which retrospective data collection and register quality checks were done in order to assess if the facility registers could be a reliable source of documents for the study. During this phase there was also a baseline qualitative assessment with adult TB patients who are parents and stakeholders to better prepare the intervention phase and assess the acceptability and feasibility of the proposed activities.
2. Intervention phase (phase II) includes implementation and participant recruitment in the two models of care and study data collection.

Study setting
This study is conducted in two high TB incidence, resource-limited African countries: Cameroon located in West Africa and Uganda in East Africa, with important differences in programmatic delivery of TB services. In Cameroon, TB care and management is centralized. Only secondary level health facilities have TB laboratory diagnostic facilities and TB patients can only access care and drugs from these health facilities. In Uganda, TB management is decentralized to the primary health care level. In both countries, national guidelines (28,29) at the time of this study development recommended contact investigation and screening as well as TPT with 6 months of daily isoniazid (6H) for eligible children. However, coverage of TPT for eligible children below 5 years is low in both countries. WHO recently reported that only 24% and 15% of eligible children were initiated on TPT in 2018 in Cameroon and Uganda respectively (1).

Facility-based model
This model implements a "passive" approach to the screening and management of household contacts (see de nition in "Study population" section) at the facility level as per current practice. Implementation follows current National tuberculosis program (NTP) recommendations, except that a 3-month regimen of daily rifampicin-isoniazid (3RH) as a xed-dose combination is offered as TPT to eligible child contacts. In the context of the study, sites also bene t from additional data collection and trainings with follow-up support for the facility staff.
When a person is diagnosed with bacteriologically con rmed pulmonary TB (index case), the facility staff in charge of TB (TB focal person) asks the index case to bring all household contacts with TB-related symptoms, all young (< 5 years) child contacts or older (5-14 years) child contacts living with HIV or exposed to HIV, irrespective of symptoms, to the health facility for evaluation for TB disease or for eligibility for TPT.
In Uganda, the NTP allows household contact tracing but evaluation, TPT initiation and follow-up are required to be done at the facility. At facility, TB investigations include clinical examination, sample collection for smear microscopy or Xpert MTB/RIF testing, and CXR when available and indicated, i.e. Xpert is negative or not done. Any contact diagnosed with TB is commenced on TB treatment, registered and provided with treatment support and follow-up as per NTP guidelines. Asymptomatic children who are eligible to receive TPT as 3RH (or 6H if drug-drug interactions with antiretroviral therapy preclude the use of rifampicin), are initiated at the facility with monthly follow-up. (Fig. 1). The schedule of the facility-based model is presented in Table 2.

Community based model
The intervention model is a decentralized, "active" approach to the screening and management of household contacts and is communitybased. When an index case is diagnosed with bacteriologically con rmed pulmonary TB, the TB focal person asks whether s/he has child contacts in the household, and if so, then asks whether s/he is willing to receive a team in his/her household for contact symptom screening. If they agree, then an appointment is made and a team comprising a trained community health worker (CHW) and a research assistant goes in the household to screen all contacts (children and adults). If the index case doesn't have contact children in their household, then s/he is not included in the study, but contact investigation is done under routine care by the TB focal person. During the contact screening visit, the contacts who present symptoms of TB are referred to the health facility for TB investigations. Those who are asymptomatic and eligible for initiate 3RH (or 6H if drug-drug interactions with antiretroviral therapy preclude the use of rifampicin). The follow-up is done at the household by the CHW after one week, two weeks, and then monthly in order to rapidly identify the children who develop TB symptoms in the community. During the follow-up visits, the CHW repeats the TB symptom screening, assesses the child's TPT tolerability and adherence and assesses presence of any critical sign. If the child presents critical danger signs, tolerability problems or TB symptoms, s/he is immediately referred to the health facility for a clinician to consult them. (Fig. 2). The schedule of the community-based model is presented in Table 2. Study population Bacteriologically con rmed (by smear microscopy, Xpert MTB/RIF or TB-LAMP assays) index cases, ≥ 15 years old who have been diagnosed less than a month prior to inclusion and declaring child contacts in the study catchment area are eligible. Exclusion criteria are: known MDR resistance, the index case being a prisoner or TB patients from an already screened household from the study.
Contacts sharing the same enclosed space for frequent or extended periods of time with the index case or having slept in the same bed during the last 3 months as per the WHO de nition of a contact case (24), are eligible unless they are already on TPT or on TB treatment.
For the qualitative assessment the study population is represented by key informants (facility managers, health staff, community health workers and community leaders) and by male and female TB patients, who are parents/guardians.

Cluster selection and randomization
The study clusters are health facilities supported by the CaP TB Program with TB diagnostic and treatment capacity after an initial assessment taking into consideration the number of bacteriological index cases identi ed from January to December 2018 (minimum of 50). Priority was given to rural, semi-rural or semi-urban facilities as there is less population movement than in an urban setting with relatively easy access. In Cameroon, there were mainly district hospitals because TB diagnosis is mainly done at secondary health care level, with ten clusters selected from two regions (Central and Littoral regions). In Uganda, as TB services are decentralized at primary health care level, the ten clusters were PHC centers in four districts in South West region, some with two facilities per cluster in order to reach the minimum of 50 index cases per year.
The randomization was strati ed by country and in each country, the 10 clusters have been allocated to one of the study models by a covariate constrained randomization (34) taking into account the number of bacteriologically con rmed TB cases from that cluster the previous year.
The randomization was performed by a statistician from the central research team 3 months prior to the start of inclusions. Participants, healthcare providers, study staff and investigators are not blinded to the allocation of the health facilities.
The cluster list can be found in the Supplementary Material.

TPT
The 3RH regimen uses the child-friendly formulation of R75mg/H50mg as a xed-dose combination(30) for eligible child contacts of < 25 kgs This formulation is procured and provided by the CaP TB project, as the NTP has not yet recommended this regimen, but has approved its use in the context of the study. Prescription is based on the body weight dose range as recommended by WHO(31). The body weight is measured at the TPT initiation visit and at the TPT outcome visit for both models and in the facility-based model it is measured at every follow-up visit.
For children of 25 kgs or more, the adult RH tablet is be provided. For children receiving an antiretroviral treatment with protease inhibitors (as lopinavir/ritonavir), nevirapine or dolutegravir, 6H is used to avoid the drug-drug interaction between R and these antiretrovirals. Along with the TPT, 10 mg of daily pyridoxine (vitamin B6) is given to each child to prevent peripheral neuropathy.

Study procedures Symptom screening
Both models use the following symptoms (32) to assess if the contact child has presumptive TB or not: Persistent non-remittent cough > 2 weeks Reported persistent fever > 10 days The presence of at least one of these symptoms requires TB investigations at the health facility.
For HIV positive children, symptoms of any duration are suggestive of TB and the child is immediately referred to the clinic. HIV testing is proposed for the 5-14 years old children with unknown status using two rapid tests, as per national HIV testing guidance. In the community model the rst test is done in the household and if positive, the con rmation test is done at the health facility.
In case a child presents a sign that is not yet suggestive of TB due to its duration (ex: cough for less than 2 weeks), the screening is repeated after two weeks' time.
In addition to the screening for TB symptoms CHWs have been trained to identify critical signs for urgent referral in case the child needs to be seen urgently by a clinician. These signs are recommended by the Integrated Management of Childhood Illness Handbook of the WHO(33) and include: lethargy or unconsciousness, chest indrawing, di culty breathing, sunken eyes, drinking poorly or not drinking, seizures, severe wasting, severe pallor, edema of both feet.

Adherence assessment
The adherence is assessed at each follow-up visit in both models of care using speci c questions on how many doses were missed in the last 4 days, by counting the number of doses taken reported by the parent/guardian in a TPT treatment card introduced by the study and by verifying empty drug blister packs. Parent/guardian received treatment adherence counseling at TPT initiation and during follow-up based on treatment adherence.

Safety assessment
At each follow-up visit the children are assessed for TPT tolerability by CHW using a standard check list of signs suggestive of hepatitis, peripheral neuropathy and rash that are classically associated with HR (nausea, loss of appetite, vomiting, jaundice, dizziness, tingling or burning sensation in the extremities). In every cluster facility a clinician was trained to act as safety monitor and examine children with problems of tolerability identi ed by the CHW. In case of serious adverse events, the safety monitor immediately noti es the event to the country principal investigator who informs the sponsor and the ethics committee of the respective country. All adverse events and serious adverse events are coded using the MedDRA dictionary (version 22.1, September 2019). On a 6 months basis a safety data review is done by the study management team that is then reported to the sponsor and scienti c advisory committee.

Sample size calculation
For the sample size calculation, we used an estimated 60% completion rate among the eligible children in the facility-based arm based on a recent systematic review (35) and a 10% difference in the community-based arm, considered to be the minimal clinically relevant difference.
We consider a cluster coe cient variability of 50% and an intra-cluster correlation of 0.01. With these parameters, we would need to include at least 1500 declared child contacts by the index case who would be eligible to the TPT to have a power of 85%. The type I error rate α, is conventionally xed at 0.05%. Based on national household statistics per country(36-38), we make the hypothesis of one child under 5 years per household. Looking at the index case TB registrations in the year prior to the intervention we estimate that it would be possible to include 1500 contact children in a 15 months period.

Data collection
Mixed methods of data collection, quantitative and qualitative, are used.

Quantitative data
Other than the facility registers, study speci c source documents are used. The data for the primary and secondary objectives are collected by the TFP in the health facilities and community health workers in the community. There is one research assistant assigned to each cluster health facility who enters data onto tablets using the REDCap mobile application version 4.9.1, February 6th 2020. Patients' cost data are collected by research assistants in the REDCap mobile application using an adapted version of the patient cost tool developed by the WHO(39). At the health system level, data are collected through literature, source documents from the Ministry of Health, primary expenditure analysis and procurement records by the cost analysis researchers.

Qualitative data
The data for the qualitative assessment is collected by the social researchers through in-depth interviews and focus group discussions in English, French or local language with the help of a local qualitative research assistant. Participants' con dentiality and privacy are respected throughout the study. During the baseline phase of the study, a qualitative assessment of social determinants has been performed to identify the perceptions of TB, prevention for child contacts and obstacles for treatment, acceptability and feasibility of the proposed intervention.
Focus group discussions have been organized with TB patients and in-depth interviews have been conducted with health staff, facility managers, CHW and community leaders. During the implementation phase, data on concurrent acceptability is assessed through periodic supervision meetings of the CHW. A second qualitative assessment will be performed at the end of the intervention focusing on the acceptability and lessons learned. These activities are planned in both models of care implemented in the study.

Process data
In the baseline phase, sites were assessed in terms of quality of data collection in the registers and speci c practices that would require adjustments for study organization of those sites. During the implementation phase, recruitment logs are lled in by research assistants to document study screening and enrolment process with reasons of refusal.

Data management
A central data manager coordinates with local data managers to ensure the data entry and veri cation according to a Data Management Plan.
There are three levels of data checking and quality control: at data entry using restricted value set or compulsory elds, at country level data management running weekly checks and at central level data management with a monthly consistency data check. The collected data is anonymized by the use of unique study identi cation numbers and followed by the investigators through a dashboard system developed at The study is overseen by a steering committee involving all investigators including representatives of the national TB program with monthly calls to discuss study inclusions, challenges and decisions on study implementation and a scienti c advisory committee composed by experts in the eld of pediatric tuberculosis and randomized controlled trials with meeting twice a year to discuss study progress, challenges and safety review. A country community advisory board is constituted to give guidance on the implementation of the community activities and support the study team on patient information and communication.

Statistical analysis Primary analysis
The denominator for the analysis of the primary endpoint is the number of child contacts < 5 years and HIV-infected 5-14 years declared by the index case at the facility during the inclusion visit. Since discrepancies can be expected between what is declared by the index case and what is observed during contact screening, a sensitivity analysis will be performed using as denominator the number of children < 5 years of age and HIV-infected children of 5-14 years of age identi ed during the screening.
A mixed model will be used to perform individual level analysis adjusting for clustering. The regression model will include two random-effects, one for the cluster and one for the household. A degree-of-freedom correction will be applied (Satterthwaite method) to deal with the type I error in ation due to the small number of clusters. The primary analysis will focus on the difference between the two study arms adjusted for country and a secondary analysis will add an adjustment for unbalanced factors (urban/rural, district size) identi ed in the baseline assessment. For the analysis of the secondary outcomes, a similar mixed model will be used with the same random effects and correction method, focusing on each endpoint of the cascade of care for initiation and completion of TPT and each endpoint of the cascade of care for TB detection.
The proportion of children noti ed in the facility TPT register among all noti ed cases during the intervention period will be compared between the two models of care and will also be compared with the same proportions before intervention for the same time period (data collected during the baseline assessment). The proportion of con rmed TB among all pediatric noti ed cases and the proportion of treatment completion will also be compared between the two models and with the pre-intervention period.

Qualitative assessment analysis
All transcripts from the in-depth interviews and focus group discussions are transcribed in French if the activity was conducted in French and in English if the activity was conducted in local language or English. For the analysis, an interim analysis process will be used. Major themes from the interviews and focus groups will be listed according to the objectives of the study before starting the analyses (a priori codes) and will be enriched if other themes will be found to be relevant to the study objectives (inductive codes). The analysis will be done using the software ATLAS.ti 8 2017.

Cost effectiveness analysis
The two models of care will be analyzed and their cost-effectiveness in each country assessed.
The analysis will be from the health care system's and the primary analysis will generate an incremental cost per Disability-Adjusted Life Year averted for the intervention model of care vs the standard of care, with a mathematical model used to extrapolate effects observed in the trial to a lifetime time horizon. Additional analyses will include reporting of patient costs incurred during illness and care-seeking, and an assetbased wealth quintile of participants, and generation of additional measures of health impact (deaths and TB cases averted).
Ethical aspects

Protocol approval
The study protocol has been submitted and approved by two central IRBs: Advarra IRB from the United States of America, which is the sponsor's institutional IRB and WHO Ethics Research Committee. In addition, the protocol was submitted and approved by the local IRBs Individual consent is obtained by the researchers during qualitative activities: focus group discussions and in-depth interviews and also during the patient cost collection for the cost-effectiveness analysis.

Handling withdrawals
At any moment a contact case can withdraw their consent without any consequence for their care and their data prior to the date of withdrawal are kept for analysis. Their case management continues under the NTP guidelines.

Con dentiality
Each participant has a unique study code. No directly identifying data is entered into the database. An identi cation log allows the research assistants to make the link between the code and the name if needed and this log is kept separately in locked study cabinets on site.

Discussion
The CONTACT study has several strengths and limitations Strengths Methodologically, the use of a randomized cluster-controlled design ensures a good level of evidence. The inter-cluster variability is taken into consideration by the use of a covariate-constrained randomization of the number of index cases per cluster. In addition, the study is using a comprehensive mixed-methods approach that looks at the study goal from different perspectives: quantitative, qualitative and costeffectiveness.
The intervention package was conceptualized in a very pragmatic and realistic manner after discussion with end-users, national TB program and community representative to ensure that it could be implemented by the NTP at the end of the research period. In addition, the intervention is evaluated in two countries with similar TB burden, but very different health system organization and level of community engagement that increases the representativeness of the study results. In Cameroon the national system is very centralized whereas in Uganda lower level health facilities are capacitated to do TB diagnosis and follow-up. In Uganda over the last 2 decades there have been several HIV-related interventions, many of which have been implemented in the communities. The population is used to community activities and a system of CHW is in place (called Village Health Teams) and articulated by the Ministry of Health. The package proposed by the CONTACT study has been inspired by the HIV and malaria community activities (42,43) and integrates very well in the Ugandan context. Finally, all cluster facilities are supported by the CaP TB program for TB diagnosis and treatment, which reduces the risk of heterogeneity between the clusters and TB detection endpoint assessment bias.
The study is constructed on the framework already existent in the health facilities and uses the health personnel of these health facilities. The main strength is represented by their training and experience in working with TB and the fact that they are already integrated in the national system, no study additional staff was hired for this purpose. The TB focal person and safety monitor receive an incentive for lling studyspeci c documentation that is outside their usual work.

Limitations
Because the clusters sites were limited to the facilities supported by the CaP TB project, it was not possible to select more than 20 clusters. It was impossible to avoid urban facilities, which increases the inter-cluster variability and may increase the risk of cluster contamination due to the more complex system of patients' reference in cities as compared to rural settings. The proportion of urban clusters is higher in Cameroon as the two selected regions where the CaP TB project takes place include the two biggest cities of the country. In Uganda, some clusters comprise two health facilities to allow for the necessary recruitment capacity and this operational limitation may introduce more heterogeneity in the measurement of the outcomes.
Another limitation is the reliability of source documents from facility registers as compared to study-speci c source documents, which can induce an information bias and risk of missing data. To minimize this limitation, a register data quality check was done during the baseline period and in sites where inconsistencies were found, a training on data collection was recommended.
The training of facility personnel on the study procedures, the reinforcement of the study source documents and the presence of research assistants is likely to increase the quality of the facility-based as compared to routine conditions and may have an effect on the expected difference of primary endpoint between the community-based and the facility-based models. Also, because the duration of the TPT is known to in uence the completion of the TPT, which is part of the primary endpoint measure and because NTP were expected to change their guidelines in the coming months, we introduced the 3 months regimen in the facility-based model as well. Therefore, the study standard of care does not fully represent the current standard of care used in both countries. The choice of a very operational and pragmatic adherence measure likely to be well accepted by NTP (recording of the dose intake on a treatment card) relies on the parent/guardian's understanding and reliability in recording the dose intake and could potentially introduce information bias or desirability bias. To prevent this risk, the CHW and research assistants are asked to systematically reconcile what is recorded by the parent/guardian on the treatment card with the pills remaining in the blisters. In the community-based model two extra visits after 1 and 2 weeks after starting TPT were requested by the NTP to ensure that no child with TB disease was missed by the symptom screening done by CHW and to verify the tolerability. This results in more frequent assessments of treatment adherence and tolerability as compared to the monthly follow-up in the facility-based model and could introduce an observation bias that could affect the comparison of adherence and safety between the two models.

Challenges
This is an implementation research which is highly dependent on the health system policy and organization. One of the many challenges to be taken into account is the change of the country guidelines during the implementation of the protocol as these changes require most of the time a protocol amendment with implications on study procedures and organization. Other challenges are related to shortages of TB medication, stock-outs of Xpert MTB/RIF cartridges that affect the identi cation of bacteriologically con rmed index cases at facility level and staff availability or turn over. The 3RH and 6H TPT drugs were provided by the study to prevent the risk of shortage. One cluster in Cameroon had to be changed after approval of the study protocol due to an unanticipated concurrent community-based intervention that could bias the study outcome measure. Additionally, in both countries, following the request from the NTP that TPT should be initiated by a nurse, the implementation of the intervention package could not rely on CHW only as initially planned and has to involve facility nurses moving to patients' household. The same applies to HIV testing that cannot be done by trained CHW in Cameroon.
Constant communication with the CaP TB program team and the NTP team at higher and lower levels is crucial to anticipate any operational issue and nd solutions to ensure the continuity of study activities according to the protocol. It also reinforces the level of ownership by the NTP and prepares the future scale up of the intervention. The cost effectiveness and qualitative research components focusing on acceptability and potential barriers such as stigma around tuberculosis and its association to HIV bring crucial information for future scale up.
The CONTACT study will bring new evidence of alternative ways for tuberculosis contact management in a more convenient manner for children and their families with an expected impact on TPT uptake, treatment completion and increase of case detection.

Study status
The study completed the rst phase and participants' enrolment started on October 14 th 2019. Enrolment is expected to be completed in  SPIRITchecklistCONTACT.pdf Clusterlist.docx