National cross-sectional cluster survey of tuberculosis prevalence in Timor-Leste: a study protocol

Abstract

Introduction

Timor-Leste has one of the world’s highest estimated tuberculosis (TB) incidences, yet the data which informs this estimate is limited and the true burden of TB disease is not known. TB prevalence surveys offer the best means of determining robust estimates of disease burden. This study aims to provide an estimate of the prevalence of bacteriologically confirmed pulmonary TB in Timor-Leste and provide additional insights into diagnostic coverage and health-seeking behaviour of persons with symptoms suggestive of TB.

Methods and analysis

A national population-based cross-sectional cluster survey will be conducted in which participants aged 15 years and older will be screened for pulmonary TB using an algorithm consisting of symptom screening and digital X-ray of the chest with computer-aided detection software for X-ray interpretation. Xpert Ultra and liquid culture methods will be used to confirm survey TB cases. Additional data will be collected from persons reporting symptoms suggestive of TB to assess health-seeking behaviour and access to TB diagnosis and care. The survey aims to screen a target sample population of 20 068 people, living within 50 clusters, representing every municipality of Timor-Leste. Bacteriologically confirmed pulmonary TB prevalence will be estimated using WHO-recommended methods.

Ethics and dissemination

Research ethics approval has been granted by the human research ethics committee of the Northern Territory, Australia, and the Instituto Nacional da Saúde, Timor-Leste. The results will be published in a peer-reviewed scientific journal and disseminated with relevant stakeholders.

Trial registration number

ACTRN12623000718640.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study will provide a robust estimate of bacteriologically confirmed pulmonary tuberculosis (TB) prevalence in Timor-Leste, which will provide valuable new information to inform national TB control efforts.

• The study makes pragmatic use of novel tools, screening algorithm and case definitions to optimise feasibility in a resource-limited setting, including rapid molecular tests, ultraportable digital X-ray of the chest and artificial intelligence-driven X-ray interpretation.

• The survey is being carried out in a way which contributes to TB active case finding and opportunistically supports system strengthening and capacity building of the National TB Programme, laboratory network and local health services.

• Budget and laboratory capacity limitations preclude liquid culture testing if sputum is negative for TB on Xpert Ultra.

Introduction

Tuberculosis (TB) is a major cause of morbidity and mortality in Timor-Leste and a recognised priority for public health research.^{1 2} In 2021, there were an estimated 6400 cases of TB nationally, equating to an incidence rate of 486 (322–684) cases per 100 000 population—the seventh highest national TB incidence rate reported globally by the WHO in 2021.³ Yet, there has never been a direct measurement of the prevalence of TB disease in Timor-Leste. National estimates of TB incidence reported by the WHO have remained largely unchanged for two decades⁴ and are limited by the quality of programmatic data used to inform them.

A national TB prevalence survey (NTPS) is a valuable tool used by countries to determine an accurate estimate of the burden of disease and monitor disease trends over time.⁵ In high TB burden and resource constrained settings, such as Timor-Leste, where the treatment coverage rate (calculated as the number of notified TB cases/estimated number of incident TB cases) is low (50%, 95% CI: 35% to 75%),⁴ they are one of the few reliable ways to obtain such estimates. NTPS has been used to generate estimates of TB prevalence in 41 countries worldwide since 2000.³ Data provided by TB prevalence surveys provide vital information to governments, donors and partners to support health service planning and implementation of TB prevention and control strategies.

The NTPS is a priority of the Timor-Leste National TB Programme and will provide a better understanding of the burden of disease caused by TB in Timor-Leste, to inform strategies for TB control and provide a baseline for future evaluation of TB control progress. The primary aim of the NTPS is to estimate the prevalence of bacteriologically confirmed pulmonary TB among the population (aged≥15 years) in Timor-Leste. These data will be used to update national estimates of TB incidence, prevalence, treatment coverage and mortality in Timor-Leste. The study will also provide insights into health-seeking behaviour and access to care among those with TB symptoms. Important secondary aims of the survey are to contribute to national TB active case finding efforts, through linking new survey-detected TB cases into routine care and treatment, and to opportunistically support health system strengthening for TB control in Timor-Leste.

Methods and analysis

Study design

The NTPS is a national cross-sectional survey which will be conducted in Timor-Leste between September 2023 and December 2023. The survey is led by coprincipal investigators from the Timor-Leste Ministry of Health and Menzies School of Health Research. The study design and implementation will be supported by a technical working group, consisting of the principal and associate investigators, other representatives of the Ministry of Health and National Health Laboratory, and technical experts in prevalence surveys, epidemiology, TB diagnostics and clinical management of TB. This technical working group will support the project from conception, through implementation, to analysis and dissemination of findings.

Setting

The Democratic Republic of Timor-Leste is a lower middle-income country located in Southeast Asia. At the time of initial survey planning, it comprised 12 municipalities located on the Eastern side of the island of Timor, as well as the Municipality of Oecusse which is an exclave located to the west of Timor-Leste’s land border with Indonesia.⁶ Timor-Leste has a population of approximately 1.2 million people, of which the majority (approximately 72%) live in rural settings outside of the country’s urban capital of Dili.⁶ Health services in Timor-Leste are predominantly publicly funded and provided, through a network of district hospitals, subdistrict community health centres and village level health posts. TB services are part of the comprehensive service package in Timor-Leste and are delivered through all levels of the health system.⁷

Study population

Eligible survey participants will include persons aged 15 years and older who have resided in one of the selected cluster areas for at least 2 weeks prior to the arrival of the survey team. Individuals residing in congregate settings, such as dormitories, military barracks or boarding schools are excluded, owing to the likely difference in risk profile for TB among these populations compared with the community. Individuals will be excluded if they are unable to provide informed signed consent.

Sample size

Sample size calculation was conducted in line with WHO guidance for TB prevalence surveys.⁵ The sample size calculation incorporated the elements of relative precision, predicted population prevalence of the primary outcome (culture-confirmed TB), co-efficient of between cluster variation, design effect and predicted participation rate.

Estimated population prevalence was 801 cases per 100 000. This is based on the official WHO estimates of prevalence in Timor-Leste in 2019.⁴ In Timor-Leste, 63% of the population is aged 15 years and older,⁸ thus we estimated the prevalence of TB in the target age group to 1268 cases per 100 000 population by dividing the population prevalence by the proportion of the population in the target age group. The WHO TB Prevalence Survey: A Handbook recommends using a relative precision of between 20% and 25%.⁵ We chose a relative precision of 20% in order to obtain a more robust estimate of prevalence and to allow some flexibility in case participation rate or predicted prevalence estimates are inaccurate.

For logistical reasons, it is recommended that cluster size is between 400 and 1000 individuals.⁵ We nominated to set a fixed cluster size of 400 individuals per cluster. This was for both statistical and logistical reasons. To ensure representativeness of the population, we wanted to ensure that at least 50 clusters were included in the survey. To ensure that field teams were not overwhelmed and had sufficient time at each cluster site, we selected a smaller cluster size. To calculate the design effect, a coefficient of between cluster variation of 0.5 was chosen as a safe midpoint, given the lack of available data on likely geographic variation in TB prevalence between clusters. This provided a design effect of 2.28.

A participation rate of 85% was chosen based on data from other NTPSs carried out in Asia.⁹ There was no similar population-based survey conducted in Timor-Leste recently, which could be used to inform this. The sample size was adjusted for the estimated participation rate by dividing the sample size by the predicted participation rate.

Based on the above estimates and calculations, the target sample will include 20 068 individuals surveyed from 50 clusters. This sample size is equivalent to approximately 2.8% of the national population aged 15 years and older.

Due to anticipated differences in the burden of TB between urban and rural areas in Timor-Leste, stratification of the population into rural and urban areas was performed to ensure overall population representativeness. The 50 survey clusters were selected using a probability proportional to size approach. Population data from Timor-Leste’s 2015 census, which was the most recent at the time of sampling, was obtained from the Ministry of Statistics and used for sample selection.¹⁰ Multistage probability proportional to size sampling was conducted, with sampling units being based on each of the administrative divisions of Timor-Leste. The primary sampling unit being municipalities (formerly districts), secondary sampling unity being administrative posts (formerly subdistricts), tertiary sampling unit being Sucos (villages) and quaternary sampling unit being Aldeias (hamlets). Probability proportional to size was applied at each stage, including the final stage, due to substantial variations in the population size of Aldeias. In selected Aldeias where the eligible population exceeds the target population by more than 10%, the cluster will be divided into sections using natural boundaries and landmarks, and randomly selected sections will be enumerated until the target cluster size is reached. In Aldeias where the eligible population is more than 10% smaller than the target size, then screening will be extended to a randomly selected neighbouring Aldeia from the same Suco. Sampled cluster areas may be excluded in the event of insecurity, natural disasters or other significant events which preclude safe access by the survey team. In this event, another cluster area will be randomly selected from within the same sampling unit as the one being replaced.

Survey procedures

Survey procedures will be carried out by one of three teams of four personnel, consisting of a medical officer, radiographer, research nurse and research officer. Prior to commencing fieldwork, the teams will undergo training in census taking, symptom screening, chest radiography and artificial intelligence software use, sputum sample collection and handling, and data collection using electronic survey tools. Socialisation activities will be undertaken at national, municipal and local levels prior to fieldwork, in order to engage key stakeholders from government, health services and partners, as well as community leaders and participating communities.

On arrival in each cluster area, the team will commence operations with census taking of the enumeration area, and establishment of a central site (such as a town hall, community centre or health facility) to conduct TB screening. The team will move from household to household over a period of between 2 and 4 days to enumerate all residents within the area and identify those residents who are eligible to participate in TB screening. Field teams will be supported by community volunteers who will assist the teams in navigating the local community and ensuring that no areas or houses are missed. All persons residing within the enumeration area will have basic demographic details, including age and gender recorded in a census database. Data will be collected electronically using tablets. Eligible participants will be assigned a unique survey identification number and invited to attend TB screening. They will be provided an invitation card which contains their unique survey identification number and information about the location and hours of the screening site.

The survey team will conduct screening within each site over a period of 10–12 days, including weekends. The hours of operation of the screening sites will be dependent on local factors such as access to the site in which screening is being done, seasonal weather conditions and community preferences and availability. However, the teams will generally offer screening between 8:00 and 18:00. On arrival at the screening site, eligible participants will be provided information on the screening process before written informed consent is sought. They will then be registered in the electronic survey database and undergo TB screening. Screening will consist of chest radiography and symptom screening. Those who screen positive will have sputum samples collected for bacteriological testing. The screening algorithm is outlined in figure 1.

Figure 1.

Survey screening and diagnostic testing algorithm. AI, artificial intelligence; MGIT, Mycobacteria Growth Indicator Tube; TB, tuberculosis.

Three ultraportable digital X-ray machines will be sourced to conduct screening. These will include two Delft Light machines and one MinXray HF120/60HPPWV PowerPlus machine. Chest radiography will include a single posterior–anterior radiograph taken by a trained registered radiographer. X-ray images will be interpreted using CAD4TB V.7 (Delft Imaging Systems, the Netherlands), which uses artificial intelligence-driven image interpretation. The software returns an abnormality score between 0 and 100. A propensity score of ≥60 will be used as the threshold to determine positive screening. This threshold was based on the manufacturer’s recommendations and deliberation of the technical working group at the time of protocol development, as conducting a local calibration study was not feasible. Symptom screening will be conducted by a trained doctor, research nurse or research officer, with participants asked if they have cough for longer than 2-week duration, and/or fever, weight loss or night sweats of any duration—based on National TB Programme guidelines in Timor-Leste.¹¹ For any participants who do not undergo X-ray screening, such as pregnant women, a more sensitive symptom screening definition of cough of any duration, fever, weight loss or night sweats will be used.

Screen-positive participants, defined as those with an X-ray of the chest abnormality score≥60 and/or one or more TB symptoms, will be eligible for bacteriological testing for TB. Two sputum samples will be collected from eligible participants, including one spot sputum collected at the screening site, and a second sample collected at least 1 hour after the first sample. Samples will be collected and packaged in accordance with national guidelines and transported daily under cold chain conditions, to a laboratory with GeneXpert MTB/RIF Ultra (Cepheid, USA) (Ultra) testing capability within the same municipality, or to the National TB Reference Laboratory in the capital, Dili.¹² An aliquot of both samples will be tested individually by Ultra. Participants who test positive by Ultra (including trace results) will have two sputum samples sent to the National TB Reference Laboratory for confirmatory testing. Where there is insufficient sputum available for liquid culture, the field teams will revisit the participant to collect additional samples. Each sample will be tested individually using Mycobacteria Growth Indicator Tube (MGIT) liquid culture (figure 1). Drug-susceptibility testing for first-line and second-line drugs will be conducted for all culture positive samples in line with routine diagnostic practice in Timor-Leste.

All participants who are positively screened via X-ray of the chest and/or symptom will be referred directly to their local community health centre for further medical follow-up and care. New cases of TB diagnosed during the survey will be reported to the National TB Programme and the district TB coordinator. An expert group involving senior physicians with experience in the clinical management of TB in Timor-Leste will be made available to survey personnel and community health centre clinicians to provide expert advice on the clinical management of survey participants where required.

Data collection

Study data will be collected and managed using Research Electronic Data Capture (REDCap) electronic data capture tools hosted on a secure server at the Menzies School of Health Research (Darwin, Australia). REDCap is a secure, web-based software platform designed to support data capture for research studies using tablet devices and REDCap software.^{13 14} Regular automated and manual data validation measures will be implemented throughout the course of the survey fieldwork to optimise data quality. Prior to initiation of fieldwork, the survey team will be trained in data collection procedures and data security.

Data collected will include age, gender, history of current or past TB diagnosis and treatment, history of household TB contact and TB screening results (symptoms, X-ray findings of the chest) and laboratory results (Ultra and MGIT). Additional data will be collected from persons reporting symptoms suggestive of TB to assess health-seeking behaviour and barriers to care related to their symptoms. These fields are detailed in the online supplemental material.

Outcomes

The primary outcome will be a national point prevalence estimate of pulmonary TB disease. Prevalence estimates will be based on the survey TB case definition, outlined in table 1. Secondary outcomes, including screening positivity, GeneXpert Ultra positive TB cases and culture confirmed TB cases, will also be reported. The survey case definitions and testing algorithm are informed by WHO guidelines and advice from the survey’s technical working group and designed to be feasible, by minimising burden on the National TB Reference Laboratory, and have programmatic relevance in Timor-Leste by aligning with routine diagnostic practices and national case definitions.¹¹

Table 1.

Survey case definitions

Classification	Definition
Screen-positive case	Cough of ≥2 weeks duration OR fever OR weight loss OR night sweats, OR A CAD4TB abnormality score of ≥60 on chest X-ray, OR Cough of any duration OR fever OR weight loss OR night sweats AND chest X-ray not done or result not available
GeneXpert-confirmed case	TB detected at a grade of very low or higher in ≥1 sputum sample by GeneXpert Ultra, OR TB detected at a grade of trace in ≥1 sputum sample AND no history of TB treatment
Culture-confirmed case	Mycobacterium tuberculosis detected by MGIT culture in ≥1 sputum samples
Survey TB case	Mycobacterium tuberculosis detected by MGIT in ≥1 sputum samples, OR Two valid negative MGIT culture results AND no previous history of TB AND TB detected at a grade of very low or higher in ≥1 sputum samples by GeneXpert Ultra, OR One valid negative MGIT culture result and one invalid culture result AND no history of TB treatment AND TB detected at any grade in ≥1 sputum samples by GeneXpert Ultra No valid MGIT culture result AND no history of TB treatment AND TB detected at any grade in≥1 sputum samples by GeneXpert Ultra

MGIT, Mycobacterial Growth Indicator Tube; TB, tuberculosis;

Statistical analysis

Cluster-level and individual-level analyses of TB prevalence will be undertaken.¹⁵ Cluster-level analysis will calculate TB prevalence for each cluster and take the mean cluster-level prevalence to produce an overall estimate. This will be used to calculate the SD of the cluster-level prevalence and the SE of the mean across clusters. Logistic regression will be used for individual-level analyses, using both robust standard errors based on observed between-cluster variability and random effects models. Three regression models will be constructed, these are: (1) logistic regression with robust standard errors, no missing value imputation and analysis restricted to survey participants, (2) logistic regression with robust standard errors, multiple missing value imputation for both survey participants and non-participants and inclusion of all persons eligible for the survey in the analysis and (3) logistic regression with robust standard errors, missing value imputation for survey participants who were eligible for sputum collection, but Ultra and/or culture results are missing, and inverse probability weighting for all survey participants.¹⁵ The results of the models will be assessed to determine robustness of the resulting prevalence estimates.⁵

We will estimate the prevalence-to-notification ratio by comparing age-specific and sex-specific survey prevalence rates to case notification rates from routine TB programme data.

Data analysis will be conducted using R statistical software.¹⁶

Patient and public involvement

Prospective participants and members of the public were not involved in the survey design. The individual participant screening results will be provided directly to all survey participants by the field teams.

Ethics and dissemination

Ethical considerations

This study has been approved in Australia by the human research ethics committee of the Northern Territory Department of Health and Menzies School of Health Research (reference: 2021-4208) and in Timor-Leste by the Ministry of Health quality control committee and human research ethics committee of the Instituto Nacional de Saúde (reference: 1306MS-INS/DE/VII/2022).

Participation in all components of the survey is voluntary, and there is no financial or other incentive provided for participation. Eligible participants will be provided briefing on the risks and benefits of taking part in the survey and will be given an opportunity to ask any questions about the survey. They will then be asked to provide informed signed consent prior to undergoing TB screening. Informed signed consent will be provided by legal guardians of participants aged 15 or 16 years, who assent to participation.

It is possible that TB screening may elicit some anxiety among some survey participants. To mitigate this risk, experienced clinicians within the survey teams will be available to provide education and support to participants. Screening results will be made available to participants rapidly (X-ray within 5 min and Ultra within 48 hours) to minimise their wait time for results.

Participation in TB screening will require a small exposure to ionising radiation for the purpose of obtaining a chest radiograph. The dose of radiation used is exceedingly small, and the risk of such an exposure is considered hypothetical. However, in line with national guidance and practice in Timor-Leste, pregnant women will be excluded from having an X-ray of the chest taken. Lead shields will not be used for participants; however, lead aprons will be available for use by the radiographers.

Dissemination

Interim analyses of the survey will be conducted after the completion of every 10 clusters, with interim reports to be shared with the survey’s technical working group and key stakeholders within the Ministry of Health. The results of this survey will be published in several formats, including scientific papers for peer-reviewed publication, a report for policy-makers, clinical and public health service administrators and TB programme partners and lay summaries for the public and other stakeholders. The findings will be shared with the WHO to inform global estimates of TB burden.

Ethics statements

Patient consent for publication

Not applicable.