Work

Data Resource Profile: Global Trachoma Serology Data Repository

Year: 2025

Type: article

Abstract: The Global Trachoma Serology Data Repository includes harmonized survey data with individual-level data on immunoglobulin G (IgG) antibodies to Pgp3 and CT694 Chlamydia trachomatis antigens, created to support operational research on the use of serosurveillance for the trachoma elimination endgame. The repository is unique in its scale: >60 000 anonymized individual records collected from children (1–9 years old) in 48 surveys in 15 countries, between 2012 and 2023, and across a range of trachoma endemicity from high-transmission settings to post-elimination. Featured variables include age, year of survey, geography (country, region, and/or district), and whether or not antibiotic mass drug administration was done prior to the survey. Individual-level records are linked to their sampling cluster, which is fundamental for population-level analyses. The repository also contains accompanying data for clinical signs of trachoma and nucleic acid amplification test-based measures of ocular C. trachomatis infection, which, together with IgG measurements, enable a more comprehensive understanding of trachoma epidemiology. All datasets are freely accessible at https://doi.org/10.5061/dryad.5qfttdzhx and will be updated regularly whenever new data become available. Data contributions are welcome via e-mail to [email protected]. Trachoma, a neglected tropical disease, is caused by repeated infection with the bacterium Chlamydia trachomatis (Ct) and is the most common infectious cause of blindness [1]. Trachoma is targeted for global elimination as a public health problem by the year 2030; by 19 May 2025, 23 countries had been validated by the World Health Organization (WHO) as having eliminated trachoma as a public health problem [2]. Surveillance of trachoma has traditionally relied on measurement of the clinical signs trachomatous inflammation—follicular (TF) and trachomatous trichiasis (TT) [3]. Country health ministries are encouraged to share point prevalence data of TF and TT with the open-access Trachoma Atlas (https://www.trachomaatlas.org/) and, for countries in the WHO’s Africa region, the ESPEN (Expanded Special Project for Elimination of Neglected Tropical Diseases) portal (https://espen.afro.who.int/) [4]. This is one example of trachoma prevalence data sharing. At low levels of transmission and after antibiotic mass drug administration, TF is an imperfect marker of Ct transmission. Serological testing—the detection of specific antibodies in blood—is an approach that is under consideration to monitor programmatic recrudescence (the return of active trachoma infection), including in post-validation surveillance. Serology is central to pathogen surveillance and the inference of infection history [5] and, in trachoma, monitoring transmission by incorporating dried blood samples into standard trachoma surveys has proven useful in supporting programs as populations approach elimination [6–10]. We set out to map the trajectory of trachoma to elimination through the analysis of population-level antibody prevalence. We assembled a comprehensive global dataset for trachoma serology matched with clinical signs and molecular measures of infection [the quantitation of Ct bacterial DNA by using nucleic acid amplification (NAAT)]. This was a collaborative effort between study investigators that was set up to provide public access to a large selection of data across various populations and geographies encompassing varied dimensions of trachoma endemicity. Contributing datasets were identified through our trachoma collaborator network, the WHO Alliance for the Global Elimination of Trachoma by 2020 (GET2020), the Global Neglected Tropical Diseases laboratory at the United States Centers for Disease Control and Prevention (CDC), and finally through a PubMed literature search in February 2021 for “(pgp3 OR ct694) AND (trachoma)” to identify other trachoma serology datasets that were already publicly available. We then invited principal investigators and corresponding authors who had published trachoma seroepidemiology studies to contribute de-identified data and accompanying metadata for our analyses and with the intention of sharing them publicly in a harmonized database. These were our leading inclusion criteria, which ensured that the contributors had primary ownership and were chiefly responsible for their data and/or the respective studies, and we were secondary custodians. Therefore, all the contributing datasets, except for those currently under peer review, have accompanying PubMed information. We provided a data-contribution assent letter to be signed or approved via e-mail by the lead data contributor. We invited up to three contributors per published study to be included as coauthors in our secondary analyses of the data and they are part of the Global Trachoma Serosurveillance Study Group. The Global Trachoma Serology Data Repository is the largest collection of publicly available data on anti-Pgp3 and anti-Ct694 immunoglobulin G (IgG) responses—two antibodies that are sensitive and specific markers of previous ocular Ct infection [11]. The repository currently contains anonymized individual-level data on 63 911 measurements from children aged 1–9 years from 15 countries across four WHO regions: the African Region (Ethiopia, Ghana, Niger, Malawi, Morocco, Togo, The Gambia, and the United Republic of Tanzania), the Eastern Mediterranean Region (Sudan), the Region of the Americas (Peru), the South-East Asia Region (Malaysia), and the Western Pacific Region (Kiribati, Solomon Islands, Vanuatu, and Papua New Guinea) (Table 1 and Supplementary Table S1). The age group of 1–9 years is the target age range for measuring Ct infection and active trachoma, as it has the highest burden of these phenomena [12, 13]. As such, we required that the data were in 1-year age bands and not aggregated into age groups. Measurements from children below the age of 1 year were excluded, to mitigate the contribution of passively transferred maternal antibodies. The repository contains harmonized data sourced from 48 surveys comparable across countries. Each contributing survey or study had received its own funding and research ethics approval at the respective host institution. Funding for the collection, creation, and harmonization of data contributed to the repository was provided by a 5-year grant (2021–2026) from the National Institute of Allergy and Infectious Diseases to investigators at the University of California, San Francisco (UCSF) (R01AI158884). Although possible, updates to the repository beyond the initial grant period are not guaranteed. Human participant oversight for the data collation, harmonization, and creation of public de-identified datasets were provided by the Institutional Review Board at UCSF (protocol #20-33198). Geographical distribution of data. Summary of all trachoma serology data by each contributing country. The table indicates the number of child measurements from each country and how many evaluation units (EUs) in a country have individual-level assessment of trachomatous inflammation—follicular (TF) and individual-level measurements of ocular C. trachomatis infection by using NAAT. Detailed data summaries for each EU are provided in Supplementary Table S1. In the Papua New Guinea surveys, C. trachomatis infection prevalence was measured by using NAAT only among children who had TF, and not in all children, as in the other included surveys. Three surveys from Ghana did not measure C. trachomatis infection prevalence by using NAAT but were classified alongside the other Ghana EUs, as they were part of the same survey series that included nine total EUs, six of which measured infection by using NAAT, and were definitively considered as not requiring programmatic actions. Only cluster-level data are available for TF and NAAT for some of the surveys in Ethiopia, Tanzania, and Niger. Geographical distribution of data. Summary of all trachoma serology data by each contributing country. The table indicates the number of child measurements from each country and how many evaluation units (EUs) in a country have individual-level assessment of trachomatous inflammation—follicular (TF) and individual-level measurements of ocular C. trachomatis infection by using NAAT. Detailed data summaries for each EU are provided in Supplementary Table S1. In the Papua New Guinea surveys, C. trachomatis infection prevalence was measured by using NAAT only among children who had TF, and not in all children, as in the other included surveys. Three surveys from Ghana did not measure C. trachomatis infection prevalence by using NAAT but were classified alongside the other Ghana EUs, as they were part of the same survey series that included nine total EUs, six of which measured infection by using NAAT, and were definitively considered as not requiring programmatic actions. Only cluster-level data are available for TF and NAAT for some of the surveys in Ethiopia, Tanzania, and Niger. The Trachoma Serology Data Repository contains individual participant data from cross-sectional trachoma serology surveys and randomized controlled trials (RCTs) conducted between 2012 and 2023. We mostly included measurements collected in population-based surveys conducted in an evaluation unit (EU)—“the normal administrative unit for health care management, consisting of a population unit between 100 000–250 000 persons”—which often corresponds to a “district” or similar second-level administrative unit and where programmatic decisions for intervention delivery are implemented [14]. Details of the data capture, serologic testing, polymerase chain reaction (PCR) testing, and clinical exam are published independently for each survey and also in the Supplementary material (Source data dictionary template; study citations are provided in Supplementary Table S1). We gathered data collected among children aged 1–9 years, with a few exceptions. Based on the original study designs, measurements in Matameye, Niger; Dosso, Niger; and Wag Hemra, Ethiopia were restricted to children aged <5 years (Fig. 1 and Supplementary Table S1). In Ethiopia in 2017 (Alefa, Andabet, Dera, and Woreta town districts), PCR testing was done at the cluster level among children aged 1–5 years, while serology and TF were measured among children aged 1–9 years. We retained all years of serology measurements for RCTs with repeated cross-sectional data (WUHA in Wag Hemra, Ethiopia; Kongwa, in the United Republic of Tanzania from 2013 to 2015; and MORDOR in Dosso, Niger). The Malaysian survey was an opportunistic study assessing the seroprevalence for multiple infectious diseases by using serological multiplex bead assays, and is therefore not an original trachoma study. The surveys in Morocco, Togo, The Gambia, and Sudan did not measure infection by using PCR, while Peru had not completed PCR testing by the time of writing of this manuscript. Age distribution of the trachoma serology data by contributing country. The different lines in each panel represent the various participating EUs. The number of measurements (y-axis) is calculated as the proportion of measurements in each age group per EU. For improved visualization, the data for the MORDOR, Niger survey were clipped at age 4 years, as there were only 17 measurements for the 5-year-olds in that survey. In the Papua New Guinea surveys, C. trachomatis infection measured by using PCR was only done for children who had TF, and not in all children, as in the other included surveys. Also, three surveys from Ghana had not measured C. trachomatis infection by using PCR but were part of the same survey series that included nine total EUs, six of which measured infection by using PCR. In Mendi, Papua New Guinea, and Kiritimati, Kiribati, each individual participant was treated as an independent observation (no cluster samples). Overall, the largest data contributors by country were Ethiopia (23.0% of all measurements), followed by Ghana (18.4%) and Niger (10.7%) (Table 1). The repository has individual- and cluster-level data. The former has 39 fields, including age, location identifiers (country and district/highest-level administrative division), clinical examination, and laboratory test results. A description of the fields is available through a data dictionary (Box1 and Supplementary Box S1). Each row represents an individual participant, except for the RCTs with repeated measurements, and individual-level records can be linked to cluster information by using cluster ID. Geographical coordinates, although provided in some surveys, are not available publicly, for confidentiality reasons. The cluster-level data have 66 fields that were prepared from the individual-level data and contain the cluster-level prevalence of TF, TI, PCR, Pgp3, and Ct694 calculated for the age bands of 1–5 and 1–9 years (Box1 and Supplementary Boxes S1 and S2). The data integrate additional measurements for three studies (Matameye, Niger; Kongwa, Tanzania from 2012 to 2015; and Ethiopia in 2017) that did not have TF and PCR data at the individual level but instead reported the data at the cluster level (Supplementary Table S1). All internal identifying information and IDs are replaced by the public IDs or excluded from the final public-facing records. The repository contains codebooks with high-level data summaries, such as means and standard deviations, plots of distributions, counts of missing values, and descriptions of missing data patterns. This level of granularity facilitates data interoperability and reuse. Each contributing study/survey has an associated “PMID” value that references the original publication PubMed ID and a “public_data” value that indicates whether the data from the study have been released publicly through other means, such as via a publication or other repository. The Global Trachoma Serology Data Repository enhances accessibility for the epidemiology and operational research communities to investigate trachoma transmission across key variables such as study population, geographical area, age, sampling methodology, antigen target, and other fields. The data have potential for downstream statistical applications to bridge between estimates of transmission intensity based on TF prevalence and infection and serological summary measures. In this regard, we used the data to study cluster-level heterogeneity in transmission and to compare consistency and correlation between serological summary measures (mean IgG, seroprevalence, and force of infection) and with paired clinical and molecular correlates of infection [15]. We also studied the data as an evidence base to synthesize a roadmap of population-level progression to elimination by estimating the seroprevalence and seroconversion rates (SCR—a measure of incidence) in different phases of the disease, from endemic transmission to post-elimination of transmission [16]. More technical studies that take advantage of subsets of the data available within the repository are possible, such as a previous analysis of the Pgp3 seroreversion rate in longitudinal samples collected from children in an RCT to evaluate the effect of comprehensive water improvements and hygiene programs on trachoma outcomes [17]. Serological surveys have the potential to facilitate evidence-based decision-making and guide strategies to detect possible recrudescence. Antibody prevalence has been used as an indicator for the elimination of African onchocerciasis and statistical models on the predictive value of serology data have supported decision-making in onchocerciasis-control programs [18, 19]. We demonstrated a similar applicability for trachoma serology data by using the harmonized data to propose SCR thresholds for transmission assessment in “baseline” and surveillance surveys [16]. Our ongoing projects using the trachoma serology data focus on developing sampling designs for serological studies in which we are exploring survey designs to monitor and capture transmission in pre- and post-elimination settings. We have shown that serological measures can be used for the comparison of infection pressure between populations. Through an open-access repository, we aim to foster collaboration and engagement among the trachoma surveillance community to explore and analyse serology data to monitor and evaluate transmission. Together with the Trachoma Atlas (https://www.trachomaatlas.org/), the harmonized serology datasets will support the generation of knowledge and contribute to a better understanding of trachoma epidemiology. The Global Trachoma Serology Data Repository is built from a broad set of studies conducted along a wide gradient of trachoma endemicity. The data cover considerable variability in transmission intensity, in which local dynamics vary from suppressed transmission, e.g. in The Gambia, to persistent prevalence, e.g. in Ethiopia (where the prevalence of TF among children aged 1–9 years in two or more trachoma impact surveys is ≥5%, the elimination threshold), and are therefore truly representative of the current status of trachoma epidemiology. Almost all serology data were generated by using the same laboratory protocol, with training and support from the Global Neglected Tropical Diseases laboratory at CDC. Also, a majority of the studies used the Tropical Data protocol for trachoma prevalence surveys [4] and most were population-based cluster samples. All source studies have been peer-reviewed. To our knowledge, this repository is the largest collection of trachoma serology data and users interested in a particular survey or population can easily section that from the consolidated data. To obtain the final data presented here, we reached out to nearly 50 contributors across multiple research groups, institutions, and governments, and created the Global Trachoma Serosurveillance Study Group with a common interest in trachoma seroepidemiology. Our efforts and outreach advocate for collaboration and data and are a example of public health and data in infectious diseases Although the serology data cover a broad range of and a large geographical area, they are not representative of all countries. surveys, such as many in the Pacific region, were conducted to trachoma epidemiology. The data were collected as part of programmatic that included serology testing, under the of operational but as which have varied in intensity and As many surveys were on trachoma few included additional and of potential such as or We the data from within our but were also in our to principal We therefore that the collection be and will not include all published studies the seroepidemiology of All datasets are freely accessible for through the at and with the The data are available in and dataset to provide for use with a of The source data as provided by contributors be for confidentiality reasons. we have to be as as possible, we will to include additional data as they become available through of PubMed using the same and we to via e-mail to additional and/or available data. of the data will to be and each will include a summary of or of published trachoma seroepidemiology studies and on the repository are through the corresponding of the users and can this along with the data repository and data users are encouraged to the original of the which can be identified by using the publication PubMed ID. In the we to more to users to search or data by using an as as to data for via Human oversight for the data collation, harmonization, and creation of public de-identified datasets was provided by the Institutional Review Board at UCSF (protocol Contributing surveys were each and approved under research reported in the primary data We all who have contributed their data to the repository and the anonymized data are and and the Global Trachoma Serology Data and updated and the repository with from and the Global Trachoma Serosurveillance Study Group. the and the with All authors and approved the manuscript. The authors are responsible for the in this and they not represent the decisions or of the with which they are or of the United States The Global Trachoma Serosurveillance Study National Health Health Organization for of Health and National Health of The Solomon Neglected Tropical Diseases World Health University of San San of and University of San San National of University of San San of University of San San Institute for Global Health University of San San Health National University of San San of University of San San Institute for Global Health University of San San of Tropical of Tropical University of Health of Diseases and Centers for Disease Control and University of San San Institute for University of Ghana, of Diseases and Centers for Disease Control and of Tropical University of San San Health of Trachoma Kongwa, Tanzania), University of San San of University of San San and of University of San San of Diseases and Centers for Disease Control and Tropical Diseases for Global of Institute for of National University of of Health of University of California, of and University of Tropical Trachoma Tanzania), of and Elimination of Health of and Disease Control of Health and of Diseases and Centers for Disease Control and of Health and Solomon Tropical Diseases for Global University of San San Trachoma of of Health and of Tropical of Papua New Guinea, Papua New of Tropical of Diseases and Centers for Disease Control and National Health of The University of San San of University of San San Institute for Global Health University of San San Solomon University of San San The of Tropical Health University of San San of University of San San of and University of San San Institute for Global Health University of San San Institute of Papua New Supplementary data are available at This was supported by the National of Health to were used for the of this manuscript. Trachoma Serology Data Data in the individual and cluster data as of of the variables are provided in Supplementary Boxes S1 and and are also available in codebooks Study information year information information information country test Study information year information country test Study information year information information information country test Study information year information country test Trachoma Serology Data Data in the individual and cluster data as of of the variables are provided in Supplementary Boxes S1 and and are also available in codebooks Study information year information information information country test Study information year information country test Study information year information information information country test Study information year information country test

Source: International Journal of Epidemiology

DOI: 10.1093/ije/dyaf183