Data for: “Quantifying excess TB prevalence among people living with HIV on ART in the universal ART era: Evidence from a TB prevalence survey in Zambia and South Africa” – Data Codebook

Persistent identifier

10.17037/DATA.00005372

Description

This dataset contains variables used for the secondary analysis outlined in “Quantifying excess TB prevalence among people living with HIV on ART in the universal ART era: Evidence from a TB prevalence survey in Zambia and South Africa”. Data was collected as part of the TREATS TB prevalence survey of people living with human immunodeficiency virus (HIV) who were receiving antiretroviral therapy (ART) in Zambia and South Africa. The secondary analysis was restricted to participants classified as either living with HIV and receiving ART or testing HIV-negative during the survey. Individuals newly diagnosed with HIV during the survey, participants living with HIV but not receiving ART, and participants without an observed HIV status were excluded from the primary analytical comparison. Variables cover the participant's sex, age, highest level of education completed, socio-economic status, marital status, current occupation, smoking status, AUDIT screen Score, crowding index, composite HIV exposure, TB status, as well as community and country of residence.

Data collection and processing

Participants attended a mobile field site (MFS) and were guided through a series of stations, including registration, TB symptom screening, chest X-ray, sputum collection and HIV testing. Data collected at the different stations were recorded using a tailor-made electronic Data Management System based on a MySQL relational database operating within a local area network. Data from each MFS were synchronised daily to a central database.

The database integrated participant-level questionnaire data collected by trained research enumerators with results generated at the different survey stations and laboratories. These included socio-demographic, socio-economic and behavioural data; self-reported HIV testing and antiretroviral therapy (ART) history; HIV rapid-test results; TB symptom-screening results; digital chest X-ray and CAD4TB scores; and bacteriological results from sputum specimens, including Xpert MTB/RIF Ultra and culture results. Records from the different data sources were linked using unique study participant identifiers.

Several analytical variables were derived from the questionnaire data prior to analysis. Household socio-economic status (SES) was estimated using principal component analysis (PCA) based on housing characteristics and household asset ownership. Because patterns of asset ownership and housing quality differed between Zambia and South Africa, PCA was conducted separately for each country. The first principal component was used as the SES score and was subsequently categorized into country-specific quintiles ranging from lowest to highest socio-economic position. In Zambia, the PCA model including all candidate housing and asset variables was retained, whereas in South Africa a reduced model excluding water source, building type and motorcycle ownership was used.

Alcohol use disorder was assessed using the 10-item Alcohol Use Disorders Identification Test (AUDIT). Responses to the 10 items were scored and summed up to generate a total AUDIT score ranging from 0 to 40. The total score was categorized as low risk (0-7), medium risk (8-15), high risk (16-19), or addiction likely (20-40). Household crowding was represented using a crowding index, calculated as the number of people living in the household divided by the number of sleeping rooms.

Following data cleaning and validation, the TB outcome and other analysis variables were generated. Prevalent pulmonary TB was defined using bacteriological results from Xpert-Ultra and culture testing. Participants who screened positive on TB symptoms or chest X-ray but had missing bacteriological results had TB status multiply imputed according to the approach used in the primary TREATS TB prevalence survey analysis. Participants who screened negative on both TB symptoms and chest X-ray were assumed not to have prevalent TB.

The repository therefore includes the analysis-ready variables used for the present secondary analysis. For the present study, the analytical dataset was restricted to participants classified as either living with HIV and receiving ART or testing HIV-negative during the survey. Individuals newly diagnosed with HIV during the survey, participants living with HIV but not receiving ART, and participants without an observed HIV status were excluded from the primary analytical comparison.

To protect participant confidentiality, directly identifying information is not included in the research dataset. Access to individual-level data is governed by the ethical approvals, participant consent provisions and data-access procedures of the TREATS study.

Data Codebook

This dataset contains 314 variables outlined below.